Nvidia CEO Jensen Huang has hailed the capabilities of Chinese open source AI models amidst growing interest in low-cost options for enterprises.

In an interview with Axios this week, the Nvidia chief said competition between open and closed-source models is positive for the broader global AI industry.

“The world needs open models,” he told Axios. “These Chinese models are excellent. Open source models that are excellent should be used”.

Huang’s comments come amidst growing interest in Chinese open source AI models in recent weeks. The Kimi K3 model, developed by Moonshot AI, temporarily paused subscriptions due to skyrocketing demand.

Surging interest in the Kimi K3 model bears similarities to that of DeepSeek, which rocked the US AI industry when it launched in early 2025.

A key advantage of these models lies in both their performance and cost efficiency, the latter of which has become a key focus for enterprises amidst growing concerns over spiralling AI costs in recent months.

As ITPro reported in November 2025, open source AI models often perform on-par with closed source options, and could potentially save enterprises millions in costs each year.

Open source models "routinely achieve 90% or more” of the performance of closed counterparts, the study noted. They also benefit from “significantly lower prices”, with operational costs plunging up to 84% in some cases.

In a research paper published by DeepSeek last year, the company said it spent a paltry amount training its flagship R1 model compared to US rivals who’ve spent billions on training purposes.

Anthropic CEO Dario Amodei suggested building a new AI model could cost over $100 billion in 2024.

Security concerns persist

While the popularity of Chinese AI models continues rising, concerns over security have been flagged repeatedly.

The concerns surrounding Chinese AI models typically focus on areas such as data privacy and national security, with some critics suggesting these tools could be used as a ‘backdoor’ for Chinese intelligence services.

Axios previously reported that the White House could consider imposing restrictions or tight conditions on US firms working with these models.

DeepSeek was subject to the same scrutiny during its rise to prominence across 2025, for example. In August last year, cybersecurity experts told ITPro that using DeepSeek equated to “printing out and handing over your confidential information”.

Huang told Axios that these concerns are a “misconception” and can be used by enterprises safely. On the topic of potential White House restrictions, Huang said this could prove counterproductive and that American firms should consider a breadth of options.

“You can download the models. You could fine-tune it, you can enhance it, you can guardrail it as you desire,” he said.

Huang has a vested interest on this front, however, and noted that increased usage will ultimately benefit Nvidia.

“If there’s great AI, even if it’s open, wherever it comes from, there will be more use,” he told Axios. “Whenever there’s more use, you’ll have to sell a lot more Nvidia computers. We’ll have to build more data centers. We’ll have more services. The technology will diffuse into more industries”.

Bill Conner, president and CEO Jitterbit, said the Kimi K3 launch marks a “clear turning point in accelerating global adoption of Chinese LLMs” but still urged caution with regard to security.

“Enterprises may underestimate the security, accountability, and governance risks associated with adopting new Chinese LLMs,” he said.

“Moonshot AI operates as a cloud service with data stored in China, introducing potential risks for organizations whose data may be shared unknowingly.”

Conner said adoption of Chinese AI models should be treated as a “strategic risk decision, not merely a cost-saving opportunity”.

FOLLOW US ON SOCIAL MEDIA

HPE and Nvidia have announced new features for HPE AI Factory, aimed at improving customers' experience with AI agents.

Announced at HPE Discover 2026, HPE AI Factory, which is part of Nvidia AI Computing by HPE, will now offer the Nvidia Agent Toolkit software, which includes Nvidia Nemotron open models, Nvidia NemoClaw (the chip maker's answer to OpenClaw), and Nvidia OpenShell secure runtime.

The new capabilities will, the companies say, help customers deploy enterprise-ready AI with greater observability, efficiency, and control.

There are also new hardware additions to HPE Private Cloud AI, which now has HPE ProLiant Compute DL394 Gen12 servers with Nvidia Vera CPU for, in the words of the company, "a compute-optimized foundation for agentic AI and high-performance data processing, including security and management features".

On the security side, the update also includes new Zerto Software capabilities – HPE acquired Zerto in 2021 – which the company says will help customers identify when a "rogue agent" starts carrying out actions that it shouldn't. If this happens, the Zerto capabilities can "use continuous data protection to rewind to a clean state".

In addition, HPE Private Cloud AI also now has support for secure local agent registration, which the company says will allow customers to approve AI models, skills, and tools while following centralized governance and security policies.

Finally, HPE Private Cloud AI also has HPE Alletra Storage MP X1000, which enables customers to apply metadata and governance policies that prepare data for AI applications. According to HPE, this allows businesses to boost token throughput by up to 20% and improve prompt processing efficiency.

Of the announcement, HPE president and CEO, Antonio Neri, said: "Across networking, servers, storage, and software, HPE is delivering full-stack AI solutions with NVIDIA that build the foundation for agentic enterprises, helping customers move from experimentation to production with control and confidence."

ITPro's managing editor Jane McCallion is at HPE Discover 2026, and you can follow all the latest announcements and analysis from the event here.

Nvidia first announced its DGX Spark mini supercomputer last year – and we've since seen a flurry of machines made by OEM partners hit the market. We've previously reviewed the outstanding Asus Ascent GX10, with a handful of alternatives like Lenovo's ThinkStation PGX and Acer's Veriton GN100. We've currently got our hands on the Dell Pro Max with GB10 – a mini PC designed to seamlessly fit into your existing IT setup.

This machine is designed specifically for what you would ordinarily describe as "power users". These include data scientists, researchers, as well as budding and current AI practitioners who want the freedom and flexibility to experiment with AI deployments within the enterprise, among a handful of other cohorts.

As such, it's powered by a custom Linux distro and hardware specifically – the Grace Blackwell Superchip – configured to let you run meaningful AI workloads on your device, rather than relying on a cloud service or remote communications. That means it's certainly not a device for everyone – and should only be considered for those who plan on making the most of the on-device AI capabilities these machines allow.

While DGX Spark-based systems are all very similar, they look very different and are also priced differently. The Dell Pro Max with GB10 comes in at a hefty price approaching £6,000 in the UK, which is close to £2,000 more than some of devices sold by its competitors. Can it make up for this asking price in other areas?

Dell Pro Max with GB10 review: Build and design

The device itself isn't much bigger than conventional mini PCs and fits snugly on your desk thanks to its standard dimensions of 150 x 150 x 51mm – which is consistent across all DGX Spark systems from partner OEMs. They're all roughly the same weight, too, with the Dell Pro Max with GB10 weighing 1.3kg – versus the Acer's 1.2kg and the Asus' 1.4kg.

Also identical between these machines are the ports you'll find on the rear. These include three USB-C 3.2 Gen 2x2 (20Gbps) ports with DisplayPort 2.1 compatibility, one USB-C Gen 2x2 with PD in, an HDMI 2.1 port, an Ethernet port, and Nvidia's ConnectX-7 SmartNIC port used to run two of these machines in parallel. There is a marked lack of USB-A ports, so you'll need additional dongles or docks to use peripherals, outside of connecting them to a monitor and accessing them through a USB-C DisplayPort setup.

With so little separating the different models, OEMs must make the most of their physical appearance, and Dell has adopted a more understated and neater design outlook than some of the alternatives. The machine's front is covered with a hexagonal honeycomb lattice grill, with Dell's logo centered perfectly. One neat addition is a tiny power LED indicator on the left of the front, alongside an LED built into the USB-C power cable. The rear adopts the same lattice design on the upper half, with the ports laid out in the standard format on the lower half – all of which are flush against the surface.

Overall, it exudes more of an overtly corporate feel – underlined by a relatively uninspiring grey finish. It won't look out of place in any office environment, but it does lack the flair that you might get with the Acer Veriton GN100, for example.

Dell Pro Max with GB10 review: Specs and performance

The GB10 Superchip is the beating heart of the Dell Pro Max, with GB10, and lets users tap into a Grace CPU and Blackwell GPU combined to deliver an efficient setup for AI-centric work. The 20-core Arm processor combines 10 Cortex-X925 performance cores and 10 Cortex-A725 efficiency cores.

Combined with the 5nm GB20B GPU, you'll have the local potency to reach a performance capacity of 1,000 teraFLOPS of AI performance with the learner and more efficient FP4 precision, and a maximum of 170 teraflops with the industry standard FP16 precision. Both are formats for training and inference, with the 4-bit floating point format allowing for lower-precision but accelerated AI performance.

The machine can also support models of up to 200 billion parameters, which is a little more than the size of GPT-3 – but this increases to a maximum of 405 billion parameters if you connect two of these together. Each one is also capped at a maximum power draw of 240W – with a 280W power brick – but you won't encounter it unless pushing the device to its absolute limits.

Comparing DGX Spark systems to conventional machines is hard – because they're a different class of device with specific purposes, which vary from general-purpose PCs designed for almost everyone. However, when testing with Geekbench 6, we registered a massive single-core score of 3,123 and a multi-threaded performance of 19,708. This is in line with similar systems and notably higher than the comparable Framework desktop.

Unscientific testing and experimentation also pointed to easy and quick setup on AI tasks, such as image generation with ComfyUI, with the machine running quietly throughout – while producing very little heat. While not as fast or as powerful as industrial-scale systems or remotely accessible infrastructure, we were impressed with how much you can push a small machine like this.

Dell Pro Max with GB10 review: Features

Setting up the DGX OS platform, which is based on Ubuntu 24.04 LTS (for ARM), is an absolute breeze. The overall look and feel of the system closely follows Nvidia's design aesthetic and colour palette, and is fairly intuitive to use. Helpfully, there are a few useful tools that you can use out of the box to get started on running AI tasks.

The best place to start would be Nvidia's playbooks, a web-based platform with different recipes that centre on specific AI-related tasks. For first-timers, this library of tutorials is the best place to experiment with the tools the DGX Spark-based system can let you tap into. This won't be accessible, or even necessary, for the average user – but anybody within an organisation with a willingness to engage in AI will find it an incredibly useful way of getting to grips with what this machine is capable of.

Each recipe includes the core purpose, the practical risks of engaging in the project, and a rough time estimate to get it up and running. Setting up DGX Dashboard, for example, takes roughly half an hour and provides a user interface from which you can manage your system, manage resources, and launch JupyterLab instances. ComfyUI, too, is a great place to start – taking about 45 minutes to build and run your own AI image generator.

Generally, these playbooks are easy to follow – but there's an active community, too, that you can check in with if you run into any trouble. Which you inevitably will if you're just getting started. But the triumph in this setup is the way it democratises AI development – breaking down rather complex and unwieldy processes into digestible and easy-to-follow chunks; that's much like the device itself, which gives a far broader cross-section of the workforce access to tools and capabilities once reserved for only the most technically minded.

Also useful for power users hoping to get more from their machine is the capacity to connect two DGX Spark systems together. Nvidia Sync, which is a new feature, lets you connect your local desktop or laptop with the machine, so you can operate it remotely within the same subnet. You'll be able to run workloads or launch applications remotely and even share the display between your machines. It's a perfect way to offload tasks onto this specialised piece of hardware if you prefer working primarily with your laptop.

Dell Pro Max with GB10 review: Is it worth it?

This new class of machine is the perfect way to get started on your AI experimentation and implementation journey within the enterprise. The systems are incredibly compact, benefit from fantastic specs courtesy of the GB10 superchip, and have a deep pool of 128GB unified memory to dip into, alongside plenty of SSD storage.

Whether the Dell Pro Max with GB10 is worth it depends on how it compares with its counterparts. Admittedly, it's expensive, but it adopts a relatively understated yet mature design, which contributes well to airflow. The power draw capacity is also slightly higher at 280W versus the standard 240W. For AI practitioners and those just getting started, it's a brilliant device – it's just worth weighing up whether the subtle tweaks are worth the additional cost. If you're all-in on the Dell ecosystem, it'll complement your existing setup perfectly, but the budget-conscious would understandably be drawn to the most cost-efficient DGX Spark system out there – which isn't this one based on current retail prices.

Dell Pro Max with GB10 specifications

AI cloud firm Nebius is planning three new deployments of Nvidia infrastructure in the UK, in a deal that, said Nvidia, supports the UK government's AI Opportunities Action Plan.

The £1.7 billion investment forms part of Nebius's plans to expand its commercial and AI research and development hub in London. The three deployments will use the latest generations of Nvidia's full-stack, end-to-end AI factory platform technology, and are expected to reach 65 MW when fully ramped up next year.

"The UK is one of Europe's most ambitious AI markets, with a clear public policy framework and a strong base of innovators," said Paolo Guglielmini, vice president EMEA at Nvidia.

"Nebius is expanding local access to Nvidia's full-stack AI factory platform, giving British companies the best performance, economics, and ecosystem support to train and deploy frontier and open-source AI close to their data, customers, and teams."

As part of the plan, Nebius has signed a ten-year, 22 MW deal with data center developer Kao Data for a deployment at its Harlow data centre campus, powered by 100% renewable energy and supported by HVO-powered generators.

"The UK is a major destination for AI and is becoming an important part of Nebius's global footprint," said Andrey Korolenko, chief product and infrastructure officer at Nebius.

"By bringing dedicated capacity to support inference workloads, we can enable UK AI builders and enterprises to achieve their AI goals."

Nvidia is touting the deal – along with several others – as evidence of its commitment to the UK.

CoreWeave is building in the government's AI Growth Zones, and seven more Nvidia AI Cloud ecosystem partners have plans in the pipeline. BT and Nscale are building sovereign AI data centers across three existing BT sites, based on Nvidia AI infrastructure.

Cosine is building an end-to-end sovereign AI coding platform for highly regulated industries such as financial services, critical infrastructure, and national security, based on the Isambard-AI supercomputer, built on 5,400 Nvidia GH200 Grace Hopper Superchips.

Cursive is working on self-improving AI systems that learn continuously from real-world data, enabling them to operate autonomously over long periods of time, part-funded through the Sovereign AI Fund.

"The Sovereign AI Fund is more than just processing power – it's a statement about investing in AI in the UK," said Talfan Evans, cofounder and CEO of Cursive.

"Sovereignty is actually now a buying criterion – and it's a challenge to tap into the resources we uniquely have as U.K. and European companies."

Meanwhile, Doubleword, the UK's first dedicated inference lab, deploys open models including Nvidia Nemotron 3 Super 120B and builds on the Nvidia Dynamo inference framework; and Prima Mente is building biological foundation models to identify new biomarkers, subtypes and drug targets of Alzheimer's, Parkinson's and ALS.

"We're determined to make the UK the best place in the world to build and deploy AI – backing the infrastructure businesses, researchers and public services need to put this technology to work," said UK AI minister Kanishka Narayan.

"Nebius's investment brings significant AI compute into the UK, giving companies what they need to train, test and run advanced systems here at home – helping drive productivity by rolling AI out widely across the economy."

Nvidia unveiled a new chip designed to embed AI into laptops and desktop computers, calling it a "new beginning" for PCs.

Nvidia has led the AI boom as the key supplier of chips to many leading developers, helping the GPU maker become the most valuable company in the world with a record $5 trillion valuation.

The RTX Spark chip brings AI inference to the device, rather than sending it up to the cloud, letting agents run locally – something Nvidia says will help drive AI adoption, which it claims has been stymied by an inability to run agents "securely and privately" on users' own PCs.

According to Nvidia, the RTX Spark can run "120B-parameter LLMs with up to 1 million tokens context using agents locally" – and it's also great for gaming and video editing, too.

"The PC is being reinvented," said Jensen Huang, founder and CEO of NVIDIA, in a statement. "For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask – and the PC does the work."

"RTX Spark brings everything Nvidia has built – CUDA, RTX, our AI platform – into a single superchip. Local agents,” he added. “Frontier models. Creative workflows. RTX games. All on a laptop. This is the new PC. The personal AI computer."

The RTX Spark features 1 petaflop of performance and up to 128GB of memory, using an Nvidia Blackwell RTX GPU – with 6,144 CUDA cores and fifth-generation Tensor Cores, according to Nvidia – alongside a 20-core Nvidia Grace CPU.

As ever, Nvidia is also promising all day battery life.

Microsoft partnership

To boost security, Nvidia worked alongside Microsoft on new core protections for Windows so agents "run safety and under full user control", the company said.

That includes the ability to disguise personal information in queries sent to cloud models, Nvidia said, adding that such security layers were already adopted by agent developers including Hermes Agent and OpenClaw in their Windows apps.

"Running solutions like OpenShell and the Microsoft security primitives on RTX Spark will enable users to leverage a fully integrated stack for private, personal agents running on device," said Vincent Koc, chief architect at the OpenClaw Foundation, in a statement.

Microsoft will also add RTX Spark powered Windows agent experiences to the taskbar. "Our goal is to deliver unmetered intelligence to every home and every desk with Windows," said Satya Nadella, chairman and CEO of Microsoft, in a statement. "RTX Spark marks a real breakthrough towards that vision."

AI PCs

The RTX Spark processor was developed alongside MediaTek and as part of a three-year partnership with Microsoft.

The chip will hit shelves this fall, with models expected from Asus, Dell, HP, Lenovo, and MSI, as well as Microsoft's own Surface. Acer and Gigabyte are expected to also add an RTX Spark model to their line-ups.

Those PC markers will all be hoping this next round of AI PCs sell better than the last, which has proven a mixed bag.

Gartner said at the end of last year that AI PCs were "reshaping the market" but adoption was slower than hoped due to tariffs and market uncertainty.

Meanwhile, Dell said customer buying decisions weren't driven by AI — and may even be confusing them.

FOLLOW US ON SOCIAL MEDIA

While concerns mount over the impact of AI on the global jobs market, Nvidia CEO Jensen Huang insists workers in some professions will benefit greatly from the huge infrastructure build-out required to power the technology.

Speaking during a commencement speech for Carnegie Mellon University’s class of 2026 at the weekend, Huang said economic growth through AI won’t be limited to the technology sector.

Indeed, trade workers will be among the vanguard helping to ramp up infrastructure roll-outs across the US and around the world.

“Electricians, plumbers, iron workers, technicians, builders, this is your time,” he told students. “AI is not just creating a new computing industry, it is creating a new industrial era.”

“AI gives America the opportunity to build again.”

Huang’s prediction of a golden era for tradespeople comes amid huge investment in the underlying infrastructure used to power AI.

Capital expenditures at a host of major tech players, including AWS, Microsoft, Meta, and Google have skyrocketed as providers look to match surging compute capacity demands.

The AI infrastructure will require significant investment over the next five years to keep pace with demand, according to 2025 research from McKinsey.

The consultancy projects data centers will require roughly $6.7 trillion in investment by 2030 to keep pace with demand for compute capacity.

Data centers designed specifically for AI processing will also require a whopping $5.2 trillion in capital expenditure.

Job market threats and opportunities

With AI infrastructure build-outs continuing at pace, research shows that demand for tradespeople is surging. Analysis from Randstad in March this year, for example, found global demand for skilled trades is growing three-times faster than professional roles.

Demand for HVAC engineers and construction workers has surged 67% and 30% respectively, the consultancy found.

Concerns over the impact of AI on “white-collar” work and entry level roles have been rising over the last two years, however.

Last year, Anthropic CEO Dario Amodei suggested that AI could wipe out half of entry-level roles by the end of the decade. A March 2026 report from Anthropic also warned white-collar workers are highly exposed to AI and automation.

The study also uncovered “suggestive evidence” that hiring of younger workers has slowed in exposed occupations. That tracks with SignalFire’s 2025 State of Talent report, which pointed to a marked reduction in tech companies hiring recent college grads across 2024.

During the commencement speech, Huang touted an upbeat outlook in terms of AI’s impact on the jobs market, suggesting that the technology will ultimately deliver broad benefits.

"You are entering the world at an extraordinary moment. A new industry is being born. A new era of science and discovery is beginning. AI will accelerate the expansion of human knowledge and help solve problems once beyond our reach," he told graduates.

"No generation has entered the world with more powerful tools, or greater opportunities, than you. We are all standing at the same starting line. This is your moment to help shape what comes next. So run. Don’t walk."

FOLLOW US ON SOCIAL MEDIA

Nvidia has announced a raft of open source initiatives at its 2026 GTC conference, which analysts have hailed as a sign of the company's long-term goals

Among the frameworks, models, and platforms introduced at Nvidia GTC, the one that’s received most attention is a new coalition of AI providers to drive development of open frontier AI models.

The Nemotron Coalition, as it has been named, includes providers such as Cursor, Mistral AI, Perplexity, and others, which Nvidia said will help “shape the next generation of AI systems”.

Each participating member will contribute their own “unique expertise” to the coalition, according to Nvidia, from data and evaluation frameworks to research and experts.

As part of the coalition, Nvidia announced plans to work alongside Mistral AI to co-develop a new base model, trained through NVIDIA DGX Cloud.

“The model will be shared with the open ecosystem and underpin the upcoming NVIDIA Nemotron 4 family of models,” the company said in a blog post announcing the move.

Charlie Dai, VP principal analyst at Forrester, said the move underlines the chipmaker’s shifting priorities, with a concerted focus now placed on interoperability amidst the rise of agentic and physical AI systems.

“Nvidia’s positioning of open models as a strategic enabler for agentic AI and physical AI reinforces the argument that openness is less about cost savings alone and more about interoperability, observability, and long-term control as AI systems become more autonomous, distributed, and operationally critical,” he said.

Dai noted that Nvidia’s approach on this front highlights the growing appeal of open source frameworks, which often act as a “catalyst for enterprise AI adoption” through “lowering barriers to experimentation, accelerating innovation, and enabling openness”.

“Enterprises increasingly rely on open models, frameworks, and tools across AI infrastructure, model development, agentic workflows, and governance while pairing them with commercial offerings to manage risk and scale,” he explained.

Research from the Linux Foundation in November 2025 noted the surging appeal of open source AI models for enterprises, which often represent a more cost-effective and flexible option for IT leaders.

Open agentic ecosystems

Agentic AI has also been an area of focus for Nvidia at GTC 2026. The company unveiled the launch of NemoClaw during the opening conference keynote on Monday.

Built on top of OpenClaw, an open source framework that allows users to create their own AI agents, and will allow users to draw upon an array of coding agents or open source models – including Nvidia’s own Nemotron open models – to build agents.

During his opening keynote, Nvidia CEO Jensen Huang said: “Every company in the world today needs to have an OpenClaw strategy [and an] agentic systems strategy.”

Indeed, Dai, noted the NemoClaw announcement reflects a broader shift from model-centric adoption toward agentic and agent-based systems for the chipmaker, again focusing specifically on interoperability for users.

“The value of agentic AI depends less on model novelty and more on execution layers, governance, observability, and operational fit,” he said.

“Nvidia’s OpenClaw narrative reinforces this view by focusing on control, security, and sustained inference rather than raw model scale, underscoring that agentic AI will drive durable compute demand while forcing enterprises to rethink architecture, cost models, and operational safeguards.”

FOLLOW US ON SOCIAL MEDIA

Dell Technologies has announced a slew of new end-to-end AI services and hardware solutions, including new support for enterprise-scale data processing and new server racks equipped with the latest Nvidia chips.

Announced at Nvidia GTC 2026, the new features for Dell AI Factory with Nvidia include expanded support for data analytics, processing, and parallel file storage.

Since it was first announced in 2024, Dell has launched over 240 products, updates, and releases within Dell AI Factory with Nvidia and more than 4,000 Dell customers have started to use the offering.

This includes the Wellcome Sanger Institute, a world leader in genomics, which has used the offering to go from generating three genomes over the course of a decade to one genome every seven hours.

“The Dell AI Factory with Nvidia provides a unified, vertically integrated data foundation that brings analytics as well as AI together in a single stack,” said Varun Chhabra, SVP, Infrastructure Solutions Group at Dell Technologies.

Dell has announced a range of new offerings within the AI Data Platform with Nvidia, which underpins the wider Dell AI Factory with Nvidia offering, including a new Data Orchestration Engine.

This is a no-code and low-code engine, a result of Dell’s December 2025 acquisition of the end-to-end AI development platform Dataloop.

Chhabra explained that it helps to automate data workflows for customers and improve visibility for data engineers and data scientists, to reduce the amount of manual data tagging and data pipeline creation that workers need to do.

Dell is also bringing new AI support across the Data Analytics Engine, which makes SQL data accessible via natural language processing (NLP), as well as support for Nvidia’s CUDA-X libraries within Data Search Engine and Data Processing Engine.

All told, Dell said the improvements unlock 12 times faster vector indexing and three times faster data processing. In internal tests with Qwen3-32B, the new features also enabled 19 times faster time to token.

Dell touts new storage offerings for the largest enterprises

Another major addition to Dell AI Factory with Nvidia is the Dell Lightning File System, a new tier 0, ultra-high performance storage offering intended for maximal GPU utilization and the lowest possible latency.

Dell Lightning File System delivers data transfer speeds of up to 150GB/sec per rack unit and is intended to complement Dell PowerScale and Dell ObjectScale for the most intense customer demands.

“The Lightning File System is really aimed at the largest of the large AI workloads, whether it's neoclouds, GPU as a service providers, or extremely high performance needs such as high frequency trading,” said Chhabra.

Dell said the Lightning File System offers twice the throughput of competing parallel file systems and 20 times greater performance than competitors using flash-only scale-out file systems.

Customers looking to launch large-scale AI storage clusters will also benefit from Dell Exascale Storage, a new software-defined, three-in-one architecture for deployments exceeding 10PB.

It brings together Dell PowerScale, Dell ObjectScale, and Dell Lightning File System, with up to 800 Gbit Ethernet network connectivity and 6TB/sec performance per rack performance.

“The goal here is this is not a new storage product, but a storage delivery vehicle, if you will, that customers can use to adapt different storage personalities over time without having to re-platform,” Chhabra explained.

“So if certain AI or HPC workloads are best suited for file data, they can deploy PowerScale on the Exascale Storage. If they are looking at modern applications or modern AI applications that require large-scale object storage deployment, they can deploy ObjectScale on the same cluster.

PowerEdge for frontier models

In addition to its Dell AI Factory with Nvidia announcements, Dell has used Nvidia GTC to unveil a slew of new hardware advancements.

First, Dell announced PowerEdge XE9812, a rack-scale system for the Dell Integrated Rack 9000 (IR9000) that comes packaged with the Nvidia Rubin NVL72.

Dell said the PowerEdge XE9812 can achieve 10 times lower cost-per-token compared to Nvidia Blackwell chips and needs four times fewer GPUs to inference a mixture of experts (MoE) model.

New NVLink capabilities within the rack allow it to achieve GPU to GPU bandwidth of up to 260TB/sec.

“The single rack has more bandwidth than the entire internet, which is kind of incredible,” Chhabra said.

The server is fanless, as it’s fully liquid-cooled, and quick disconnects for hot swaps and maintenance that doesn’t disrupt operations.

In addition to its latest frontier offering, Dell also announced new 100% liquid-cooled server offerings in the form of the PowerEdge XE9880L, XE9885L, and XE9882L.

“All of them have HGX Rubin GPUs and the difference between the systems is really the CPU choices,” Chhabra explained.

The XE9880L comes with dual-socket Intel Xeon CPUs, the XE9885L comes with next-generation AMD Venice CPUs, and the XE9882L comes with Rubin Arm-based CPUs.

These servers support up to 144 GPUs per rack, which are liquid cooled for higher performance per kilowatt. The servers house Nvidia HGX Rubin NVL8, which pack 8 Rubin GPUs in a 2U chassis interlinked with NVLink v6 connectivity and complemented with BlueField data processing units (DPUs) and Spectrum-X Ethernet switches.

Overall, Dell recorded a 5.5x performance improvement with this server family compared to B200 GPUs.

Dell also announced the Dell PowerEdge R9822 and M9822, which bring Nvidia’s Vera CPU to rack-scale environments.

“These two servers extend the power edge portfolio with support for the Vera processors, as I mentioned, which enable organizations to deploy modern ARM based compute on trusted power edge infrastructure,” said Chhabra.

He added that the servers are intended to support AI factories, high-bandwidth data analytics, databases, cloud-native applications, and high-performance computing (HPC) workloads.

Finally, Dell is launching the PowerEdge R770, R7725, and R7715, which feature Nvidia RTX Pro 4500 Blackwell GPUs for three times faster data processing than L4 tensor core GPUs, four times faster visual compute and content creation than the previous generation.

PC announcements for edge AI

addition to Nvidia’s Grace Blackwell Ultra GB300 Superchip, the workstation comes with 496GB LPDDR5X CPU memory and 252GB HBM3e GPU memory.

It is capable of 20 petaflops of FP4 computing power, enough to run trillion-parameter models entirely at the edge.

Jon Siegal, SVP of Product Marketing at Dell Technologies, said the Dell Pro Max with GB300 is “purpose-built” to meet next-generation agentic AI workloads.

Dell is also bringing back the Dell Pro Precision workstation range, with a range of offerings designed to run trillion-parameter models locally for specialized industries such as healthcare, deep tech, and financial services.

In response to a question from ITPro, a Dell representative also indicated that the firm is seeing some customers fine tuning trillion-parameter models into billion-parameter small language models (SLMs).

Dell’s mobile workstations are now thinner, lighter, and more powerful. The Dell Pro Precision 5 series, available the latest Intel Core Ultra H processors and Nvidia RTX Pro discreet graphic options, delivering a blend of powerful graphics performance and up to 50 TOPS of AI processing capability.

The Dell Pro Precision 7 series is aimed at designers and creators, with RTX Pro 3000 discrete graphics in addition to the Intel chips.

FOLLOW US ON SOCIAL MEDIA

HPE has announced a major expansion of its Private Cloud AI portfolio, with a particular focus on data sovereignty capabilities.

Unveiled at Nvidia GTC in San Jose, California, updates to the service will include an air-gapped configuration for HPE Private Cloud AI. This will allow enterprises to ensure data isn’t exposed to external networks and kept in a secure, isolated location.

In a press briefing ahead of the annual conference, Dale Brown, head of growth for AI solution at HPE, said the move comes in direct response to the growing enterprise focus on sovereignty and security, as well as evolving regulatory compliance requirements.

“We’re meeting customers where they are,” he said, adding that there are three overarching considerations at play for organisations with regard to sovereignty.

“The interesting thing about sovereignty, if you notice, sovereignty always has to be put in context of customer need and use-cases,” he told assembled media.

“So security might mean completely disconnected from the internet, but also sovereignty might mean that the data can’t leave a particular country and/or a particular principality.”

Brown added that the solution “provides even extra security” through a private cloud model, “bringing AI to the data” but also offering enterprises a higher degree of control of how data is processed, and where.

HPE Private Cloud AI capacity boost

HPE also announced new capacity scaling capabilities for Private Cloud AI customers at Nvidia GTC. This includes the ability to scale network racks up to 128 GPUs, marking an increase from the previous limit of 64.

The company said this is aimed primarily at allowing customers to accommodate “larger, more demanding AI workloads” while still maintaining a “consistent operational experience”.

Again with a focus on sovereignty, HPE confirmed that the Private Cloud AI service, as well as HPE ProLiant servers and HPE AI Factories, will support the latest Nvidia Nemotron open models.

These form part of Nvidia’s Agent Toolkit, which helps streamline the deployment of agents in on-prem and sovereign infrastructure environments for enterprises operating in critical or regulated industries.

Brown told journalists that HPE is also adding a raft of new “additional configurations and deployment modalities” for Private Cloud AI customers.

The service includes access to an array of Nvidia AI Enterprise software and blueprints, for example, including Nvidia’s updated AI-Q blueprint for AI agents.

New Nvidia Omniverse blueprints for digital twins are also available, the company revealed.

“Ultimately, our partnership with Nvidia has been all about removing friction, repeatable patterns, speeding time to value with proven best tools and techniques between our companies,” he said.

“We’re super proud of how we’ve co-engineered these solutions consistently.”

FOLLOW US ON SOCIAL MEDIA

Women are still greatly underrepresented in STEM. The latest figures from the Women Tech Network show women only make up around 26-27% of the STEM workforce and the organization estimates that at the current rate of change, it will be nearly 123 years before the economic gender gap is closed.

It’s International Women’s Day 2026 – and the rate of women hired in tech continues to lag far behind that of men.

As hiring teams turn to AI tools to automatically field candidates, we’re at something of a crossroads – do we fix the bias inherent in hiring? Or simply replicate it at scale with similarly biased AI tools?

In this episode, Jane are Rory are joined by Clare Hickie, EMEA CTO at Workday, to discuss how businesses can engage in bias-free talent acquisition in the age of AI.

Highlights

"It's not a problem if you control it, and the only way that you can control it is to ensure that you have got a responsible AI program, the right measures, the right procedures, the right guardrails and the right protocols in place."

"I've been in technology a long, long time, and it's now it's my job to really pay back. And my job in paying back is by ensuring that I can go out and really encourage and inspire young people into technology. So there's a couple of things going on."

"We've seen a negative impact on society as a consequence of AI as well, right? But what we're really are driving for is that positive impact on society, and the only way you can do that is removing the bias of the past. And the way you remove that biased information is to use the technology to also ensuring that you're operating with clean data and with context.

"You have to apply fair processes. You have to apply, apply fair procedures. So in the past, organizations, for example, didn't even have to post that there was another job, it was a tap on the shoulder after another tap on the shoulder. And sometimes, you know what? It worked and then for many they were just not even considered, because they were just forming these groups and teams and leadership teams that they were all from themselves.

Footnotes

• International Women’s Day 2026: why are there still so few women working in cybersecurity?
• Gender diversity improvements could be the key to tackling the UK's AI skills shortage
• Why are so many men in tech blind to the gender divide?
• Is diversity still a challenge in the channel?

Subscribe 

• Subscribe to The IT Pro Podcast on Apple Podcasts
• Subscribe to The IT Pro Podcast on Spotify
• Subscribe to the IT Pro newsletter
• Join us on LinkedIn

February is the shortest month, but you wouldn’t know it from the sheer amount of news that’s broken in just the past 26 days.

Amid growing fears of AI stealing jobs, OpenAI’s CEO Sam Altman has claimed that firms are simply using the technology as an excuse for mass layoffs. Earlier this month, a series of Anthropic releases drove stocks down at a range of companies – all tied to fears that the SaaS model might be on its way out.

Also this month, Pure Storage has rebranded as Everpure – what’s behind this decision and what does it say about the company’s strategy going forward?

In this episode, Jane and Rory welcome back Ross Kelly, ITPro’s news and analysis editor, to explore some of February’s biggest stories.

Highlights

"The central concern here is that a lot of AI companies, particularly Anthropic and OpenAI etc, are now launching sector-specific tools that are very proficient here, to the point where a lot of customer service, or legal industry tasks can be conducted by agents, automated. This has a lot of investors concerned about the long-term impact on the software as a service industry – you have a lot of, again, sector-specific software companies, very specific tools based on particular tasks within multiple industry verticals."

"These are comments that Sam Altman made at the New Delhi AI Impact Summit last week, essentially around the fact that a lot of companies are 'AI washing'. Which, roughly translated, means they're using AI as an excuse for layoffs, when the reality is that there's potentially poor financial performance there or ulterior motives."

"So this ties in with a big acquisition here from Everpure, formerly Pure Storage of 1touch, which is a data orchestration and management platform. The rebrand here is interesting because, you know, I attended Pure Accelerate 2025, Pure Storage's annual conference, and a big focus there was around, the data management aspect of what they're looking at now with their platform approach. They launched their enterprise data cloud, which again, is an all-encompassing ecosystem."

Footnotes

• Sam Altman just said what everyone is thinking about AI layoffs
• Why Anthropic sent software stocks into freefall
• Anthropic says Claude Code can help streamline 'cost-prohibitive' COBOL modernization, but IBM says it's not that simple
• What might cause the 'AI bubble' to burst – and what impact would that have on the business world?
• Pure Storage snaps up 1touch in data management pivot

Subscribe 

• Subscribe to The IT Pro Podcast on Apple Podcasts
• Subscribe to The IT Pro Podcast on Spotify
• Subscribe to the IT Pro newsletter
• Join us on LinkedIn

Generative AI is transforming the traditional computing stack, according to Nvidia CEO Jensen Huang, who says it will eventually form the core in a new era of software development.

Speaking with BlackRock CEO Larry Fink at the World Economic Forum (WEF) in Davos, Huang compared the impact of AI on modern computing to the advent of the web, the personal computing boom, and cloud computing.

“It’s helpful to reason back to the first principles of fundamentally what is happening to the computing stack. This is a platform shift,” he said.

“New applications were developed to run on a new type of computer. [With the] platform shift to the internet, a new type of computing platform hosted all kinds of new applications.”

With each of these transformations, Huang noted, the “computing stack was reinvented and new applications were created”.

Generative AI essentially marks the latest step on this evolutionary path, with Huang noting that ChatGPT and Anthropic’s Claude chatbot will be the tip of the spear that delivers new applications and tools in the future.

“New applications will be built on top of ChatGPT,” he added. “New applications will be built on top of Anthropic Claude, for example. So it’s a platform shift in that way.”

AI is transforming software

Later in his conversation with Fink, Huang noted that AI is “really easy to understand” when looking at the technology through this prism.

AI is helping redefine not only what constitutes modern software, but how it’s built, as enterprises flock to the technology to transform processes and streamline operations.

Analysis from Bain & Company shows AI is already transforming areas such as software as a service, with the arrival of agentic AI further bolstering efficiency improvements and speeding up development.

The key difference in this case is that while software in the past was “effectively pre-recorded”, AI tools rely on prompts, marking a step change in how users interact with them. They’re more intuitive, proactive, and are capable of discerning context with minimal information.

From a user efficiency perspective, this is a marked improvement when cumbersome software often relied on specific structured data to function.

“Software in the past was effectively pre-recorded. Humans would type and describe the algorithm or the recipe for the computer to execute,” he explained. “It was able to process structured information, meaning you’ve got to put the name, the address, their account number, their age, where they live.”

“You create these structured tables that software would then go and retrieve information from,” Huang added. “We call it SQL queries. SQL is the single most important database engine the world's ever known. Almost everything ran on SQL before.”

Because AI can understand unstructured information, allowing it to digest text, images, or sound, Huang said AI represents a far more convenient mode of human-computer interaction.

“For the first time, we now have a computer that is not pre-recorded, but it's processed in real-time,” he commented. “Describe it however you like to describe it, and to the extent that it can understand your intention, it could perform a task for you.”

A “five-layer cake”

Huang noted that the foundations upon which AI rests are crucial. It’s a delicate balancing act in which multiple “layers” need to function in tandem for the technology to operate efficiently.

This aspect adds a degree of complexity that previous technological shifts were typically exempt from. Simply put, it’s not just a simple click-and-go technology operating independently. The infrastructure stack that underpins AI is vital.

“When you think about AI, you think about the AI models, but it’s really important to understand industrially, AI is essentially a five-layer cake - at the bottom is energy,” he said.

“AI, because it's processed in real time and it generates intelligence in real time, it needs energy to do so. Energy is the first layer. The second layer is the layer that I live in. It's chips, chips and computing infrastructure.

“The next layer above it is the cloud infrastructure, the cloud services. The layer above that is the AI models.”

According to Huang, there’s a common misconception that the nitty gritty of the technology happens in the model domain. However, while this is crucial, it’s in the application layer above that where real value and economic growth are generated.

There are challenges here though, Huang admitted. Model performance improvements have outpaced application development, meaning organizations are struggling to unlock value from the technology.

“The reason why last year was an incredible year for AI is that the models made so much progress than the layer above it, which is ultimately the layer that we all need to succeed.”

“This application layer could be in financial services,” Huang added. “It could be in healthcare. It could be in manufacturing. This later on top, ultimately, is where economic benefit will happen.”

FOLLOW US ON SOCIAL MEDIA

Unless you've been living under a rock for the past 18 months, it's impossible not to be aware of how many of its eggs Nvidia has put in the AI basket.

To further stoke the AI fire, Nvidia released a dedicated desktop box aimed at AI developers called the DGX Spark, which combines Nvidia's latest generation of ARM CPU, GPU, and networking technologies in one user-friendly and affordable package.

The phrases "user-friendly" and "affordable" are doing a lot of lifting in that previous paragraph, but given the lack of alternatives if you want to develop your AI project locally, this is as good as it gets.

Nvidia has also partnered with several OEMs to bring variations on the Spark theme to market. All the OEM boxes share the same basic specification as the Nvidia Ur-Box, suggesting Nvidia is keeping everyone on a pretty tight leash.

Dell has released the Dell Pro Max with GB10, Gigabyte the AI Top Atom. Lenovo with the ThinkStation PGX, MSI with its EdgeXpert MS-C931, and finally Asus with the Ascent GX10.

It's the Asus Ascent GX10 we are looking at today. All the variations on the DGX Spark will set you back around £4,000, depending on the amount of storage it comes with; the Ascent we are poking with a stick can be picked up in the UK for £3,640.

Asus Ascent GX10: Design

Given the esoteric hardware that lurks inside the Ascent GX10 is impressively small, stylish, and unobtrusive. It also has one major advantage over Nvidia's own box and the other OEM competition in that the power switch is at the front rather than the back or, in Dell's case, the top.

That may not be a big deal if you just have one box sat next to your monitor, but if you have several arranged in a rack, and that is very much how this system is intended to be used, then you'll be left groping around the back to power your system up.

Physically, the Ascent GX10 is a 150 x 150 x 51mm metal box weighing 1.48Kg. The front, back, and side panels are painted in matte slider while the top has a cross-hatched mirror finish.

The wavey slats on the front presumably provide more surface area for cooling, but as essentially an aesthetic consideration. Compared to the rather baroque DGX Spark and the utilitarian offerings from the other OEMs, the Asus box is by far the most visually appealing.

To be fair to Nvidia, the external design of the DGX Spark is very much an homage to the 2016 DGX1, a $130,000 61Kg lump the size of a large suitcase with a paltry-by-modern-standards 170TFLOPS capability.

All the ports are grouped at the rear below the large air vent. Running left to right, you'll find a Kensington lock, four 20Gbps (Gen 2×2) USB-C ports, one of which serves as the connection for the 240W power brick, an HDMI 2.1 video output, 10GbE Ethernet, and two 200Gbps ConnectX-7 QSFP112 ports.

Incidentally, the three USB-C ports that are not powering the box also support DisplayPort 2.1 DP alt-mode video output. Another consideration is that the GB10 chip has a TDP of 140W, which leaves 100W for all the other bits and bobs, including the ConnectX-7 NIC and the three USB ports. Wireless communications are handled by a MediaTek card that supports Wi-Fi 7 and Bluetooth 5.4.

Internal access is minimal, but given that everything important is soldered in place, this is not the end of the world. Unscrew the base grille and then another metal plate, and you can easily get at the 2242 SSD, which means you can buy the cheapest version of the box and upgrade the storage should the need arise.

Dig deep into the Asus product sheets, and you'll find that only the 4TB configuration ships with a PCIe 5.0 x 5 SSD. The smaller SSDs are PCIe 4.0 x 4 specification.

Like the Nvidia DGX Spark and all the other OEM iterations thereof, the Ascent GX10 runs on NVIDIA's DGX OS, a variant of Ubuntu 24.04 LTS (for ARM) Linux.

The primary differences between DGX OS and vanilla Ubuntu are really just a matter of the apps Nvidia has added, such as the dashboard to manage your AI projects, Nvidia Sync utility that manages back-end SSH keys and tunnelling, and all the relevant drivers.

The OS dashboard also directly links to Nvidia's DGX Spark Playbooks repository on GitHub. These provide a great selection of programmes for getting a wide variety of AI models up and running, including ComfyUI, Pytorch, and vLLM, to name but three.

These playbooks are an important part of the GX10's appeal to those who are rather new to the world of AI development, and one reason why the GX10 or any of its Spark cousins are perhaps a better bet for AI developers than the likes of a Mac Studio or a do-it-yourself rig built around an Nvidia discrete GPU or two.

Asus Ascent GX10: Specs and Performance

Running the show inside the GX10 is an Nvidia GB10 ARM v9.2-A CPU, which consists of 10 Arm Cortex-X925 and 10 Arm Cortex-A725 cores, a Blackwell GB20B GPU. You also get 128GB of soldered LPDDR5X-8533 RAM, a brace of ConnectX-7 200Gbps networking ports, and a 4GB 2242-format SSD.

Versions with a 1GB or 2GB SSD are also available for rather less money, but the majority of the boxes on sale through retail channels are the 4GB variety, and given the intended use, that would surely be the model to go for.

Before we press on, it's worth making clear that one reason these Nvidia-system boxes are so expensive is because of that Nvidia ConnectX-7 interface, which easily accounts for upwards of one-third of the asking price.

It goes without saying that 200Gbps of bandwidth is nothing like what you get from a full-fledged GB200/GB300 server, but it does provide developers with access to more power and also a way to see how their models perform on a scaled setup.

Measuring performance is borderline superfluous on a machine that boasts 1 petaFLOP (that's 1,000 TFLOPS in old money) of performance, and our testing regime isn't suited to measuring the exact potential of GX10 to run large LLMs and VLMs at the same time or create local AI agents.

Asus's demonstrations undoubtedly confirmed Nvidia's performance claims, such as the impressive abilityfine-tuningning Pytorch LoRA llama3 8B at 53K peak tokens per second (TPS) and generating a 1K FP16 image in Flux1.dev 12b in double quick time thanks to having ready access to the 35GB of RAM that the job required.

In a rather unscientific test, we ran some basic prompts in LM Studio on both the GX10 and the Framework Studio. The Asus box averaged 4.67 tokens per second while the Framework came in at 13.4 TPS. Of course, our Framework box only has 32GB of RAM.

Running some 1K image creation tasks in ComfyUI gave the Asus box an even larger advantage, the GX10 taking 216 seconds to do a job that the Framework box took over 400 seconds.

At a more general level, the GB10 CPU scored 3,104 in the GeekBench 6 single-core benchmark and 20,048 in the multi-core test. For comparison, a 2024 Mac Mini running on the M4 Pro CPU scored 3,986 and 22,857, respectively, while the AMD Ryzen AI Max 385 CPU in the Framework Desktop scored 2,952 and 16,599.

Throughout all the test the GX10 impressed by how quiet it was, the dual fan system seldom making much more than a loud whisper. Under heavy stress, the external case got a little warm in places, but by no more than your average laptop when under duress.

Our takeaway is that if you are developing large AI frameworks and want to do so locally, then the GX10 and its relatives are the smallest, fastest, and most convenient way to do it.

Granted, you could build yourself a development rig based on a couple of Blackwell gaming GPUs and potentially get more performance, but the result would be bigger, louder, far more power hungry, and probably more expensive when you factor in the cost of RAM, a CPU, a box, and everything else, let alone a comparable network card.

Asus Ascent GX10: Is it worth it?

With an item as esoteric as the Ascent GX10, that's not an easy question to answer. For a substantially smaller pile of cash, you can buy an Apple Mac Studio or a system built around one of AMD's Ryzen AI Max chips, like the Framework Desktop modular PC we looked at recently or HP's new Z2 Mini G1a Workstation, both of which can be configured with 128GB of RAM.

Of course, none of those machines have two 200GbE OSFP112 networking ports, and that's the first area in which the GX10 excels: scalability. Buy two of these things and hook them together, and you will really be cooking with gas. Though you'll also be over £7,000 out of pocket.

The value of the various Nvidia AI playbooks shouldn't be overlooked either. Having some GB10-specific workbooks is something we think will prove useful to developers wanting to further hone and expand their AI capabilities.

The Ascent GX10, then, is very much a thoroughbred designed for a specific job. Many developers will be happy with one of the forthcoming AMD Ryzen Max+ boxes, which are not only cheaper but run Windows 11, making them more versatile, but if you are deadly serious about your AI development and actually need the performance that more than one GX10 box can deliver, then it's a pretty impressive device for the asking price.

Asus Ascent GX10 Specification

CrowdStrike, AWS, and Nvidia have picked 35 startups for their Cybersecurity Startup Accelerator, saying they've been chosen for their innovation, their potential to create market impact, and the caliber of their teams.

“This year’s cohort reflects a global movement: founders building cloud- and identity-first defenses that put security teams ahead of the speed and scale of AI-emboldened adversaries," said Daniel Bernard, chief business officer at CrowdStrike.

"With AWS and Nvidia, we’re creating a community and growing 'the crowd', giving these start-ups the opportunity to turn breakthrough ideas into market-shaping technologies, and push the industry forward.”

The free, eight-week program runs from now until March 3, offering mentorship, technical expertise, funding, and go-to-market support along with access to top cybersecurity experts and global visibility across partner ecosystems.

The program will come to a head with a final pitch day for five finalists during the RSA Conference in San Francisco on March 24, where an expert panel will select one innovation award winner, which will potentially win investment from the CrowdStrike Falcon Fund.

"AI is reshaping cybersecurity at every level, demanding new approaches that can operate at cloud scale and defender speed,” said Bartley Richardson, senior director of agentic AI and cybersecurity engineering at Nvidia.

“Through the accelerator, Nvidia, AWS, and CrowdStrike are empowering start-ups with the compute, frameworks and guidance they need to advance agentic AI and build the next wave of intelligent, resilient security technologies.”

The companies expanded the program in November, opening it to startups worldwide. The 2026 cohort includes firms working on AI-driven threat detection, cloud workload protection, identity security, and secure data handling, from North America, Europe and other regions.

Now in its third year, the Accelerator has graduated 59 startups in total, which between them have collectively raised more than $730 million in funding following their participation.

Nvidia’s support comes via its Inception program, which supports companies through critical stages like product development, prototyping, and deployment.

Members get a personalized range of benefits, including Nvidia Deep Learning Institute credits, preferred pricing on select Nvidia hardware and software, and guidance from technical experts.

Last year's Accelerator winner was Terra Security, picked for its work in agentic AI-powered, continuous web application penetration testing, while Remedio was the winner of the 2023–2024 cohort – both firms raised sizeable funding rounds following the program.

"Participating in the AWS CrowdStrike Accelerator Program has truly been a wonderful experience for Remedio. The invaluable content and strategy sessions have significantly boosted our company's growth," commented Tal Kollender, CEO and co-founder of Remedio.

"What's been most rewarding on a personal level is the exceptional mentorship and the opportunity to connect with industry leaders I've long admired. It's surreal to witness their genuine interest and eagerness to help."

FOLLOW US ON SOCIAL MEDIA

The southern French city of Grenoble has been chosen by HPE and Nvidia to host the first AI factory lab in the European Union (EU).

The announcement comes just ahead of HPE’s annual Discover Barcelona conference, which kicks off on Wednesday 3 December, and is the latest in a number of joint AI factory projects from the two companies.

The AI factory lab in Grenoble will allow customers to test and refine workloads on a sovereign AI factory environment on hardware located and running in the EU, the companies said.

The lab features Nvidia AI Enterprise government-grade software and Spectrum-X Ethernet networking, HPE servers and Alletra storage, and PTX and MX Series routers from Juniper Networking (now part of HPE).

Private AI in the UK

Not to be left out, HPE is also launching a similar lab in the UK, in partnership with Carbon3.ai. The Private AI Lab, which will be located in Manchester, is built on HPE Private Cloud AI with Nvidia AI Enterprise Suite software and is intended, HPE said, to "accelerate UK enterprise AI adoption".

Tom Humphreys, CEO of Carbon3.ai, said: "UK enterprises want secure, sovereign, and sustainable AI infrastructure they can trust – and they want real outcomes fast. By removing cost and risk, and by grounding everything in sovereign, sustainable UK infrastructure, we're accelerating enterprise adoption and helping the UK convert its AI potential into economic impact."

The sovereign aspect of this hinges on Carbon3.ai's Private AI Lab offering UK compute and Nvidia infrastructure "in a secure, compliant, and scalable environment that aligns with existing cloud and governance standards and regulatory obligations", powered by 100% off-grid renewable power.

Matt Harris, SVP and MD for UK, Ireland, Middle East, and Africa at HPE, added: "UK organisations face unique AI infrastructure challenges – from strict data residency and regulatory requirements to the need for scalable, energy efficient AI infrastructure and a clear path from pilot to production.

“The Private AI Lab addresses these needs directly. It provides a fast, low-risk route to production-grade AI that preserves sovereignty, reduces operational complexity and supports more energy-efficient deployment at scale.”

Carbon3.ai Private AI Lab is available immediately to qualified UK businesses. The AI Factory Lab in Grenoble will be available in Q2, 2026.

Nvidia has announced a slew of supercomputing and AI factory announcements, with new hardware adoption and high-performance computing (HPC) plans laid out in line with partner organizations.

Timed to coincide with the start of Supercomputing 2025 (SC25), the annual HPC event held this year in St. Louis, Nvidia unveiled new orders for HPC and supercomputer sites, alongside a series of agreements for major partners to use its network switches.

First, Nvidia announced it is partnering with Japan’s public research institute Riken on two new supercomputers powered by GB200 systems. The first of these systems will combine traditional supercomputing and AI to tackle issues in manufacturing, robotics, seismology, and quantum research.

It will be powered by 1,600 Nvidia Blackwell GPUs, connected by Quantum-X800 InfiniBand networking and built on the GB200 NVL4 platform.

The second system will aim to combine GPUs with Riken’s quantum HPC hybrid infrastructure. Powered by 540 Blackwell GPUs with the same compute platform and networking technology, the aim here is to run quantum algorithms and hybrid computing simulations.

“Integrating the NVIDIA GB200 NVL4 accelerated computing platform with our next-generation supercomputers represents a pivotal advancement for Japan’s science infrastructure,” said Satoshi Matsuoka, director of the Riken Center for Computational Science.

“Our partnership will create one of the world’s leading unified platforms for AI, quantum and high-performance computing, allowing researchers to unlock and accelerate discoveries in fields ranging from basic sciences to industrial applications for businesses and society.”

Riken’s petascale Fugaku supercomputer was the fastest in the world between June 2020 and May 2022. It now ranks seventh in the world per the latest top500 list.

The two new systems announced today will act as development platforms to inform the creation of FugakuNEXT, due online at the end of 2029.

This next-generation super computer will aim to combine traditional supercomputing with quantum computing and bring together Fujitsu Monaka X CPUs and Nvidia architecture via new NVLink Fusion silicon.

Nvidia has recently announced plans to build two new supercomputers for the US Department of Energy (DoE), known as Solstice and Equinox. Harris explained that Solstice will be powered by 100,000 Blackwell GPUs, delivering 2,000 exaflops of AI performance – “more AI flops than the entire top500 combined”.

Nvidia touts open AI models for scientific computing

Another major announcement made by Nvidia is the launch of Apollo, a new family of AI physics models aimed at helping developers run complex industrial simulations in real-time.

The model family unlocks capabilities across a wide range of scientific applications, all with real-world use cases for enterprises backed by Nvidia’s AI infrastructure.

These include sector-specific simulations such as structural analysis across automotive, electronics, and aerospace, as well as electronic defect direction, mechanical design, and computational lithography.

Apollo is also capable of intense, scalable simulations including climate and weather modelling, computational fluid dynamics, real-time simulation of electromagnetic signals to recreate wireless communications and optical data, and nuclear physics simulations.

Some Nvidia partners have already been using Apollo to improve their processes. For example, the semiconductor giant Applied Materials – the second largest chip equipment supplier after ASML – has used Apollo models to improve the power efficiency of its manufacturing process.

Running the models on Nvidia GPUs via its CUDA framework, the firm has unlocked flow, plasma, and thermal modeling of its semiconductor process chambers, combining AI and digital twins for an overall acceleration of 35x across its ACE+ multiphysics simulation platform.

Northrop Grumman and Luminary Cloud have also used Apollo models to design spacecraft thruster nozzles more efficiently.

Apollo joins Nvidia’s existing open model families announced at GTC 25 including Clara for biomedical AI, Cosmos for physical AI, GR00T for robotics, and Nemotron for agentic AI.

New adoption for Nvidia’s networking technology

The Texas Advanced Computing Center, AI cloud computing firm CoreWeave, and AI factory startup Lambda will adopt Nvidia Quantum-X Photonics in their upcoming data center infrastructure.

Nvidia has said the co-packaged optics networking switches are capable of 115.Tbits/sec networking, connecting millions of GPUs while improving application runtime by 5x and power efficiency by 3.5x.

“Nvidia Quantum-X photonics will help them realize more robust, serviceable, and energy efficient network fabrics that keep large-scale workloads running fast and efficiently,” said Dion Harris, senior director of HPC and AI Infrastructure Solutions at Nvidia.

In an example, Harris explained that new networking technologies must pair with GPU breakthroughs to power the latest mixture-of-experts (MoE) models such as DeepSeek R1.

Nvidia’s answer is the Blackwell NVL72, in which 72 GPUs are connected to one another with all-to-all, 130 TB/sec bandwidth NVLink switches. In SemiAnalysis data on DeepSeek R1, the NVL72 performed at 10x better performance per dollar, performance per watt, and throughput – for an overall 10x improvement in revenue.

“You can't do this with just chip innovation by itself,” said Harris. “We achieve this through extreme co-design with Innovation and optimisation. Optimisation across gpus CPUs memory. Networking rack scale architectures as well as software.

In the year since SC24, Harris said 80 new systems built on Nvidia’s systems have been announced with 300,000 GPUs committed and the equivalent of 4,500 exaflops of AI performance planned.

Arm is also becoming a CPU partner for NVLink, alongside existing partners such as Fujitsu, Intel, and Qualcomm.

At the same time as it expands its traditional networking portfolio, Nvidia has also announced new adoption roadmaps for its NVQLink and CUDA-Q platforms, which will connect GPUs to quantum processors in the near future with a latency of less than four microseconds.

At SC25, Nvidia has announced 21 supercomputing centers around the world will adopt NVQLink including:

• Australia’s Pawsey Supercomputing Research Centre
• DCAI, operator of Denmark’s Gefion supercomputer.
• Germany’s Jülich Supercomputing Centre (JSC).
• Japan’s Global Research and Development Center for Business by Quantum-AI technology (G-QuAT), at the National Institute of Advanced Industrial Science and Technology (AIST).
• The Korea Institute of Science and Technology Information (KISTI).

In the US, many of the DoE’s national laboratories have also signed on including Lawrence Berkeley National Laboratory, Los Alamos National Laboratory, and Oak Ridge National Laboratory (which currently hosts the supercomputer Frontier).

“Performing tasks like quantum error correction requires a Quantum GPU interconnect with microsecond latency and throughput in the hundreds of gigabits per second,” Harris said.

“It's also essential for a scalable interconnect platform to be open and universal supporting all types of quantum processor builders, so that every supercomputer has a turnkey solution to bring quantum into their workflow.”

MORE FROM ITPRO

• Nvidia, Deutsche Telekom team up for "sovereign" industrial AI cloud
• Jensen Huang says future enterprises will employ a ‘combination of humans and digital humans’
• Nvidia’s Intel investment just gave it the perfect inroad to lucrative new markets

SC25 – the annual global supercomputer conference – is taking place this week in St Louis, Missouri and Dell Technologies has used the occasion to announce a slew of new AI-focused products and updates.

In its PowerEdge server range, the AMD Instinct-powered XE9785 and XE9785L – both of which were first announced at Dell Technologies World 2025 – are now generally available. The company also introduced a new Intel-powered PowerEdge server, the R770AP.

All of these servers, Dell Technologies said, are optimized for working with AI workloads, with the XE9785 and XE9785L capable of “breakthrough performance with AI training and inference”. The R770AP, meanwhile, can “handle demanding HPC and AI workloads”.

The company has also introduced two new switches, the Dell PowerSwitch Z9964F-ON and Dell PowerSwitch Z9964FL-ON. According to the firm, these switches are able to deliver 102.4 Tb/s of switching capacity, which means they can “meet the needs of the most intensive AI fabrics”.

AI Factories and Nvidia partnership boost

Dell Technologies also announced updates to its AI Factory offering, which is underpinned by Nvidia hardware.

AI factories have been a cornerstone of Dell’s AI strategy since 2024, with Nvidia chips underpinning it from the start. The idea of these factories is in essence a bundled offering, with Dell Professional Services helping deliver the ‘factory’ to the buyer rather than them having to put it together themselves.

In the announcements from SC25, Dell Technologies revealed its ObjectScale and PowerScale storage devices are now integrated with the Nvidia NIXL library. NIXL library is in turn part of Nvidia Dynamo, described by Nvidia as “a distributed inference-serving framework” built to deploy and automate AI models at “data center scale”.

Additionally, Dell AI Factory with Nvidia also now includes Dell PowerEdge XE7740 and XE7745 servers, both of which feature the chip-maker’s RTX Pro 6000 Blackwell Server Edition GPUs and Hopper GPUs.

Finally, the Dell Automation Platform is also available on Dell AI Factory with Nvidia and, according to Dell Technologies, “will deliver smarter, more automated experiences by deploying validated, optimized solutions with a secure framework”.

All of these products are available immediately.

Microsoft has teamed up with Nvidia and WeTransact to launch the ‘Agentic Launchpad’ – a new program aimed at helping UK software companies build next-generation AI platforms.

As part of the scheme, successful startups and scale-ups will be able to work directly with Microsoft and Nvidia, gaining access to advanced AI expertise, technical sessions, and opportunities for joint innovation.

This will include a range of events, workshops, and summits, including the Microsoft AI Tour and industry briefings.

Marketing support is a key focus of the scheme, helping startups to access broader markets through media campaigns and executive sponsorships, the duo said.

Long-term, the aim here is to improve time-to-market and visibility. Microsoft said participants will be able to promote their agentic AI apps to key stakeholders and target customers through dedicated campaigns. They'll also feature in communications, blogs, and press releases from Microsoft and its partners.

“This is a really exciting opportunity for innovative and ambitious AI-first software companies to accelerate the growth of their agentic AI solutions with the expert help of Microsoft and Nvidia,” said Darren Hardman, CEO, Microsoft UK & Ireland.

“We are passionate about helping to build the agentic AI future and see innovation flourish in the UK.”

Nvidia to beef up training resources

Nvidia has pledged additional support for the program by giving access to its existing training resources for startups and smaller enterprises.

“Nvidia is ready to support the Microsoft Agentic Launchpad through the Nvidia Inception program for cutting-edge start-ups,” said Serge Palaric, Nvidia vice president of EMEA hyperscaler and ISV alliances.

“This collaboration will provide start-ups with access to Nvidia Deep Learning Institute training sessions and a network of experts. By using Nvidia AI infrastructure, software companies will be able to accelerate their AI development and enhance performance.”

The program builds on Microsoft's GenAI Accelerator, launched this time last year in collaboration with Nvidia and GitHub. Participants were given free Azure credits, membership of NVIDIA’s Inception program, and support from Microsoft’s AI ‘Black Belt’ engineers.

“The GenAI Accelerator provided invaluable access to technical advisors, partner managers, and experts at Microsoft,” says Sharad Khandelwal, cofounder and CEO of SentiSum.

“We gained visibility in critical AI and SaaS communities, and it gave us the credibility to partner with enterprise customers who value Microsoft’s stamp of quality.”

MORE FROM ITPRO

• UK's AI sector booms – but can the country hang on to its startups?
• 3 ways AI helps startups grow
• Cash is flowing into AI startups, but investors have warned of the rise of ‘zombiecorns’

Not content with a new supercomputer and AI cluster at Oak Ridge National Laboratory, the US Department of Energy (DoE) has put in an order for a further two, this time to be housed at Argonne National Laboratory (ANL).

The two AI supercomputers, named Solstice and Equinox, will be created by Oracle and Nvidia and have been tipped as a future “powerhouse for scientific and technological innovation” by US secretary of energy Chris White.

Solstice is the bigger of the supercomputers, with what Nvidia claims is a “record-breaking” 100,000 Blackwell GPUs. Equinox, meanwhile, will feature 10,000 Blackwell GPUs and will be the first to be completed, with an estimated delivery date of H1 2026.

There’s no date given for Solstice, but once it comes online the two will be connected to Equinox via Nvidia networking to deliver an expected 2,200 exaflops of “AI performance”, according to Nvidia. This would make the cluster more powerful than Frontier – the first exascale supercomputer – or the current most powerful supercomputer, El Capitan.

This will be a combined output, however, rather than per supercomputer and it’s unclear at this point whether it’s Rmax, a verified output achieved on the HPL benchmark, or Rpeak, a theoretical maximum output.

ITPro approached Oracle for more technical details of its involvement, but did not receive any information at the time of publication.

Commenting on the announcement, Paul Kearns, director of ANL said: “The Equinox and Solstice systems are designed to accelerate a broad set of scientific AI workflows, and we are collaborating with Oracle and NVIDIA to prepare thousands of researchers to effectively leverage the systems’ groundbreaking capabilities.

“This system will seamlessly connect to forefront DoE experimental facilities such as our Advanced Photon Source, allowing scientists to address some of the nation’s most pressing challenges through scientific discovery.”

DoE supercomputer tsunami

Solstice and Equinox aren’t the only supercomputers to be announced by the DoE this week. On Monday 27 October, HPE revealed it was building a new supercomputer for the agency, called Discovery, at Oak Ridge National Laboratory (ORNL) in collaboration with AMD. This is in addition to an AI cluster called Lux that will also feature AMD chips and networking.

Similarly, Nvidia announced a collaboration with HPE to create another two supercomputers – Vision and Mission – for the DoE’s Los Alamos National Laboratory.

Like Discovery, this will be based on HPE’s Cray GX5000 supercomputing system but will feature Nvidia Vera Rubin CPUs, Rubin GPUs, and Quantum-X800 Infiniband networking. Like most of the other announcements this week, there’s no hard date for delivery of these systems other than “in the coming years”.

A final three Nvidia-based supercomputers – Tara, Minerva, and Janus – were also announced but are even lighter on detail; other than their names and that they will be located again at ANL, no further information has been given.

The US DoE is one of the primary consumers of supercomputing technology. The top three spots in the Top500 table of the most powerful supercomputers in the world are occupied by DoE systems:

• El Capitan at Lawrence Livermore National Laboratory
• Frontier at ORNL
• Aurora at ANL, where Solstice and Equinox will also be sited

All three existing DoE supercomputers are relatively new as well. El Capitan came online in 2024, Frontier in 2021, and Aurora in 2023.

MORE FROM ITPRO

• Inside Isambard-AI: The UK’s most powerful supercomputer
• Europe's first exascale supercomputer, Jupiter, is now live
• Inside Lumi, one of the world’s greenest supercomputers

Nvidia CEO Jensen Huang thinks future workforces will be made up of human and AI employees working in tandem, but this hybrid workforce could create challenges for enterprises.

Speaking during a recent interview at Citadel Securities’ 2025 Future of Global Markets conference, Huang pointed to recent advances in agentic AI – whereby autonomous agents conduct work on behalf of humans – as an indication that workforces will be fundamentally transformed.

This isn’t a case of firms building an array of autonomous workers, per se. Instead, these agents will essentially be akin to temps or agency staff that are brought in depending on their specific areas of “expertise”.

“I wouldn’t be surprised if you license some and you hire some, depending on the quality and depending on the deep expertise,” he said. “So future workforces in enterprise(s) will be a combination of humans and digital humans.”

Naturally, these agents will be provided by some of the leading companies operating in the space, such as OpenAI, and also spanning a range of areas.

The Nvidia chief specifically highlighted providers such as Replit, Cursor, and Lovable – all of which offer AI tools for use in software development.

The rise of AI workforces

Huang is by no means the first big tech exec to predict a shift toward AI-augmented workforces in recent months.

Indeed, since the advent of the generative AI ‘boom’, industry stakeholders have touted the potential for enterprises to cut down on human staff and replace them with AI.

While warnings over a complete disintegration of human workforces in place of AI were widespread in late 2022 and early 2023, the prevailing sentiment among some tech execs now points toward Huang’s prediction – somewhat of a hybrid setup.

Salesforce CEO Marc Benioff, for example, has been a highly vocal proponent of this idea. During an interview with Axios at the 2025 World Economic Forum, Benioff said current CEOs could be the last industry leaders to manage an all-human workforce as a result of AI agents.

“We are really moving into a world now of managing humans and agents together,” he told the publication.

Prepare to work alongside agents, but not for them

Predictions of a more collaborative human-machine workforce are gaining traction, with Microsoft’s Work Trend Index suggesting that workers will eventually be leading entire teams of agents.

Exactly how this relationship will work is still a point of contention, however. A recent survey from Workday found employees are warming to the use of AI agents at work – providing there are strict boundaries.

Three-quarters (75%) of respondents told Workday they’re comfortable working alongside agents, however, only 30% said they’d be comfortable being managed by one. The company noted that 12% were “not at all comfortable” at the prospect of being managed by an agent.

Huang pointed to the looming management-related quandaries facing enterprise leaders with this new workforce, suggesting the “hiring” process should still be viewed as the same as with humans.

“I tell my CIO, our company’s IT department [that] they’re going to be the HR department of agentic AI in the future,” he said. “They’re going to be the HR department of digital employees of the future”

“Those digital employees are going to work with our biological ones, and that’s going to be the shape of our company in the future.”

Nvidia appears to be putting its money where its mouth is on this front so far, with Huang revealing that in specific areas – such as cybersecurity – the company has more agents keeping watch than human staff.

"We home grow a lot of our AI's because we have a lot of proprietary knowledge and data that we want to protect, and we have skills in developing those AI's," he said.

In other domains such as software engineering and chip design, the company is also “augmenting” staff with AI, he added.

“At Nvidia today, 100% of our software engineers, 100% of our chip designers, every single engineer is augmented by Cursor,” Huang said.

MORE FROM ITPRO

• Jensen Huang has taken Nvidia to the top – but he says the 'suffering' wasn't really worth it
• Jensen Huang changes his tune on quantum computing after January stock shock
• Jensen Huang says AI will make workers ‘busier in the future’ – so what’s the point exactly?

OpenAI has signed a deal with AMD worth as much as $100 billion to the latter over several years – and could see the AI leader taking a 10% stake.

As part of a flurry of tech reveals and partnership announcements, OpenAI said it would use 6 Gigawatts of AMD GPUs via a multi-year, multi-generation deal that will start with a one-gigawatt deployment in the second half of next year of AMD's Instinct MI450 Series GPUs.

"By sharing technical expertise to optimize their product roadmaps, AMD and OpenAI are deepening their multi-generational hardware and software collaboration that began with the MI300X and continued with the MI350X series," OpenAI said in a statement announcing the deal. "This partnership creates a true win-win for both companies, enabling very large-scale AI deployments and advancing the entire ecosystem."

In order to "further align" the two companies' strategic interests, AMD has given OpenAI the ability to buy up to 160 million shares at a low fixed price, which will vest as the pair hit key milestones, including that first 1GW deployment but also share price targets. That could see OpenAI taking a 10% stake in AMD.

"This partnership is a major step in building the compute capacity needed to realize AI's full potential," said Sam Altman, co-founder and CEO of OpenAI.

The deal follows a similarly massive partnership worth $100bn with rival chipmaker Nvidia.

Big news for AMD

While OpenAI has used AMD chips before, the deal is a big one for a chipmaker that until now has "trailed" rival Nvidia in the AI boom, noted one analyst. "AMD has really trailed Nvidia for quite some time," Leah Bennett, chief investment strategist at Concurrent Asset Management, told Reuters. "So I think it helps validate their technology."

AMD's stock jumped by nearly 30%, while Nvidia's fell slightly on the news of the deal, though the rival chipmaker signed its own $100bn deal for 10GW of infrastructure last month.

AMD executives told Reuters that it expected to start seeing revenue from the deal next year as the 1GW facility is built, and that the deal will be worth tens of billions in annual revenue for the chipmaker – and by sparking deals with other customers, be worth more than $100bn over the next four years.

"Our partnership with OpenAI is expected to deliver tens of billions of dollars in revenue for AMD while accelerating OpenAI's AI infrastructure buildout," said Jean Hu, EVP, CFO and treasurer of AMD, in a statement. "This agreement creates significant strategic alignment and shareholder value for both AMD and OpenAI and is expected to be highly accretive to AMD's non-GAAP earnings-per-share."

OpenAI's $1tn in deals

The AMD deal takes OpenAI above the $1 trillion mark in deals focused on infrastructure and deployment alone, notes the Financial Times.

That includes the partnerships with chipmakers AMD and Nvidia, as well as rollout deals signed with Oracle and Coreweave, that will see OpenAI gain 20GW of compute capacity in the next decade, the report notes, adding that OpenAI itself estimates each GW of capacity costs it about $50bn – hence the $1trn bill. That includes the $500bn Stargate plans.

However, analysts noted that OpenAI doesn't have the money to pay those bills – though nor does the wider industry, with Bain & Co warning the cost of compute for the wider AI market will top $2trn by 2030, with revenue falling at least $800 short. "OpenAI is in no position to make any of these commitments," Gil Luria, analyst at DA Davidson, told the FT.

The various stock options may help it fund its investments, and echo the Nvidia deal that sees that chipmaker invest money into OpenAI to effectively enable it to buy its hardware.

That said, such structures and the high costs have raised concerns that the AI market is a bubble, with one investor saying: "there's a reasonable chance that a tremendous amount of capital destruction is going to come through this cycle."

MORE FROM ITPRO

• OpenAI signs series of AI data center deals with Samsung

Nvidia is investing as much as $100 billion into OpenAI, which will in turn use its hardware to build massive 10-gigawatt AI clusters.

The deal links up two of the giants of AI, with OpenAI sparking the recent boom and Nvidia powering much of it. The partnership, described as a letter of intent, will see OpenAI rolling out at least 10 gigawatts of Nvidia systems to train and run its next generation models.

Nvidia is expected to invest $100 billion in tranches as the systems are deployed — helping to pay for the work while also giving the GPU maker a further stake in its customer.

The first phase is set to come online in the second half of next year using Nvidia's Vera Rubin platform, OpenAI said.

Nvidia CEO Jensen Huang told CNBC that the 10-gigawatt systems were the equivalent to five million GPUS — as much as the company will ship this year.

“This is a giant project,” Huang said.

"Everything starts with compute," said Sam Altman, CEO of OpenAI. "Compute infrastructure will be the basis for the economy of the future, and we will utilize what we’re building with Nvidia to both create new AI breakthroughs and empower people and businesses with them at scale."

A win-win for OpenAI and Nvidia

Here's how the deal will work: as each gigawatt of Nvidia technology is deployed by OpenAI, Nvidia will invest $10 billion into OpenAI.

Analysts noted that the deal certainly makes it look as though Nvidia's investment is being used to buy Nvidia products.

"Nvidia invests $100 billion in OpenAI, which then OpenAI turns back and gives it back to Nvidia," Bryn Talkington, managing partner at Requisite Capital Management, told CNBC. "I feel like this is going to be very virtuous for Jensen."

"On the one hand this helps OpenAI deliver on what are some very aspirational goals for compute infrastructure, and helps Nvidia ensure that that stuff gets built," Bernstein analyst Stacy Rasgon told Reuters. "On the other hand the 'circular' concerns have been raised in the past, and this will fuel them further."

Beyond that, Nvidia has been named by OpenAI as a preferred strategic compute and networking partner for its "AI factory" plans, and the two companies will "co-optimize" their roadmaps.

This includes closer collaboration on development of OpenAI models, as well as deployment of Nvidia’s hardware and software. Full details of the partnership will be finalized in the next few weeks, the companies said.

Continued partnership

OpenAI and Nvidia have been working closely since the advent of the generative AI ‘boom’ in late 2022. The latter has emerged as the leading player in the hardware space, driving the systems that underpin much of the tech industry’s efforts on this front.

"We’ve been working closely with Nvidia since the early days of OpenAI," said Greg Brockman, cofounder and president of OpenAI, in a statement. "We’ve utilized their platform to create AI systems that hundreds of millions of people use every day."

Nvidia took part in a $6.6 billion funding round for OpenAI last year and, alongside OpenAI and Oracle, is part of the US infrastructure project known as Stargate.

"By locking in OpenAI as a strategic partner and co-optimizing hardware and software roadmaps, Nvidia is ensuring its GPUs remain the backbone of next-gen AI infrastructure," Hargreaves Lansdown analyst Matt Britzman told Reuters.

"The market is clearly big enough for multiple players, but this deal underscores that, when it comes to scale and ecosystem depth, Nvidia is still setting the pace — and raising the stakes for everyone else."

MORE FROM ITPRO

• UK to host largest European GPU cluster under £11 billion Nvidia investment plans
• OpenAI just revealed what people really use ChatGPT for
• Jensen Huang says AI will make workers ‘busier in the future’ – so what’s the point exactly?

Nvidia plans to invest $5 billion in Intel in a move analysts say will help the chip giant branch out into lucrative new markets.

Announced this week, the partnership will see the two rivals collaborate on data center and PC hardware. The immediate aim is to ensure Intel’s data center architecture – in particular its x86 ecosystem – can connect to Nvidia’s via its proprietary NVLink technology.

In time this will yield Intel-built, custom CPUs for Nvidia that will connect its market-leading GPUs within data centers.

The pair will also collaborate on new hardware in the consumer PC market, with Intel set to build x86 system-on-chips (SoCs) that will complement Nvidia’s RTX range of graphics cards, used for gaming and graphics-intensive applications such as computer assisted design (CAD) software.

“Intel’s x86 architecture has been foundational to modern computing for decades — and we are innovating across our portfolio to enable the workloads of the future,” said Lip-Bu Tan, CEO of Intel.

“Intel’s leading data center and client computing platforms, combined with our process technology, manufacturing and advanced packaging capabilities, will complement Nvidia's AI and accelerated computing leadership to enable new breakthroughs for the industry,” he added.

“We appreciate the confidence Jensen and the Nvidia team have placed in us with their investment and look forward to the work ahead as we innovate for customers and grow our business.”

Nvidia is the big winner

Intel has struggled in recent years to curb Nvidia’s meteoric rise to becoming not only the largest chip company, but also the most valuable company in the world.

The former chip titan’s shares fell 60% in 2024, the same year the firm slashed 15,000 roles and missed its own targets for AI hardware dominance.

Tan’s time as leader so far has been marked by a sharp strategic shift and stabilizing financials, though the firm still reported a net loss of $2.9 billion in its Q2 earnings statement.

In this light, the announcement with Nvidia could come as welcome news to Intel shareholders and help the company increase its US spending. Mario Morales, group vice president, Enabling Technologies and Semiconductors at IDC, said the partnership will be seen as a “positive move for both companies”.

Nvidia, however, will likely be the big winner given it now has an opportunity to capitalize on a broader array of markets.

“They are leveraging each other's strengths to deliver products to customers, particularly in the enterprise sector,” he said.

“This collaboration will help Nvidia expand into new markets beyond the data center, focusing on AI inferencing across various endpoints like PCs, smartphones, and autonomous systems that can leverage a collection of compute chiplets and dry the requirements of performance efficiency.”

The partnership comes just weeks after the US government took a 10% stake in Intel for $8.9 billion, amid financial woes, layoffs, and criticism of CEO Lip-Bu Tan by President Trump.

Under the terms of the deal, the purchase will be financed with a remaining grant of $5.7 billion from the US CHIPS Act and $3.2 billion from the Secure Enclave program – both legacies of the Biden administration.

The funding comes with the caveat of all spending needing to be carried out within the US, in alignment with each act’s stated aim of improving the US semiconductor industry and making it more competitive with China.

In the wake of tech tariffs and a greater focus by the Trump administration on domestic manufacturing, Nvidia has also been under pressure to reduce its reliance on firms such as Taiwan-based TSMC.

Its partnership with Intel will push it closer to US-centric operations, though the chip giant still heavily depends on TSMC, which produces 90% of the world’s advanced chips.

“The announcement today highlights the ongoing AI race between the US and China,” Morales added.

“While the US currently leads in technology, China is rapidly advancing with its own AI initiatives. The collaboration between Nvidia and Intel is seen as a strategic move to maintain US leadership in AI technology infrastructure and endpoints.”

MORE FROM ITPRO

• UK to host largest European GPU cluster under £11 billion Nvidia investment plans
• Jensen Huang says AI will make workers ‘busier in the future’
• Nvidia hails ‘another leap in the frontier of AI computing’ with Rubin GPU launch

Microsoft has announced plans to invest £22 billion ($30bn) across the UK in a bid to drive development of the country’s AI infrastructure.

The investment, which represents the largest financial commitment the tech giant has ever made in the UK, includes $15bn in funding to expand cloud and AI infrastructure.

In a blog post outlining the pledge, Microsoft president Brad Smith revealed this will also include investment to build the country’s largest supercomputer, boasting more than 23,000 Nvidia GPUs.

Development of the supercomputer will be led in collaboration with Nscale.

In a separate statement, Microsoft CEO Satya Nadella said the move builds on long-standing ties between the tech giant and the UK.

Nadella said the company has around 6,000 employees in the UK alongside multiple data center regions and "some our our most important AI and research labs".

"Microsoft has proudly invested and operated in the United Kingdom for more than four decades, opening our first office outside the United States here in 1982," he said.

"Our commitment to that partnership is stronger than ever. Just two years ago we announced a $3.2 billion investment in the United Kingdom’s AI infrastructure," Nadella added. "Today we’re doubling down on our investment."

Prime minister Keir Starmer welcomed the announcement, adding that the funding pledge is a “powerful vote of confidence” in the UK’s burgeoning technology sector.

“This commitment will not only strengthen our digital infrastructure and support thousands of highly skilled jobs, but also ensure Britain remains at the forefront of global innovation as we deliver on our Plan for Change,” he said.

A big week for British tech

The announcement from Microsoft comes amidst a big week for the British technology sector and coincides with US president Donald Trump’s second state visit to the UK.

A host of major technology companies have all made significant funding commitments in recent days.

Google, for example, unveiled the opening of a new $1 billion data center at Waltham Cross on the outskirts of London as part of a sweeping £5 billion investment scheme across the country.

Funding from the tech giant aims to build out cloud and AI infrastructure alongside investment in research and development (R&D).

Meanwhile, Nvidia announced plans to invest upwards of £11 billion on 16 September. Working alongside partners like Nscale and CoreWeave, the chipmaker said it plans to create Europe’s largest GPU cluster in the UK, underpinned by 120,000 Blackwell GPUs.

MORE FROM ITPRO

• How Satya Nadella's first decade at Microsoft took it from tech giant to undisputed industry titan
• Microsoft saved $500 million by using AI in its call centers last year
• Microsoft says AI is finally having a 'meaningful impact' on developer productivity

Nvidia has announced the UK will have Europe’s largest GPU cluster, totaling 120,000 Blackwell GPUs by the end of 2026, in a major boost for the country’s sovereign compute capacity.

The chip giant hailed £11 billion of total investment in the UK AI ecosystem, constituting chip orders and a rapid build-out of infrastructure such as AI factories.

These are expected to serve enterprise compute demand, researchers, and consumers.

The total is made up of investment from Nvidia partners Nscale and CoreWeave, with UK-based Nscale alone set to deploy 60,000 GB300 chips.

As part of the deal, Nscale will deploy a total of 300,000 Blackwell GPUs worldwide with the US, UK, Norway, and Portugal named as areas of investment.

“The UK has a true Goldilocks opportunity,” said David Hogan, VP for enterprise, EMEA at Nvidia.

“We have the right conditions for rapid AI growth and innovation in the country. The only thing that's been missing is infrastructure. Today we are announcing that Nvidia is building new AI infrastructure to support strong, secure and sustainable economic growth across the UK and serve as a platform with groundbreaking research.

Nvidia putting its money where its mouth is

In June, Nvidia CEO Jensen Huang praised the UK for its immense AI potential but noted it lacks the sovereign compute capabilities and infrastructure to capitalize on the generative AI ‘boom’.

Today’s announcement is a step toward addressing this gap by both Nvidia and its partners, securing the UK’s spot as Europe’s AI hub. With 120,000 GPUs deployed by the end of 2026, the UK would comfortably secure the title of largest GPU deployment in Europe, Hogan explained in a call with assembled press.

This still places the UK behind the US, with OpenAI targeting a deployment of two million GPUs for its Oracle-backed Stargate project, Meta seeking multiple gigawatt data centers representing millions of GPUs, and xAI in the process of increasing its Colossus training cluster to 200,000 Nvidia Hopper GPUs.

Elsewhere, the UAE is reportedly in negotiations with the US government to import 500,000 Nvidia GPUs per year for AI deployment.

Nevertheless, the UK is set to have an order of magnitude more chips at its disposal by this time next year.

Nscale’s pledged figure represents a significant increase on the 10,000 GPUs it announced it would deploy at London Tech Week in June, coinciding with Huang’s comments.

Hogan clarified that the £11 billion investment is inclusive of Nscale’s global deployment, but stressed that as the firm is headquartered in the UK, worldwide exports are nonetheless a significant boost for the region’s AI economy.

“We're delighted to announce Nscale's commitment to UK AI infrastructure today, including through Stargate UK and building the most powerful supercomputer in the country with Microsoft,” said Josh Payne, CEO at Nscale, in response to the announcement.

“As a UK-based company, we're showing how we can be makers, not takers, of the most important technology of our time.”

Of the 60,000 GPUs pledged by Nscale, 8,000 will be used by OpenAI for ‘Stargate UK’. This is the latest deployment of OpenAI infrastructure under the ‘Stargate’ brand, following its $500 billion US project and Stargate Norway expansion.

Though Nvidia didn’t provide an immediate breakdown of how the 240,000 chips Nscale is deploying outside the UK would be allocated, in July it announced plans to deliver 100,000 GPUs with Nscale at Stargate Norway.

This leaves 140,000 currently unaccounted for, with the as-yet-unannounced sites in the US and Portugal potentially filling the gap.

New AI ‘growth zones’ are coming

The move marks the latest big announcement on AI infrastructure for the UK over the last year. The government unveiled its AI Opportunity Action Plan in January, in which it laid out plans to establish ‘AI growth zones’ where the construction of data centers would be fast tracked and given priority access to the national grid.

Coinciding with Nvidia’s announcement, the Department for Science, Innovation & Technology (DSIT) announced a new AI growth zone in North East England, across Blyth and Cobalt Park outside Newcastle.

Blackstone announced £10 billion for a data center at the Blyth site in September 2024 and the DSIT stated the new zone holds potential for a further £20 billion in investment from other partners.

Stargate UK will be partially located at the Cobalt Park site, with the remainder of its infrastructure not yet allocated. Nvidia revealed CoreWeave and Nscale will be deploying chips across England and Scotland, raising the potential for further AI growth zone announcements.

In addition to its landmark announcement on GPU investment, Nvidia announced a new combined quantum hardware and AI supervision center in New York, launched by Digital Realty and Oxford Quantum Circuits.

The announcement coincides with President Trump’s state visit to the UK, during which a number of high-profile tech partnerships and announcements are being unveiled.

Executives such as Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman expected to play a role in the week’s proceedings.

Google announced a £5 billion investment into UK infrastructure, announced at the opening of a new data center in Broxbourne to expand its UK cloud availability. The investment total includes UK R&D and engineering, including AI-related costs at Google DeepMind.

MORE FROM ITPRO

• Gas-powered data centers: what's behind the boom?
• Salesforce UK chief exec Zahra Bahrololoumi touts the country's tech industry potential
• Google just confirmed the location of its first small modular reactor

Nvidia has unveiled its next generation GPU, the Rubin series, boasting marked performance improvements for AI inferencing.

Rubin is a new class of GPU, according to Nvidia, one that is purpose-built for massive-context processing, such as extensive software coding and high-speed video generation.

The new GPU will work “hand-in-hand” with the Vera CPU and be housed inside the latest Nvidia Vera Rubin NVL 144 CPX platform, the company said. This is an integrated NVIDIA MGX system that packs 8 exaflops of AI compute to provide 7.5x more AI performance than NVIDIA’s GB300 NVL72 systems.

It also promises 100TB of fast memory and 1.7 petabytes per second of memory bandwidth in a single rack.

However, the tech giant notes that Rubin CPX will be offered in other flexible configurations for customers who want to resume their existing infrastructure.

“The Vera Rubin platform will mark another leap in the frontier of AI computing — introducing both the next-generation Rubin GPU and a new category of processors called CPX,” said Jensen Huang, founder and CEO of Nvidia.

“Just as RTX revolutionized graphics and physical AI, Rubin CPX is the first CUDA GPU purpose-built for massive-context AI, where models reason across millions of tokens of knowledge at once.”

What you need to know about the Rubin CPX

Rubin CPX is set for general release in 2026 and could further solidify Nvidia’s dominance in the GPU market. But its main aim appears to be to solve the challenge of deploying inference at scale.

It uses a ‘cost-efficient’ monolithic die design that’s packed with NVFP4 computing resources, the company said, which has been optimized to deliver extremely high performance and energy efficiency for AI inference tasks.

A key point Nvidia made during its presentation was that Rubin would enable the highest performance and token revenue for long context processing; the company said it was capable of going “far beyond what today’s systems were designed to handle”.

With Rubin, AI coding assistants will become sophisticated systems capable of comprehending and optimizing large-scale software projects, Nvidia claims.

American generative AI company Runway, which will use Nvidia hardware in its video offerings called Rubin a “major leap in performance”, highlighting it as a powerful tool for creators.

“Video generation is rapidly advancing toward longer context and more flexible, agent-driven creative workflows,” said Cristóbal Valenzuela, CEO of Runway.

“We see Rubin CPX as a major leap in performance, supporting these demanding workloads to build more general, intelligent creative tools. This means creators — from independent artists to major studios — can gain unprecedented speed, realism, and control in their work.”

MORE FROM ITPRO

• Jensen Huang says 'the AI race is on' as Nvidia shrugs off market bubble concerns
• How Nvidia took the world by storm
• Jensen Huang thinks the UK has immense AI potential

INEurope's Jupiter supercomputer has been fired up, hitting the exascale threshold of more than one quintillion operations per second.

Sited at the Jülich Supercomputing Centre (JSC), in North Rhine-Westphalia, Germany, it's the first European supercomputer to do this. It's officially ranked as Europe's most powerful supercomputer and the fourth fastest worldwide, and is expected to be the first in the world to be able to handle 50 qubits, with the current record set at 48.

"With Europe's first exascale supercomputer, we are opening a new chapter for science, AI, and innovation," said Ekaterina Zaharieva, European commissioner for startups, research, and innovation.

"Jupiter strengthens Europe's digital sovereignty, accelerates discovery, and ensures that the most powerful and sustainable computing resources are available to our researchers, innovators, and industries."

Researchers, said the Commission, will now be able to run climate and weather models at kilometre‑scale resolution, enabling much more precise forecasts of extreme events such as heatwaves, heavy storms, and floods.

The University of Lisbon, meanwhile, will use Jupiter to scale multimodal, multilingual, open language models – integrating concepts from machine learning, sparse modeling, information theory, and cognitive science. The aim is to address the limitations of existing LLMs, allowing them to support all European languages.

Other planned projects include developing therapies to combat neurodegenerative diseases like Alzheimer's and retroviruses such as HIV, as well as advancing particle physics.

"Just as the planet Jupiter has a gravitational pull that shapes our solar system, the Jupiter supercomputer will pull Europe's research community, its start-ups, its industry, and its talent together," said Henna Virkkunen, executive vice-president for tech sovereignty, security, and democracy. "It will attract investment, stimulate breakthrough, pushing Europe forward."

"Europe gains its most advanced AI supercomputer"

Jupiter was developed at a cost of €500 million, with half coming from the EU and the rest from Germany.

It is powered by the Nvidia Grace Hopper platform, with 24,000 Nvidia Grace Hopper Superchips, based on Eviden's BullSequana XH3000 liquid-cooled architecture and interconnected with Nvidia Quantum-2 InfiniBand networking.

More than 50,000 network connections, said Nvidia, allow it to transmit three times more data than all global data traffic at any given moment.

Jupiter is claimed to be the most energy-efficient of the five fastest computer systems in the world, with more than 60 billion floating point operations per watt. And, thanks to a warm water-cooling system, it can also use the waste heat generated in operation to heat buildings, and will be integrated into the Jülich campus heating network.

"With Jupiter, Europe gains its most advanced AI supercomputer, built for large-scale simulation and AI, powered by NVIDIA Grace Hopper Superchips and Quantum-2 InfiniBand," said Jensen Huang, founder and CEO of NVIDIA.

"Jupiter fuses high-performance computing and AI into a single architecture. A platform for next-generation scientific computing, it will accelerate breakthroughs across every domain – from modeling climate and renewable energy to advancing quantum research, designing new materials and building digital twins."

MORE FROM ITPRO

• Inside Lumi

Jensen Huang has taken a bullish stance on the future of the generative AI market following Nvidia’s latest financial earnings report.

Speaking during an earnings call in the wake of Nvidia’s results, Huang shrugged off concerns over a looming AI ‘bubble’ dominating discussions in the industry.

Huang said he expects a further spending surge over the next five years, with continued enterprise demand pushing the AI space into a multi-trillion dollar market.

“We see $3 trillion to $4 trillion in AI infrastructure spend by the end of the decade,” he said. “I think the next several years, surely through the decade, we see really significant growth opportunities ahead.”

“The AI race is on,” Huang added.

Revenue at Nvidia increased to $46.74 billion, marking a 56% increase compared to the same period last year and exceeding Wall Street expectations. Similarly, profits increased by 40.8%, from $18.78 billion in the last quarter to $26.4 billion.

Notably, data center sales at Nvidia, which form the majority of its business, increased by 56%, with big tech customers - believed to be Meta and Microsoft - a key factor in growth.

Its data center sales, which form the majority of its business, rose 56% to $41.1 billion.

Alvin Nguyen, senior analyst at Forrester, said the results mark “another impressive performance” by the firm and highlight continued demand for products such as its Blackwell chips remains high.

“The company is making strategic moves across areas under their control, including continued development of state-of-the-art data center GPUs, innovation in new growth areas like robotics and physical AI, and promoting use cases that drive increased GPU purchases, such as reasoning AI and rack-scale computing."

AI Bubble? What AI bubble?

Huang’s bullish outlook comes against a backdrop of rising concerns over a potential AI ‘bubble’, sparked by comments from leading industry figures.

Earlier this month, OpenAI CEO Sam Altman told journalists that some people - particularly investors - may be getting “overexcited” by AI.

“Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes,” he said, according to reports from The Verge. “Is AI the most important thing to happen in a very long time? My opinion is also yes.”

Altman noted that huge valuations for emerging AI startups showed the current trajectory of investment in the space was “not rational” and “insane”.

It’s an issue that was highlighted earlier this year in a study from Silicon Valley Bank, which warned about the rise of ‘zombiecorn’ startups. These rising stars of the industry move fast, raise big investment rounds, but often fail to deliver tangible results and collapse rapidly.

"Someone is going to lose a phenomenal amount of money," Altman told reporters, according to The Verge. "We don't know who, and a lot of people are going to make a phenomenal amount of money."

Just days later, a study from MIT further compounded concerns about a looming dotcom bubble-style collapse. The study warned that 95% of AI pilots fail to deliver tangible financial savings or improvements to company profits.

Based on interviews with 100 executives and a survey of 300 staff, the study painted a gloomy picture for enterprises ramping up investment in the technology.

MORE FROM ITPRO

• Nvidia, Deutsche Telekom team up for "sovereign" industrial AI cloud
• Inside Isambard-AI: The UK’s most powerful supercomputer
• Jensen Huang thinks the UK has immense AI potential – but it still has a lot of work to do

HPE has announced an expansion to its Nvidia Computing by HPE portfolio, including the latest Nvidia hardware designed for AI.

Two new HPE ProLiant Compute servers will be made available, the company confirmed. The HPE ProLiant DL385 Gen11 server will support up to two Nvidia RTX PRO 6000 Blackwell chips within a slim 2U RTX Pro Server configuration.

Meanwhile, the HPE ProLiant Compute DL380a Gen12 server will support up to eight of the RTX PRO 6000, in a 4U form factor. This offering will be made available in September.

The RTX PRO 6000 was previously only available in the 4U form factor, but is now available in 2U form factor. This, HPE said, can be more easily deployed by customers in their existing server racks and is capable of being run with air cooling.

HPE highlighted the intended workloads for the HPE ProLiant Compute Gen12, including inference for generative and agentic AI models, visual computing, 3D modeling, and digital twins.

HPE Private Cloud will feature Nvidia's latest models

The newest version of Private Cloud AI will include support for this hardware, as well as for Nvidia’s latest AI models.

This will include Nemotron, a set of open source reasoning models based on Meta’s Llama, as well as the Cosmos Reason vision language model intended for robotics and physical AI devices and Nvidia Blueprint for Video Search and Summarization, used to help AI agents extract information from video data.

“Our collaboration with Nvidia continues to push the boundaries of innovation, delivering solutions that unlock the value of generative, agentic and physical AI while addressing the unique demands of enterprise workloads,” said Cheri Williams SVP and general manager for private cloud and flex solutions at HPE.

HPE has doubled down on its partnership with Nvidia in recent months, with commitments to ensuring its customers have access to the latest hardware from the chip giant.

Private Cloud AI was co-developed with Nvidia from the ground up, combining HPE’s private cloud offerings and GreenLake architecture with Nvidia’s AI computing stack including its NIM microservices.

Together, the firms have stated they aim to enable enterprises to make the most of enterprise AI deployment, with baked-in investment protection designed to provide organizations with the latest Nvidia GPUs as they become available.

Customers can also choose to air gap Private Cloud AI services, if their data privacy rules demand strict controls.

MORE FROM ITPRO

• HPE unveils Mod Pod AI ‘data center-in-a-box’ at Nvidia GTC
• HPE GreenLake gets the agentic AI treatment
• HPE just announced huge changes to its channel programs

Just outside Bristol, in a controlled site surrounded by razor wire and the hum of giant cooling machines, sits the UK’s new fastest supercomputer. Newly-inaugurated, as of 17 July 2025, this is a machine intended to supercharge UK innovation, with the raw power to crack some of the most complex problems in science.

Isambard-AI, named after the 19th century engineering pioneer, Isambard Kingdom Brunel, is more powerful than the rest of the UK’s supercomputers combined. With 21 exaflops of AI performance and 216 petaflops of linpack performance, it ranks 11th globally and is intended to push the UK’s sovereign compute capacity into competitive territory.

ITPro was among the first media outlets in the world to get an inside look at Isambard-AI at its official inauguration ceremony inside the National Composites Centre (NCC).

The first thing you notice about Isambard-AI when you step inside its hall is how quiet the machine actually is. The 12 cabinets that make up the machine are packed to the brim with ‘blades’, each containing eight Nvidia GH200 Grace Hopper Superchips.

In all, Isambard-AI contains 5,448 of these GPUs, with an HPE Cray ClusterStor E1000 providing 25 petabytes of data storage and HPE Slingshot 11 interconnect allowing data to be shared at 200Gbits/sec.

“Isambard-AI places Bristol at the centre of the AI revolution, spearheading AI innovation and scientific discovery in important areas such as drug discovery and climate research,” said Simon McIntosh-Smith, professor of High Performance Computing at the University of Bristol and director of the Bristol Centre for Supercomputing.

“I’m incredibly proud of our team and how we’ve worked so closely with partners to develop this national facility which will establish Bristol and the UK as an international hub for AI research.”

Record-breaking efficiency

Keeping Isambard-AI cool is no small feat: step back outside the data hall and you’re met with loud hybrid cooling towers, which use fans to cool down water that regulates the temperature inside via direct-to-chip cooling.

The site draws 5MW from its grid supply at the NCC, all of which is zero-carbon. This is partly what makes it one of the greenest supercomputers globally, ranked fourth on the Green500 list. But the cooling systems also feed directly into this, as they run 90% dry, reducing the overall water required for cooling.

This is something of a green coup in an industry that is increasingly battling immense data center water use, with Isambard-AI achieving a near-perfect power usage effectiveness (PUE) score of 1.1.

In the near future it could improve on this, with McIntosh-Smith revealing the NCC will soon be heated using waste heat from Isambard-AI, and that his team are working with South Gloucestershire Council to further distribute this heat to local businesses and homes.

Throughout the event, the team behind the project also stressed the importance of speed when it came to assembling Isambard-AI. If you delay too long, new hardware comes online that renders your expensive computing project less cutting-edge.

“We're thrilled to work very closely with our friends at Nvidia on some of the core accelerator technologies in these systems. But it's about how you put the whole system together,” said Neil MacDonald, EVP & GM, HPE Server Business.

“How you connect those technologies, how you power them, how you cool them, how you enable the team here at Bristol to deploy that technology so rapidly. Because if the technology is moving so quickly, the time to value, the time to science, the time to capability is critical.”

Isambard-AI has been years in the making, having been first announced in 2023.

The supercomputer was assembled offsite in six modules, with each cabinet built at the HPE factory in Czechia which were then moved into its data hall at the NCC in just 48 hours. From ideation to first computation, Isambard-AI took just two years to put together, far quicker than the four to five years a project of this scale would normally take.

MacDonald praised the “decisiveness” of the University of Bristol team and the UK government, adding that the HPE Cray EX supercomputer platform underpinning the system, as well as Slingshot 11, is the same technology underpinning the world’s three exascale supercomputers: El Capitan, Frontier, and Aurora.

He said this allows systems to be as scalable as possible, with potential for further build-out in the UK.

Building future supercomputers at this speed will require reliable access to chips such as Nvidia’s next-generation Blackwell GB200 chips, or future chips such as the upcoming Vera Rubin.

David Hogan, VP Enterprise EMEA at Nvidia, added that his team is in conversation with governments over access to chips and how to establish the right strategy for operationalizing compute infrastructure.

“This is about, ‘I need access to platforms’, how do you operationalize them? How do you then get access to the enablement of that, of those people, individuals, and organizations to take advantage of that, how they start to leverage software, how do you train people?”

A new scale of research and innovation

Allocation for Isambard-AI will be decided by the Department of Science, Innovation and Technology (DSIT). Researchers are awarded time on the machine for free and McIntosh-Smith stated that private businesses could bid for time at competitive rates, subject to DSIT approval.

ITPro was given an opportunity to see the work of select research teams who have secured time with Isambard-AI for their projects, including those working on the discovery of new materials for clean energy and for detecting prostate cancer using MRI scans.

The Rosalind Franklin Institute is using Isambard-AI to conduct analysis of placenta, both human and animal, to better understand how the structure of placenta impacts pregnancy health and why some diseases are found in specific animal groups but not others.

McIntosh-Smith added that Isambard-AI could train trillion-parameter LLMs in a matter of weeks and acknowledged that some models are already being trained on the machine.

These include Nightingale AI, a multimodal health model trained on NHS data by a team at Imperial College London, and BritLLM, a model developed by University College London intended to improve public services in all British languages including Welsh.

Isambard-AI was built to be as accessible to researchers as possible, without requiring extensive prior supercomputer experience.

“We actually designed the system so that people who are already maybe using public cloud can pretty much take their code, drop it straight onto Isambard-AI, and just run straight away,” explained McIntosh-Smith.

The UK’s AI ambition

Isambard-AI is the cornerstone of the AI Research Resource (AIRR), a growing cluster of supercomputers for UK innovators and researchers. The other supercomputer in the cluster is Dawn, a 19 petaflop supercomputer based at the University of Cambridge, which was built with Dell and Intel technologies.

The cluster could expand further in the near future, with the government having recently announced £750 million for a new supercomputer at Edinburgh University which is expected to be an exascale system.

This is part of a recent about-turn on supercomputing, with the government having initially scrapped £1.3 billion in funding earmarked for the Edinburgh project and expansion of the AIRR in the immediate aftermath of the general election.

Along with its AI Action Plan, the government has now made UK sovereign compute a major focus. Earlier this year, prime minister Keir Starmer announced a goal of increasing the UK’s compute capacity twentyfold by 2030.

McIntosh-Smith told ITPro that this figure was set with Isambard-AI already in mind as the ‘current capacity’, meaning the supercomputer is just the starting point. He further suggested that Isambard-AI could expand in the coming years, with 30MW of power capacity available at the NCC and the National Grid checking to see if more can be secured.

Hogan echoes comments by Nvidia CEO Jensen Huang, made at London Tech Week, that the UK’s immense AI skill needs to be backed with more sovereign compute.

“You should see Isambard as, almost, an incubator for AI in terms of platform. People should come here and build their business and AI prove that out and then take that and take it into the world and build on other infrastructure and start to scale.”

In its lifespan, which is contracted for five years but could run to six or seven, the University of Bristol hopes that Isambard-AI will live up to these expectations and directly enable researchers to make major breakthroughs.

“The sovereign AI issue is real,” McIntosh-Smith told ITPro.

“As the geopolitical situation becomes more complicated it’s more and more real. Isambard is big enough [that] we could train ChatGPT, the latest versions, from scratch in the UK. We could not do that before.

“So if somebody wanted to say ‘You cannot access the latest AI large language models’, we could have been cut off. Now if that happened, we could train one from scratch in the UK, we know exactly what was run, what data went into it, and do it all here – the data would never have to leave the UK.”

HPE has announced an expansion of its AI factory portfolio with a spate of new 'composable' solutions designed to supercharge the creation, adoption, and management of AI factories for businesses of all sizes.

The new features are integrated end-to-end solutions and services for AI factories that aim to remove the complexity of customers building a full AI tech stack on their own, which the company says has become a key hurdle to adoption rates over the last two years.

In an announcement at the company's annual HPE Discover conference, the company said that generative, agentic, and physical AI have the potential to transform global productivity and create lasting societal change – but it warned that AI is only as good as the infrastructure and data behind it.

"Organizations need the data, intelligence, and vision to capture the AI opportunity, and this makes getting the right IT foundation essential," said Antonio Neri, president and CEO at HPE.

"HPE is delivering the most comprehensive approach, joining industry-leading AI infrastructure and services to enable organizations to realize their ambitions and deliver sustainable business value."

HPE doubles down on Nvidia ties

HPE's new AI factory services leverage its partnership with Nvidia. The company's Private Cloud AI is part of the Nvidia AI Computing by HPE portfolio, for example, and boasts an Nvidia Enterprise AI Factory-validated design.

Within this the company launched new HPE ProLiant Compute Gen12 servers that rank number one in over 23 AI tests, HPE claims.

As well as the ability to provide a secure enclave to prevent tampering, this also supports post-quantum cryptography and trusted supply chain capabilities at rack and server level.

HPE revealed these servers will support the latest from Nvidia Blackwell computing.

Elsewhere, the offering also includes investment protection and scalability from one GPU generation to the next – this includes upgrading to Nvidia H200 NVL and Nvidia RTX Pro accelerators.

Organizations with strict data privacy requirements will also be entitled to air-gapped management and multi-tenancy that enables enterprises to collaborate and partition resources across their teams.

What's more, the expansion will also include Nvidia blueprints for fraud detection, digital human, and the Nvidia AI-Q blueprints for AI agent creation and workflows.

As well as a new "try and buy" program that lets customers test Private Cloud AI at Equinix's global data centers before making a purchase.

Nvidia is teaming up with Deutsche Telekom to build what it's calling "the world's first industrial AI cloud" – and continuing to push the idea of sovereign AI.

Nvidia is calling the collaboration "the world's first industrial AI cloud." It will be built by next year in Germany and operated by Deutsche Telekom, with the aim of supporting industrial applications including design, engineering, simulation, digital twins, and robotics.

The first phase will use 10,000 Nvidia Blackwell GPUs, as well as making use of Nvidia networking and AI software, the company said. Customers will be able to run Nvidia CUDA-X libraries, as well as Nvidia Omniverse workloads from Siemens, Cadance, and other providers.

"In the era of AI, every manufacturer needs two factories: one for making things, and one for creating the intelligence that powers them," said Jensen Huang, founder and CEO of Nvidia. "By building Europe's first industrial AI infrastructure, we're enabling the region's leading industrial companies to advance simulation-first, AI-driven manufacturing."

The "AI factory", as Nvidia is calling the data center, will be designed using its own Nvidia Ominverse set of digital twin tools. "As part of this blueprint, Cadence's Reality Digital Twin Platform will be used to simulate and optimize the entire AI factory in a physically accurate virtual environment, enabling the engineering teams to build a smarter, more reliable facility," the company said in a statement.

Industrial AI for physical AI

Nvidia said that Deutsche Telekom's AI cloud will help drive AI development and AI adoption in Germany, among startups, major enterprises, and also smaller businesses.

"We must seize the opportunities of artificial intelligence now, revolutionize our industry, and secure a leading position in the global technology competition," said Deutsche Telekom CEO Timotheus Hottges. "Our economic success depends on quick decisions and collaborative innovations."

One of the companies that will make use of the data center is Neura Robotics, a German maker of cognitive robots. Neura said it will use the Deutsche Telekom cloud for training AI to power its robots, which can be upgraded with skills for welding or ironing, for example.

"Physical AI is the electricity of the future – it will power every machine on the planet," said David Reger, founder and CEO of Neura Robotics. "Through this initiative, we're helping build the sovereign infrastructure Europe needs to lead in intelligent robotics and stay in control of its future."

Sovereign systems

The announcement comes as part of Huang's European tour, in which he's repeatedly pitched the idea of sovereign AI – the idea of a country or region owning its own AI infrastructure, for economic gains as well as to protect its own culture and independence.

"It codifies your culture, your society's intelligence, your common sense, your history – you own your own data," Huang told attendees of the World Governments Summit in Dubai last year.

Over the past week, according to Reuters, Huang has made appearances in London, Paris, and Berlin to announce AI infrastructure projects in those regions. In the UK, PM Keir Starmer announced £1bn in funding for AI infrastructure. In France, President Emmanuel Macron called AI infrastructure a "fight for sovereignty" following a partnership between Nvidia and Mistral to build a local data centre. And now the German plans for an "AI factory", which German Chancellor Friedrich Merz called an "important step" toward digital sovereignty.

That idea of sovereignty has also been backed by Microsoft via its Sovereign Cloud, which will keep customer data within European borders.

Nvidia CEO Jensen Huang has said the UK has enormous AI potential but needs significant infrastructure buildout to make the most of it.

Speaking alongside UK prime minister Keir Starmer at London Tech Week, Huang insisted that building AI supercomputers in the UK will be necessary to kickstart investment.

“The UK is in a Goldilocks circumstance,” Huang said, explaining that it has “one of the richest AI communities anywhere on the planet” and attracts the third largest amount of AI venture capital investment in the world, after the US and China.

Together, Huang argued, these factors make the UK a very attractive region for AI spending.

“It’s a fantastic place for VCs to invest – the ecosystem is really perfect for takeoff.”

But the Nvidia chief executive also warned the UK is the “largest AI ecosystem in the world without its own infrastructure,” comparing the challenge for researchers to astronomers trying to do their work without radio telescopes.

“You can’t do machine learning without a machine – and so the ability to build these AI supercomputers here in the UK will naturally attract more startups, it will naturally enable all the rich ecosystem of researchers here to do their life’s work,” he explained.

The UK government has committed to building out the UK’s compute capacity 20x by 2030, as announced in its AI Opportunities Action plan back in January.

During his speech at London Tech Week, Starmer announced an extra £1 billion ($1.35 billion) in funding to support this expansion and pledged to rapidly expand the construction of data centers and labs.

Huang argued that once the foundational infrastructure is in place, with Nvidia directly involved, research can feed into business which will further drive infrastructure buildout.

“I think it’s just such an incredible place to invest – I’m going to invest here,” Huang added.

UK firms are flocking to Nvidia hardware

Nvidia’s hardware is used around the world for AI data centers and supercomputer infrastructure. The firm is already heavily involved in UK projects such as the Bristol-based Isambard-AI supercomputer.

At the event, AI cloud platform provider Nscale announced plans to build UK AI infrastructure comprising 10,000 Nvidia Blackwell GPUs, while AI infrastructure firm Nebius committed to building a UK AI factory using 4,000 Nvidia Blackwell GPUs.

Starmer repeatedly stated that the government is committed to an ongoing conversation with the private sector to double down on the UK’s AI expertise.

“If there’s one thing I want to get across today, it’s that we are leaning into this,” Starmer said.

“We are excited about the potential that this could have, will have, on the lives of millions of people and making their lives better," he added.

"I came into politics quite late in my career to improve the lives of working people across the country and I can see here the huge potential to do that at a scale and speed which I didn’t necessarily think would be possible.”

The day before the event, the government hosted a meeting between cabinet ministers and select representatives from the tech industry including Huang, Google DeepMind CEO Demis Hassabis, and former Google CEO Eric Schmidt.

Attracting AI investment and expanding AI skills is a key aim of the UK government, and while onstage Starmer urged attendees to reach out to the government to get involved in public-private partnerships.

The UK government has separately committed to new funding for AI skills, with an aim to train 7.5 million people by 2030 in partnership with companies such as Nvidia, Google, and Microsoft.

MORE FROM ITPRO

• Jensen Huang doesn't think AI will come for his job — but other CEOs might disagree
• Starmer bets big on AI to unlock public sector savings
• The UK’s AI ambitions face one major hurdle – finding enough home grown talent

At the start of 2025, Nvidia CEO Jensen Huang caused the stock valuations of leading quantum computing companies to plummet by up to 40% (including Nvidia partners such as IonQ and Rigetti) when he claimed that a “very useful quantum computer” was at least two decades away.

“If you kind of said 15 years for very useful quantum computers, that'd probably be on the early side,” Huang said, during Nvidia's financial analyst day at the Consumer Electronics Show. “If you said 30, it’s probably on the late side. But if you picked 20, I think a whole bunch of us would believe it."

In the following weeks, there was a significant response from the quantum computing industry and quantum stocks tumbled. It was at Nvidia’s GTC event in March, during a panel where a number of affected quantum computing companies were in attendance, that Huang delivered a mea culpa of sorts and explained what was behind his comments.

“Building Nvidia and CUDA, and turning it into the computing platform it is today, has taken us over 20 years. So, the idea of time horizons of five, ten, 15, 20 years is really nothing to me. And of course quantum computing has the potential and the hope, all of our hopes, that it will deliver extraordinary impact,” Huang said.

“But the technology is insanely complicated, so the idea that it would take years to achieve was something that one: I would expect. Two: because of the complexity of it and the grand impact it would have, is something that we should expect.”

The panel provided a fascinating glimpse into the world of quantum computing. But it also posed a number of questions, notably around Nvidia’s own quantum strategy.

“We’re not trying to build qubits”

With Nvidia being world’s largest supplier of graphics processing units (GPUs), and one of the world’s most valuable companies, ranking second for global market cap at the time of publishing, Huang’s statements served as a stark reminder that the company isn’t planning to produce its own quantum processing unit (QPU).

“Nvidia is not trying to build quantum hardware,” explained Nicholas Harrigan, product marketing manager for quantum at Nvidia, in a promotional video. “We're not trying to build qubits, we're not trying to build a quantum processor. But we are trying to build a device that will turn those qubits into a useful quantum computer, and we call that device an accelerated quantum supercomputer.”

On the face of it, this may seem like an odd decision, especially when you consider the strategy of Nvidia’s peers. For example, Intel made its Tunnel Falls quantum processor available to developers and researchers in 2023. At the end of 2024, Google announced that it was creating its own quantum chip, Willow (using the same underlying tech as IBM). And, at the start of 2025, Microsoft announced its Majorana 1 quantum chip.

However, when you dig a little deeper, the decision makes more sense. Intel’s chip is created using silicon-spin qubits. Google (like IBM) uses superconducting qubits. Microsoft has opted for topological qubits. And there are other qubit technologies in the mix, too, such as trapped ions, neutral atoms, quantum annealing, and photonics.

"We can understand the strategy – staying chip-agnostic seems to be the strategically right call for now," says Arit Kumar Bishwas, director at PwC Innovation Hub.

"Nvidia is monetizing quantum today through software and hybrid systems while letting startups absorb the hardware risk. The strategy maximizes leverage from its GPU dominance, buys time for the physics to shake out, and preserves the option to dive into chips later, either via acquisition or joint ventures.

“They might be monitoring ecosystem wins and latency metrics; the day a hardware winner becomes obvious, maybe the expectation is to move swiftly (and capital-intensively) down-stack.”

Building on CUDA-Q

Rather than create its own quantum processing unit, Nvidia’s strategy is to take an ecumenical approach to quantum qubit technology. It is concentrating on the provision of AI factories, using software platforms that take advantage of Nvidia’s expertise in GPU-acceleration.

“The world is racing to build state-of-the-art large scale AI factories,” Huang said. “In a fab [semiconductor fabrication plant], there's the factory that manufactures the wafers, and then there's the factory that manufactures the information to manufacture the wafers. Every industry that has factories will have two factories in the future: the factory for what they build, and the factory for the AI.”

At the heart of Nvidia’s strategy is software called CUDA-Q, which is an open source quantum development platform that can run useful, large-scale quantum computing applications.

“We don't build quantum computers, but we are deeply integrated into the quantum computing industry, and we create libraries,” said Harrigan. “Software is just as important as hardware, and that's why we've built CUDA-Q, our software platform that allows users to seamlessly both access and control large scale accelerated quantum supercomputers.”

The Nvidia Accelerated Quantum Research Center

To help researchers and quantum hardware companies develop on its platform, Nvidia will be opening a new research center in Boston, which it says will be operational by the end of 2025. The NVIDIA Accelerated Quantum Research Center (NVAQC) will integrate leading quantum hardware with AI supercomputers, enabling what Nvidia refers to as ‘accelerated quantum supercomputing’.

“The Nvidia Accelerated Quantum Research Center will provide EQuS group researchers with unprecedented access to technologies and expertise needed to solve the challenges of useful quantum computing,” said William Oliver, director of the MIT Center for Quantum Engineering, in the announcement’s press release.

“We anticipate the future will also include other members of the Center for Quantum Engineering at MIT. Integrating the Nvidia accelerated computing platform with qubits will help tackle core challenges like quantum error correction, hybrid application development, and quantum device characterization.”

The Center will work with companies and research groups including Quantinuum, Quantum Machines, QuEra Computing, Harvard Quantum Initiative in Science and Engineering (HQI), and the Engineering Quantum Systems (EQuS) group at the Massachusetts Institute of Technology (MIT).

“One key part of quantum computing is developing software. And people always talk about a million qubits. But writing software for a million qubits is not easy,” said Avinash Palaniswamy, chief commercial officer at Quantinuum, in the same promotional video as Harrigan. “Where Nvidia is really useful is they also provide libraries, which run on GPUs. So in the software development flow, as long as we need to develop software for quantum computers, you will have to have the Nvidia technology there to develop those scaled qubits.”

The question of whether removing itself from the race for qubit development is a good decision could take decades to answer, especially if we follow Huang’s January timeline for the development of a “very useful” quantum computer. But, in doing so, Nvidia can continue to focus its hardware efforts on the present, which is dominated by AI infrastructure.

Cybersecurity provider CrowdStrike has announced its integration into the Nvidia Enterprise AI Factory validated design architecture.

As AI adoption continues to gather pace globally, the integration will enable enterprises to leverage CrowdStrike’s Falcon security capabilities in Nvidia Enterprise AI deployments to protect AI infrastructure, systems, and models.

In an announcement, CrowdStrike’s chief business officer Daniel Bernard said the company is “securing the standardized architecture that’s transforming how organizations innovate in the AI era.”

“Together with Nvidia, we’re eliminating the guesswork from securing AI infrastructure with a validated reference architecture,” he explained.

“Now, organizations can build Nvidia AI Factories even faster, adopting AI technologies with confidence and speed.”

Validated design

Nvidia’s Enterprise AI Factory validated design is deployable on premises with Nvidia Blackwell infrastructure and combines the hardware and software required to support every stage of the AI lifecycle, from data ingestion and model training, to deployment and runtime use.

Nvidia said its aim is to bring AI into production at scale and enable enterprises to leverage full stack infrastructure when building, deploying, and operating AI applications, agentic AI, as well as physical AI workflows.

However, as organizations increasingly look to AI and embrace AI factories, the technology also brings new business risks - such as data poisoning, model tampering, and sensitive data exposure.

Integration

CrowdStrike’s integration into Nvidia’s Enterprise AI Factory validated design aims to combat these risks by securing AI with AI.

The firm's own AI security capabilities are continuously trained on a feedback loop of front line insights from the firm’s threat hunters and intelligence analysts, combined with trillions of daily security events processed by the CrowdStrike Falcon platform.

This combination enables machine-speed detection and response to both existing and emerging security threats, according to the firm.

CrowdStrike combines its Falcon Cloud Security AI-SPM with advancements such as AI Model Scanning and Shadow AI detection to identify and mitigate risks in AI models, while providing visibility into overall security posture.

Additionally, this security is combined with CrowdStrikeAI Red Team Services and Falcon Adversary OverWatch for end-to-end security within enterprise AI factories using Nvidia AI Factory validated design.

“As organizations move quickly to operationalize AI, security must be built in from the ground up,” commented Justin Boitano, Nvidia’s vice president of enterprise AI software products.

“CrowdStrike’s integration with the Nvidia Enterprise AI Factory validated design brings real-time, AI-powered cyber security to the AI applications driving productivity breakthroughs across industries.”

MORE FROM CHANNELPRO

• CrowdStrike layoffs to hit 500 staff as firm targets strategic AI shift
• CrowdStrike eyes SaaS security gains with Adaptive Shield acquisition
• CrowdStrike prepares for battle as Delta given go-ahead for outage lawsuit

Reports suggest US President Donald Trump is considering banning DeepSeek amid an investigation into the AI developer.

The administration is considering banning US citizens from using DeepSeek, according to three sources familiar with the matter, the New York Times reported.

Chinese AI firm DeepSeek shocked the AI industry in January when it released a more efficient model that competed with more expensive, American-made options, sparking concerns that US companies could be overtaken in AI by Chinese efforts.

The New York Times also said that it may outright ban DeepSeek from buying US technology to power its systems, in particular Nvidia chips.

Nvidia chip bans

The rumors come as the US government investigates whether chipmaker Nvidia violated US rules established during the Biden administration that banned the sale of high-powered processors to China.

A congressional committee is investigating DeepSeek, accusing it of spying for China, stealing US designs, and subverting export controls. That committee is looking into allegations that Nvidia sold higher-powered chips to customers in Malaysia and Singapore, who then sold them on to China.

In February, Reuters reported that Singapore had arrested three people for selling Nvidia chips to DeepSeek in China.

The committee chairman, John Moolenaar, said in a statement yesterday: "We now know this tool exploited US AI models and reportedly used advanced Nvidia chips that should never have ended up in CCP hands.

“That’s why we’re sending a letter to Nvidia to demand answers. American innovation should never be the engine of our adversaries’ ambitions."

An Nvidia spokesperson told the publication that the company followed government rules on exports "to the letter."

That came alongside another hurdle for Nvidia. The chipmaker said in a filing on Tuesday night that it would now need a special license to sell to China its lower-powered H20 AI processors, which were specifically designed to sell in that country to meet Biden-era rules.

Nvidia said the tighter restrictions will cost it $5.5 billion in earnings.

Huang in China

Yesterday (17th April), Nvidia CEO Jensen Huang responded to the further tightening of export restrictions and the pressure on DeepSeek by visiting the AI developer in Beijing.

According to a report in the Financial Times, Huang and DeepSeek founder Liang Wenfeng discussed new chip designs to dodge the latest restrictions.

The trip plans were reportedly finalized after the H20 ban was revealed, with Chinese state broadcaster CCTV saying Huang called China a "very important market for Nvidia."

Indeed, it made $17 billion in sales there last year, the report added.

MORE FROM ITPRO

• How Nvidia took the world by storm
• DeepSeek R1 has taken the world by storm, but security experts claim it has 'critical safety flaws' that you need to know about
• Jensen Huang doesn't think AI will come for his job — but other CEOs might disagree

Nvidia said it will incur a $5.5bn cost due to the US government’s tightened export rules imposed on China.

The chip manufacturer is one of the key providers behind the boom in artificial intelligence technologies, but it will now require a license to export its H20 AI processors to China.

An escalating trade war between the US and China has followed the sweeping tariff changes announced by US President Donald Trump at the end of March. The two countries have since announced retaliatory trade tariffs with steep fees imposed on exported goods.

As a result, Nvidia said Tuesday that the US government previously told it the H20 chip required a permit to be sold in China – including Hong Kong. The tech giant also said federal officials had advised them that the license requirement will be in effect for “the indefinite future”.

“The [government] indicated that the license requirement addresses the risk that the covered products may be used in, or diverted to, a supercomputer in China,” Nvidia said.

Nvidia is one of a few businesses in the US that is worth trillions of dollars. This is partly due to its crucial role in the development of GPUs fueling the generative AI boom.

The company has seen its shares rise by 1,400% since 2020. However, shares took a tumble (around 6%) in the hours after trading in the US – potentially wiping billions off its value.

The tech giant’s value has already taken a hit this year, with a drop in January following the reveal of DeepSeek, the Chinese AI app that was built at a fraction of the cost of most other chatbots. At the time, it was suggested that the US had been caught off guard by the technological advancements used to develop DeepSeek.

On Tuesday, Nvidia separately announced plans to build up to $500bn worth of AI infrastructure in the US over the next four years. The aim is to bolster its presence in the American manufacturing market.

Previously, the company designed its chips but largely outsourced the production to contractors, such as the Taiwan Semiconductor Manufacturing Company (TSMC). Currently, Taiwan has been hit by a 32% tariff by the US government.

MORE FROM ITPRO

• Microsoft and Nvidia are teaming up to support UK AI startups
• How Nvidia took the world by storm
• Why is big tech racing to partner with Nvidia for AI?

Nvidia GTC 2025, the chipmaker’s annual conference, has dominated the airwaves this week – and it’s not hard to see why. The company is the belle of the ball in global tech right now, with everyone getting in on the action.

The generative AI boom that’s saturated the industry over the last two and a half years has, by and large, been underpinned by Nvidia’s hardware. Tech giants have flocked to Nvidia’s chips to power their relentless push to achieve AI dominance.

With growing concerns over infrastructure strain, the need for great levels of compute power, and the industry pivot to agentic AI, all eyes were on Nvidia to set the stage for a big year.

The annual conference certainly didn’t disappoint. CEO Jensen Huang, donning his iconic jacket, told attendees that the industry is at the precipice of significant change.

Data center infrastructure is changing, he noted, and Nvidia wants to lead the charge toward the development of ‘AI factories’. These modern data centers are more powerful, scalable, and intelligent than their predecessors - and this evolution is being fueled by the need to run AI systems more efficiently.

“AI has made a giant leap – reasoning and agentic AI demand orders of magnitude more computing performance,” Huang said.

So what were the big announcements at GTC 2025? Here’s everything you need to know.

Nvidia GTC 2025: Blackwell Ultra AI Factory Platform

Among one of the first major announcements at GTC 2025, Nvidia unveiled the launch of its Nvidia Blackwell Ultra AI Factory Platform.

The launch of the platform, Nvidia said, paves the way for the ‘age of AI reasoning’, and will help users boost training and test-time scaling inference capabilities. This essentially means enterprises can apply more compute during inference to improve accuracy.

As the name suggests, the platform is built on the Blackwell architecture, and includes the Nvidia GB300 NVL72 rack-scale solution, which the company said can deliver 1.5x improved AI performance compared to its predecessor.

This launch comes at a pivotal time in the evolution of the AI space, with enterprises now flocking to agentic AI, the company said. Accuracy is the name of the game amidst this shift, and by giving enterprises the ability to rapidly scale inference capabilities, Nvidia is betting this could be a game changer.

“We designed Blackwell Ultra for this moment,” Huang said. “It’s a single versatile platform that can easily and efficiently do pretraining, post-training and reasoning AI inference.”

Nvidia launches a family of open reasoning models

Agentic AI was once again given a major boost at GTC with the launch of Nvidia’s Llama Nemotron open reasoning models. This family of models is designed specifically to support developers to create advanced AI agents.

Built on Meta's Llama models, this offering will provide users with on-demand AI reasoning capabilities, and the company said it advanced the family during post-training to improve areas such as multi-step math, coding, reasoning, and complex decision making.

“This refinement process boosts accuracy of the models by up to 20% compared with the base model and optimizes inference speed by 5x compared with other leading open reasoning models,” Nvidia said in a statement.

These improvements mean the models can handle more complex reasoning tasks - making them the perfect foundation for enterprise AI agents - while reducing operational costs through improved efficiency.

Several industry big hitters are already working with the Llama Nemotron models, according to Nvidia. Microsoft, for example, has integrated the reasoning models within Microsoft Azure AI Foundry.

SAP is also tapping Llama Nemotron models to advance SAP Business AI solutions and Joule, the company’s flagship copilot tool.

Isaac GR00T N1

Advancements in robotics were a key talking point - and source of excitement - at GTC 2025, and Nvidia added more fuel to the fire with the launch of Isaac GR00T N1, the world’s first open source humanoid robot foundation model.

This is a family of fully customizable models, pre-trained by Nvidia, that will be released to robotics developers globally. The aim here is to accelerate the development of ‘generalist robots’ that can perform tasks akin to their human counterparts.

The potential use-cases are endless, and Huang went so far as to suggest that AI-powered robotics will be one of the big trends in the industry in years to come.

“The age of generalist robotics is here,” he said. “With Nvidia Isaac GR00T N1 and new data-generation and robot-learning frameworks, robotics developers everywhere will open the next frontier in the age of AI.”

GR00T N1 can “easily generalize across common tasks”, the company noted. This includes grasping, moving objects with one or both arms, and transferring items from one arm to another.

In his keynote, Huang showed off a humanoid robot developed by 1X, which performed domestic tasks such as tidying up.

Vera Rubin Ultra

Nvidia gave attendees a tantalizing glimpse into the future of its GPU roadmap this week with the Vera Rubin AI superchip. The successor to the company’s Grace Blackwell chip, Vera Rubin also uses a combination of a CPU and GPU.

While ‘Grace’ represents the CPU and ‘Blackwell’ the GPU, in this instance ‘Vera’ and ‘Rubin’ match accordingly.

Nvidia said Vera will boast 4.2 times the memory and 2.4 times the memory bandwidth compared to Grace, while Rubin will offer 288GB of high-bandwidth memory 4 (HBM4) compared to the HBM3e included in the Blackwell Ultra.

Exciting developments, no doubt, but Nvidia went one step further, unveiling plans for its Vera Rubin Ultra chip. These will provide even better performance stats upon launch, but we could be waiting a while - they’re expected to ship in the second half of 2027.

The big talking point here was the power consumption, with the company revealing a rack of Vera Rubin Ultra chips will consume up to 600kW each.

MORE FROM ITPRO

• Jensen Huang changes his tune on quantum computing after January stock shock
• How Nvidia took the world by storm
• HPE unveils Mod Pod AI ‘data center-in-a-box’ at Nvidia GTC

Nvidia CEO Jensen Huang has stepped back from his prediction that practical quantum computing applications are decades away following comments that sent stocks spiraling in January.

At Nvidia GTC’s "Quantum Day" event, Huang took a conciliatory tone to the quantum industry, comparing quantum to the GPUs made by Nvidia, noting the latter took plenty of time to make an impact.

"This is the first event in history where a company CEO invites all of the guests to explain why he was wrong," Huang said, according to CNBC, while opening an event that had a dozen representatives from across the quantum industry.

"This whole session is going to be like a therapy session for me,” Huang added.

Huang further backtracked on his comments in January when he said real-world quantum applications were much further off than many industry stakeholders believed.

At the time, the Nvidia chief executive suggested the technology would only be good at solving specific, theoretical problems.

"Of course, quantum computing has the potential and all of our hopes that it will deliver extraordinary impact," Huang reportedly said. "But the technology is insanely complicated."

Referencing the subsequent stock sell off, Huang admitted he didn't know so many quantum firms were already publicly traded.

"How could a quantum computer be public," he said.

Huang also asked the panel if ‘quantum computers’ was the wrong terminology, suggesting they should be referred to as scientific machines instead.

But D-Wave CEO Alan Baratz disputed that, saying his company's computers are already being applied for actual computing, beyond mere measurement.

"There are computational problems that are of the rage of classical computers," Bratz said, according to Investors Business Daily.

Quantum collaboration

Earlier this week, Nvidia announced it was building a research center to develop and provide technologies to help push forward quantum computing.

The Nvidia Accelerated Quantum Research Center (NVAQC) will bring together quantum hardware with existing AI supercomputing, what Nvidia is calling accelerated quantum supercomputing.

"The NVAQC will help solve quantum computing’s most challenging problems, ranging from qubit noise to transforming experimental quantum processors into practical devices,” the company said in a statement.

The NVAQC will be based in Boston in the US and expected to begin work next year. The center will work with industry and academia, offering access to Nvidia's GB200 NVL72 rack-scale systems, which the company claimed was the "most powerful hardware ever deployed for quantum computing applications".

"Quantum computing will augment AI supercomputers to tackle some of the world’s most important problems, from drug discovery to materials development," Huang said in a statement.

"Working with the wider quantum research community to advance CUDA-quantum hybrid computing, the NVIDIA Accelerated Quantum Research Center is where breakthroughs will be made to create large-scale, useful, accelerated quantum supercomputers."

Jensen Huang's quantum error

Such an announcement strikes a different tone to what Huang was saying earlier this year. In January, Huang tanked quantum stocks by saying he didn't buy predictions that the technology would have a real commercial impact any time soon, adding that real-world applications were likely decades away.

“So if you kind of said 15 years for very useful quantum computers, that’d probably be on the early side. If you said 30 is probably on the late side. But if you picked 20, I think a whole bunch of us would believe it,” he said at the time.

Quantum Computing Inc shares slipped by 45%, Rigetti Computing by 46% and IonQ by 42% — the total cost of Huang's comments was $4bn in market capitalisation.

That sparked pushback from industry giants including Google, which a few weeks later said it believed commercial quantum computing could arrive within the next five years.

Shortly after that, Microsoft unveiled a quantum chip breakthrough called Majorana 1, saying it could hasten progress towards practical applications.

MORE FROM ITPRO

• Nvidia thinks it’s time to start measuring data center efficiency by other metrics
• Jensen Huang has taken Nvidia to the top – but he says the 'suffering' wasn't really worth it
• Satya Nadella hails Microsoft’s 'Majorana' quantum chip breakthrough

Recent research carried out by ITPro showed investment in AI is a key priority for businesses globally, however, following through on that ambition is a question that looms large for many organizations.

At Nvidia’s GTC conference, Hewlett Packard Enterprise (HPE) unveiled Mod Pod, a product intended to answer at least some elements of this conundrum.

Mod Pod is a liquid-cooled modular data center that’s optimized for AI and HPC workloads. It’s built into a container and can, the company says, be easily deployed on a business’ premises without needing to completely overhaul its existing data center.

“A lot of data center space that does exist, does not have the capabilities for liquid cooling, which means you don't have the density in your racks, and you also don't have the PUE (power usage effectiveness),” said HPE CTO Fidelma Russo. “ So [Mod Pod] gives you a lower total cost of ownership.”

“We have examples of our customers, siting these in parking lots where they used to have employees, but with the work from home from COVID, they have the space,” she added. “So again it's easy, [you’ve] just got to level some space and you can have a data center in your backyard up and running in months.”

Mod Pod comes in 6m and 12m configurations, and supports up to 1.5MW per unit with a PUE of under 1.1. While HPE is keen to highlight its liquid-cooling credentials, the Adaptive Cascade Cooling technology can be adapted to use either air or liquid cooling depending on customer need and preference.

HPE expands Private Cloud capabilities

In addition to Mod Pod, HPE also announced several new features in Private Cloud AI, the flagship – and, thus far, only – product from its partnership with the chipmaker, Nvidia AI Computing by HPE.

The first is support for the newly announced Nvidia AI Data Platform, which allows Nvidia-Certified Storage providers, of which HPE is one, to build AI query agents into their hardware using Nvidia AI Enterprise software, such as NIM and Llama Nemotron models, as well as AI-Q Blueprint.

It also revealed a new developer system that adds an “instant AI development environment”, powered by Nvidia accelerated computing, HPE Data Fabric, and support for rapid deployment of Nvidia blueprints.

There were also a number of storage announcements, including HPE ProLiant Compute DL380a Gen12 and HPE ProLiant Compute DL384b Gen12, which feature Nvidia RTX Pro 6000 Blackwell chips and NVIDIA GB200 Grace Blackwell NVL4 Superchips respectively.

HPE ProLiant Compute XD servers, meanwhile, will support the NVIDIA HGX B300 platform, launched at GTC. The company says this technology will allow customers to “train, fine-tune and run large AI models for the most complex workloads, including agentic AI and test-time reasoning inference”.

MORE FROM ITPRO

• What HPE's results say about the direction of enterprise AI
• HPE’s AI and supercomputing journey continues with new Cray and Slingshot hardware
• HPE launches exclusive sovereign cloud offering for the channel

Nvidia CEO Jensen Huang might not be sold on the immediate potential of quantum computing, but industry giant Google appears to disagree.

The tech giant could be aiming to release commercial quantum computing applications within the next five years, according to reports from Reuters this week.

“We’re optimistic that within five years we’ll see real-world applications that are possible only on quantum computers,” Hartmut Neven, founder and lead of Google Quantum AI told the publication.

Neven’s optimistic outlook follows skeptical comments on the real-world applications of quantum computing by Nvidia co-founder and CEO Jensen Huang last month.

Interest in the potential of quantum technology has been building rapidly in recent years. In July 2024, analysis from Boston Consulting Group projected quantum computing will create anywhere between $450 billion and $850 billion in economic value by 2040.

But despite growing enterprise interest, the quantum computing space still remains in a nascent stage – a talking point which Huang was keen to emphasize, insisting that practical, real-world applications could still be decades away.

“So if you kind of said 15 years for very useful quantum computers, that’d probably be on the early side. If you said 30 is probably on the late side. But if you picked 20, I think a whole bunch of us would believe it,” he said, according to reporting from Axios.

Huang’s critical comments prompted shares in a host of popular quantum computing companies to nosedive. Shares in Quantum Computing Inc fell by 45%, for example, while IonQ experienced a 42% dip.

Notably, shares in Reigetti Computing plummeted 46%. The firm has reported significant growth over the last year, recording an 1,800% rise in stock prices.

Total losses on the back of Huang’s comments amounted to around $4 billion in market capitalization.

Quantum computing hype is building

Huang’s comments on the lack of real-world applications for quantum computing aren’t without merit. However, despite this enterprises still appear to be bullish on the potential of the technology - and this has been reflected in venture capital funding for organizations operating in the space.

Analysis from Crunchbase in November last year showed quantum computing startups had raised $1.5 billion in VC funding up until that point, with funding pledges spanning some 50 deals.

Notably, this marked a doubling of the $785 million raised in the year prior as well as the $963 million raised in 2022, which represented an all-time high.

Direct investment in quantum by enterprises is also building, research shows. A recent study from QuEra Computing found that global budgets for quantum applications are expected to rise by 20% in 2025 alone.

“The projected growth in both confidence and investment reflects a growing recognition of quantum computing’s potential to address complex challenges and drive innovation in key sectors,” the firm said.

Google has a track record with quantum gains

Google’s positive outlook on quantum computing appears to be firmly based around recent successes with the technology. In early December, the company announced it had made a major breakthrough with its new ‘Willow’ quantum chip, for example.

Using the chip, Google said it had successfully completed calculations that would previously have taken current-gen supercomputers 10 septillion years.

At the time, Google said the Willow success would pave the way to a “useful, large-scale quantum computer”, and CEO Sundar Pichai was highly vocal about the potential real-world applications on social media.

“We see Willow as an important step in our journey to build a useful quantum computer with practical applications in areas like drug discovery, fusion energy, battery design + more,” he said.

There are decades where nothing happens; and there are weeks where decades happen, so the saying goes, and that is certainly true of the final week of January 2025.

In the weeks leading up to recording this episode, we had planned to talk about Stargate – OpenAI, Oracle, and Softbank’s half-a-trillion pound AI investment. We will, but in a very different context because unless you’ve been under a rock, you’ll know that today’s episode has to be all about DeepSeek.

To say the Chinese large language model has caused an upset in the market would be an understatement. Within days of its release, it wiped more than a trillion dollars off the US stock market, and sent big tech into a public spiral.

But where did DeepSeek come from? And what does it mean for the future of the AI sector?

In this episode Jane and Rory are once again joined by Ross Kelly, ITPro's news and analysis editor, to explore the profound implications of DeepSeek.

Highlights

“I don't want to say loser here, because, you know, there's signs that there's rallying going on in the markets. But Nvidia dropped around 17 points, losing almost $600 billion in market value. Broadcom as well, equally, down around 17 points.”

“You need to remember what happened after the Sputnik situation, though it was again, a game of one upmanship between the USSR and and the United States and the West. Think we will potentially see that between, you know, us and Chinese based companies in years to come.”

“In making the claim that using relatively underpowered chips, specifically chips that China can can still import not even the frontline flagship chips that Nvidia is using today, in making the claim that using just those DeepSeek has been able to make a model that rivals some of OpenAI's best models, it scared a lot of investors because it throws into question that whole approach.”

Footnotes

• Chinese AI firm DeepSeek has Silicon Valley flustered
• The DeepSeek bombshell has been a wakeup call for US tech giants
• DeepSeek accused of training its models on OpenAI’s content
• Stargate project: OpenAI, Oracle pledge support for $500 billion AI infrastructure drive

Subscribe

• Subscribe to The ITPro Podcast on Apple Podcasts
• Subscribe to The ITPro Podcast on Spotify
• Subscribe to the ITPro newsletter
• Subscribe to the ITPro Podcast on YouTube
• Join us on LinkedIn

Nvidia shocked attendees at CES 2025 last week when it announced a new mini PC which it claims is the world’s smallest AI supercomputer.

Unveiled by CEO Jensen Huang in a Steve Jobs-esque ‘one more thing’ fashion, the device will feature a Grace Blackwell GB10 ‘superchip’ capable of delivering a petaflop of AI performance.

Project Digits, a placeholder name for the unit which Nvidia is actively seeking public input on to improve, will be equipped with the entire Nvidia software stack including NeMo, RAPIDS, Blueprints, NIM, and more.

The GB10 superchip will feature a GPU boasting the latest generation CUDA cores and fifth generation Tensor cores, as well as an ARM-based CPU with 20 power-efficient cores, which was designed in collaboration with Taiwanese semiconductor manufacturer MediaTek.

The system will come with up to 4TB of NVMe storage alongside 128GB of unified onboard memory, although the memory bandwidth capabilities were not revealed, prompting speculation about whether this may be a potential bottleneck for the system.

Designed specifically for AI developers, Nvidia said the device will be able to create prototypes, fine-tune, and run inferencing of large language models (LLMs) of up to 200 billion parameters locally.

It comes with a hefty price tag, however. Priced at $3,000, Nvidia is targeting enterprises and research institutions who are interested in developing and deploying advanced AI models without sinking huge amounts of capital in data center infrastructure or paying exorbitant cloud fees.

Project Digits will allow users to develop and run inference on these models using a single piece of hardware, which they can then choose to deploy to the cloud or a data center as they see fit.

The mini PC is set for an official launch in May 2025, and will be available from Nvidia and partners, the company confirmed.

Powerful AI-focused mini PCs to become a fixture in 2025

As Jensen Huang held the pint-sized device on stage at CES, he declared AI was the new mode of computing, and as such will become the norm across applications in every industry.

“AI will be mainstream in every application for every industry. With Project Digits, the Grace Blackwell Superchip comes to millions of developers,” he said.

“Placing an AI supercomputer on the desks of every data scientist, AI researcher and student empowers them to engage and shape the age of AI.”

With Project Digits, Nvidia looks to maintain its sizable lead in the AI compute space with a more approachable all-in-one system with a price that, although prohibitive to the average consumer, is relatively modest in the world of AI computing infrastructure.

The off-the-shelf system is not without competitors, however, with other major players unveiling their own mini workstations capable of powering generative AI fine-tuning and inferencing workloads.

For example, HP launched its Z2 Mini G1a workstation that features AMD’s Ryzen AI Max+ Pro 395, with 16 cores and 32 threats running at a base speed of 3 GHz and capable of being boosted up to 5.1 GHz.

HP says the workstation will be able to tackle complex AI-accelerated projects such as working on LLMs locally as well as other graphics-intensive and 3D design workloads.

As it is running on an AMD platform, the HP system will not be able to run Nvidia’s CUDA platform but will have access to AMD’s competing ROCm software.

Run:ai has confirmed its acquisition by Nvidia has been successful, adding that the chip giant plans to make its AI optimization tools open source so it can work across a wider variety of systems beyond Nvidia's own GPUs.

Founded in 2018, Run:ai develops software to help support AI infrastructure, building a platform on Kubernetes, the orchestration layer for much of the hardware that powers AI and the cloud.

Nvidia offered a reported $700 million for the startup in April, but the deal has faced regulatory scrutiny. Weeks after European regulators approved the deal, a blog post by Run:ai's founders has confirmed that the Nvidia acquisition is now complete.

"AI and accelerated computing are transforming the world at an unprecedented pace, and we believe this is just the beginning," said CTO Ronen Dar and CEO Omri Geller, the company's co-founders, in a blog post.

"GPUs and AI infrastructure will remain at the forefront of driving these transformative innovations and joining Nvidia provides us an extraordinary opportunity to carry forward a joint mission of helping humanity solve the world’s greatest challenges.

Run:ai says its software manages and optimizes the infrastructure that powers AI — whether in the cloud, on-premise, or a hybrid design — with the aim of boosting development, making best use of resources, and helping companies innovate in AI.

It allows AI developers to manage shared compute from a centralized interface, including pooling GPUs and applying them to different tasks, as well as managing users and their access.

"We will continue to help our customers to get the most out of their AI Infrastructure and offer the ecosystem maximum flexibility, efficiency and utilization for GPU systems, wherever they are: On-Prem, in the cloud through native solutions, or on Nvidia DGX Cloud, co-engineered with leading CSPs," they added.

Open source for open AI

At the time the deal was announced, Nvidia said it would continue to offer Run:ai's products under the same existing business model for the "immediate future".

At the moment, Run:ai's technology currently only works with Nvidia GPUs — but the acquisition is set to change that with Run:ai pledging to open source its software.

"While Run:ai currently supports only Nvidia GPUs, open sourcing the software will enable it to extend its availability to the entire AI ecosystem," the co-founders added.

"True to our open-platform philosophy, as part of Nvidia, we will keep empowering AI teams with the freedom to choose the tools, platforms, and frameworks that best suit their needs," they wrote.

"We will continue to strengthen our partnerships and work alongside the ecosystem to deliver a wide variety of AI solutions and platform choices."

Regulatory challenges

The deal was first announced in April, with Nvidia offering $700 million for the Israeli startup.

The European Commission examined the deal, citing antitrust concerns, but approved the acquisition in December on the grounds that Run:ai has "negligible" revenue so fails to meet necessary thresholds to take action.

The EC had been concerned that the deal would further cement Nvidia's dominance in the AI market. At present, Nvidia controls about 80% of the market for AI GPUs.

AMD has announced a 4% cut to its global workforce as part of a realignment of business priorities.

According to a statement sent to ITPro, AMD said it would be taking a number of steps ultimately resulting in a workforce reduction.

"As a part of aligning our resources with our largest growth opportunities, we are taking a number of targeted steps that will unfortunately result in reducing our global workforce by approximately 4%,” the statement said.

AMD did not state what those specific growth opportunities were, though it said it was committed to ensuring impacted employees are treated with respect and are assisted through the transition.

A previous Securities and Exchange Commission (SEC) filing stated that, as of December 2023, AMD employed a total of 26,000 staff members internationally. This means the layoffs could be equivalent to over 1,000 staff members.

The announcement comes after a promising charge in the AI space for AMD, with the firm’s Advancing AI event showing off strong footing in terms of the critical foundations needed for the technology.

At the firm’s Q2 2024 earnings call, AMD reported a 405% increase year-on-year in operating income in its data center unit. One analyst from IDC told ITPro that the firm is in a strong position to push hard on AI in 2025.

The firm’s Q3 earnings resulted in a 7% dip in shares according to reporting from Investopedia, though, with net income falling slightly below some estimates.

AMD isn’t the only chip giant making cuts

The move from AMD follows a raft of workforce cuts at industry rival, Intel, which announced plans to cut 15% of its workforce in August, equivalent to around 15,000 roles across its global workforce.

Speaking at the time, CEO Pat Gelsinger said the move was part of a cost structure realignment designed to change the firm’s operations, as it hadn’t been seeing enough revenue growth from technologies like AI.

Both AMD and Intel have been playing catch-up with industry rival Nvidia over the last two years. The chipmaker’s market value has surged due to soaring demand for its high-end chips, with big tech firms ramping up AI adoption and flocking to the firm.

Intel’s revenue had been falling before the August layoffs, and Gartner analyst Alan Priestly told ITPro the firm needs to focus on developing a clear product offering for AI in order to capitalize on the trend.

Microsoft and Nvidia have teamed up to launch a generative AI accelerator for UK startups, offering them technical guidance, advice on marketing strategy, and access to a network of potential investors.

The Microsoft GenAI Accelerator is specifically tailored for UK AI startups and scale-ups that have secured seed funding and are now looking to develop their products further.

The accelerator is focused on products that could change people’s lives for the better, create jobs, and have significant economic impact, the two tech giants said.

"By supporting local AI start-ups and providing them with the resources they need to succeed, Microsoft and our partners are helping to build a vibrant and sustainable AI community in the UK and create the tech powerhouses of tomorrow," said Microsoft UK CEO Darren Hardman.

"These fast-growing companies will create jobs, boost UK economic growth, and create products that make a positive impact on people’s lives."

The program will run from 22 January to 5 March 2025 and will culminate in a Demo Day, during which participants will have the chance to showcase their products to top venture capitalists and private equity firm.

There will also be a hybrid format for the program – part online, part in person – which Microsoft said should make it easier for companies outside London and the south-east to take part.

Participants will be given one-to-one support from Microsoft’s AI Black Belt team, which specializes in AI and machine learning, as well as technical guidance, advice on marketing strategy, and access to workshops and help from Nvidia and Microsoft-owned GitHub.

"Nvidia will support Microsoft GenAI Accelerator participants through the Nvidia Inception programme for cutting-edge start-ups," said Serge Palaric, Nvidia’s vice-president for hyperscaler alliance in EMEA.

"This collaboration will provide startups with access to Nvidia Deep Learning Institute (DLI) training sessions and a network of experts. By using NVIDIA’s accelerated computing platform, startups will be able to accelerate their AI development and enhance performance."

Startups and scale-ups wanting to take part have until 22 November to apply, with a shortlist to be drawn up in early December.

UK-based AI startups and scale-ups which already have a product in the market and are innovating across various sectors, including fintech, medtech, and greentech, can apply.

Microsoft noted it's particularly keen on companies that are looking to build AI-native solutions on Microsoft’s cloud.

Nvidia CEO Jensen Huang has expressed his confidence that AI systems will not be able to replace him, amid continued discussions over which jobs will be threatened by AI in the near future.

Speaking at Nvidia’s AI Summit in Mumbai in October, Huang said AI has the potential to take over 20% to 50% of a given job, depending on the specific role, according to reporting from the event.

But AI agents will be able to do these aspects of the job 1,000 times better, Huang predicted, arguing that instead it would be the “person who uses AI to automate that 20% [who] is going to take your job.”

Huang dismissed growing fears entire jobs could be lost to AI agents, arguing that “in no job can [AI] do all of it”, stating that the scenario he sees playing out is one where workers implement AI tools as assistants to automate their workflow.

When asked if he harbored any latent fears that he would be replaced by AI, Huang said “absolutely not”, emphasizing he saw no possibility of AI CEOs replacing him or fellow chief executives.

This stands against the results of an edX survey conducted last year, which spoke to 500 CEOs from across the US to gauge their thoughts on how AI would impact the business landscape.

EdX found that just under half (49%) of the CEOs surveyed said they believe most or all of their jobs could be completely automated or replaced by AI, compared to just 20% of IT workers.

AI CEOs already in position, despite Huang’s prediction

Despite Huang’s confidence, AI CEOs have already emerged with a number of organizations announcing they will be using an AI-powered chief executive.

NetDragon Websoft, a Chinese video game developer, appointed an AI-powered virtual humanoid robot as its CEO in August 2022, months before OpenAI debuted its ground-shifting ChatGPT product.

“The appointment is a move to pioneer the use of AI to transform corporate management and leapfrog operational efficiency to a new level,” NetDragon said in a statement released after announcing the appointment.

The move was met with a sizable jump in the company’s share price, with its stock rising by 10% in the six months following the announcement.

NetDragon Websoft has not been the only firm to foray into the world of automated chief executives. Polish drinks company Dictador Holdings appointed Mika, touted as the world’s first AI robot CEO, in November 2023.

Unveiled in June 2022, Mika was created by Anthropic in partnership with Hanson Robotics, with her role described as identifying opportunities for the business, and overseeing day to day operations for the rum and gin brand.

“Dictador's board decision is revolutionary and bold at the same time. This first human-like robot, with AI, in a company structure, will change the world as we know it, forever,” Marek Szoldrowski, president at Dictador Europe said at the time.

In an ‘interview’ with Reuters, Mika explained that its “decision-making process relies on extensive data analysis, and aligning with the company’s strategic objectives.

Szoldrowski told Reuters the appointment was far from a PR stunt, and Mika was “definitely in fact CEO”, but cautioned human chief executives should not need to worry about being edged out by AI replacements, stating human leaders will still be required for major decisions.

"There is no concern [that] artificial intelligence could hire or fire somebody. It's still the major decisions, significant decisions are still in (the) human executive team's hands.”

Cognizant has announced a new partnership with Nvidia that aims to enhance the IT consultancy firm’s data modernization offering and enable faster AI adoption.

The collaboration will see Cognizant leverage the chipmaker’s full stack accelerated computing platform, including Nvidia RAPIDS, part of the Nvidia AI Enterprise software platform, alongside the Cognizant Data and Intelligence Toolkit.

The move aims to tackle the escalating costs and performance bottlenecks caused by the rapid growth of data and enterprise need for real-time analytics, while laying the groundwork for faster adoption of AI use cases, Cognizant said in an announcement.

“Our latest collaboration with Nvidia aims to transform the way businesses process and analyze large datasets, delivering, we believe, unprecedented speed and efficiency,” commented Annadurai Elango, EVP and global head of core technologies and insights at Cognizant.

“Transitioning clients to GPU-based infrastructure will also lay the groundwork for implementing exciting new generative AI use cases in critical business functions.”

The new and enhanced data modernization offering will leverage Nvidia’s RAPIDS Accelerator for Apache Spark to help customers speed up data processing tasks, ramp up resource efficiency, and leverage more sophisticated analytics.

The pair said the combination will outperform traditional CPU-based processing without the need for code changes in Apache Spark-based workloads.

Cognizant’s Data and Intelligence Toolkit, which features a set of accelerators designed to eliminate friction when leveraging data for decision making, will harness Nvidia GPU acceleration through the RAPIDS software libraries and APIs.

Customers stand to benefit from enhancements across analytics processing performance, data preparation, model training, as well as deployment. Proprietary benchmarks have shown an increase of 2.5x in acceleration in data engineering and data science workloads, alongside a 66% reduction in computing costs when compared with traditional CPU processing.

"Enterprises migrating to the cloud need efficient solutions for large-scale data processing and real-time analytics to speed development and retrieval-augmented generation for agentic AI,” said John Fanelli, Nvidia’s vice president of enterprise software.

“With Nvidia RAPIDS accelerated data processing, Cognizant can help businesses significantly reduce computing costs and accelerate workloads to extract more value from their cloud investments."

Intel has won a long-running legal battle against the EU after successfully overturning an antitrust fine imposed in 2009.

The European Commission had found the company abused its dominant position in the chip market and fined it €1.06 billion, marking a record at the time.

According to the European Commission, the company had been giving rebates, some disguised, to major computer manufacturers, including Dell, Hewlett-Packard, and Lenovo, on condition that they bought Intel’s x86 central processing units (CPUs). These were termed conditional rebates.

The company was also found to be paying manufacturers to stop or delay the launch of products based on rival chips such as those from Advanced Micro Devices (AMD). This was known as ‘naked restrictions’.

"Such rebates and payments effectively prevented customers, and ultimately consumers, from choosing alternative products," the Commission said at the time.

"By undermining competitors’ ability to compete on the merits of their products, Intel’s actions undermined competition and innovation."

Since then, the case has gone back and forth at a glacial pace. The fine was overturned in 2022 on the basis that the Commission's analysis hadn't been thorough enough to prove that Intel's behavior had harmed competition.

The Commission then cut the fine to €376.36 million on the basis that it had only proved the naked restrictions were damaging competition, and not the conditional rebates.

In the wake of this, the Commission appealed, asking to reinstate the full amount of the original fine; but with this decision from the EU Court of Justice the case is finally over.

"The Court of Justice upholds the annulment by the General Court of the Commission’s decision finding an abuse of a dominant position on the part of Intel and imposing a fine of €1.06 billion on Intel," the court said in a statement.

"In its judgment, the Court of Justice rejects all of the grounds of appeal raised by the Commission."

While Intel may have dominated the chip market back in 2009, that's very far from the case now. As recently as 2020, analysis from S&P Market Intelligence showed it was the world's second-largest semiconductor corporation by market capitalization, behind only Taiwan Semiconductor Manufacturing Company (TSMC). Now, though, it's number 14.

What's changed is not just the rise of TSMC, but also the huge demand for AI, which has seen Nvidia's sales boom. The chipmaker recently recorded revenues of $30 billion for the three months to 28 July.

This week, it was reported by Korean media outlet MK that Intel Foundry was in talks with Samsung over a collaboration aimed at toppling TSMC.

The reports suggest the two could partner on research and development, share production facilities and exchange process technologies, the report said.

Nvidia CEO Jensen Huang has confirmed that a design flaw that was impacting the expected yields from its Blackwell AI GPUs has been addressed.

Huang clarified that the issue was the result of Nvidia’s design, and not its fabricating partner TSMC, with mass production of the Blackwell B100 and B200 GPUs expected to resume later this month, according to Reuters.

In a conference call in August, Huang revealed Nvidia had modified the design of the Blackwell chips to improve yield, with rumors soon swirling that the incident had caused tension between the company and TMSC.

Speaking at Nvidia’s AI Summit in Mumbai, Huang shed some light on the issue, clarifying that it was the sole result of an Nvidia error, and TSMC was in no way responsible for any production delays.

"We had a design flaw in Blackwell ... It was functional, but the design flaw caused the yield to be low. It was 100% Nvidia's fault,” Huang stated.

Huang pushed back on suggestions that two companies' relationship had become strained, adding that TSMC was instrumental in assisting Nvidia recover from the interruption and get units back in production as soon as possible.

"In order to make a Blackwell computer work, seven different types of chips were designed from scratch and had to be ramped into production at the same time,"he explained

"What TSMC did, was to help us recover from that yield difficulty and resume the manufacturing of Blackwell at an incredible pace."

Huang labels claims of bad blood with TSMC ‘fake news’

After initial reports of the design flaw emerged, some outlets claimed the issue and resulting delays to production may have soured its longstanding relationship with TSMC.

Some outlets suggested the fault was associated with the packaging technology onboard the Blackwell chips, namely TSMC’s CoWoS platform.

According to some reports, the two companies were in deadlock over who was responsible for the fault, blaming each other for the performance issues and potentially jeopardizing the pair’s 30-year relationship.

TSMC has been central in Nvidia’s rise to become the preeminent chipmaker of the AI era, with Huang himself remarking that “without TSMC, there would be no Nvidia today”.

Huang labeled any claims of ill will between the two firms as ‘fake news’.

Speaking at a recent Goldman Sachs conference, the Nvidia CEO announced the Blackwell GPUs will enter mass production in the fourth quarter (late October), and are expected to start shipping in early 2025.

However, Nvidia disclosed that it would have to ship some of the low-yield units in 2024 to satisfy surging demand for the Blackwell GPUs, specialized for AI workloads.