Amazon Web Services (AWS) is planning to invest up to $50 billion in purpose-built AI and high-performance computing (HPC) infrastructure for the US government.

The data center build, set to start next year, will see an extra 1.3 gigawatts of compute capacity added across AWS Top Secret, AWS Secret, and AWS GovCloud (US) regions across all classification levels: Unclassified, Secret and Top Secret.

Federal agencies will gain expanded access to AWS's AI services, including Amazon SageMaker AI for model training and customization, Amazon Bedrock for model and agent deployment, Amazon Nova, Anthropic Claude, and leading open weight foundation models, with the use of AWS Trainium AI chips as well as Nvidia AI infrastructure.

"Our investment in purpose-built government AI and cloud infrastructure will fundamentally transform how federal agencies leverage supercomputing,” said AWS CEO Matt Garman.

“We're giving agencies expanded access to advanced AI capabilities that will enable them to accelerate critical missions from cybersecurity to drug discovery. This investment removes the technology barriers that have held government back and further positions America to lead in the AI era."

The investment, the company said, will speed up discovery and decision-making across applications ranging from national security to scientific research and innovation, including autonomous systems development, cybersecurity, energy innovation, and healthcare research.

AWS said the move is aimed at supporting the priorities laid out in the US government's recently-published AI Action Plan, which promotes the development of AI infrastructure, as well as other advanced computing initiatives deployed on secure, US-based AI and cloud infrastructure.

AWS boasts deep government ties

The hyperscaler is already a major cloud provider to the US government, serving more than 11,000 government agencies including US intelligence communities, the Department of Defense, federal agencies and defense industrial companies.

In 2011, it became the first cloud provider to build infrastructure specifically for government security and compliance requirements with the launch of AWS GovCloud (US-West).

Since then, it's launched AWS Top Secret-East, the first air-gapped commercial cloud accredited to support classified workloads; AWS Secret Region, making it the first cloud provider accredited across all US government data classifications; and AWS GovCloud (US-East), AWS Top Secret-West, and AWS Secret-West regions.

"Federal customers and the supporting industrial base share a vision of AI and HPC convergence. This includes orchestrating expert AI models, agents, and natural language interfaces to enable researchers and engineers to explore complex problems through conversational interaction," said Amazon.

"This represents a fundamental shift from traditional HPC workflows to AI-accelerated discovery, where scientists can specify challenges and receive AI-driven recommendations backed by high-fidelity simulations and analysis."

MORE FROM ITPRO

• AWS to give AI skills to 100,000 people in the UK by 2030
• Is AWS' cloud dominance waning?
• Channel focus: All you need to know about AWS' partner program

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

Good morning from Barcelona on the east coast of Spain where HPE CEO Antonio Neri will be delivering his HPE Discover Europe keynote from the Fira exhibition center.

Neri's keynote will kick off at 1.30pm CET, but ahead of that you can read our predictions of what we can expect to hear about this afternoon.

It's a very breezy day in Barcelona with a few branches coming off trees immediately in front of the Fira north entrance. Inside it's bustling with people picking up coffee and finding their way around the conference center as the main conference day starts to get underway.

There's just over an hour until Antonio Neri's keynote starts. In the meantime, why not check out one of the company's other announcements from this week – El Capitan. The supercomputer, which was built by HPE using AMD chips, has come online and immediately been crowned the world's fastest by Top500.

The wait is almost over. Antonio Neri will shortly be on the stage and the keynote theatre is filling up with delegates. But first, we have some FC Barcelona fans (or performers playing Barcelona fans) coming through the crowd and on stage.

FC Barcelona is one of the headline customers at this year's European Discover – the opening VT is talking up the company's pedigree in data analysis in sport (this also includes other sports like motor racing).

Neri is now on stage and talking up again the relationship with FC Barcelona – HPE is now the club's official technology partner.

The Miriam Ferrando, CIO of FC Barcelona, has joined Neri on stage and is explaining the partnership. "Our vision is to fully immerse fans during their entire visit and deliver fast, stable and secure, customized experience," she says.

The football segment (or soccer, depending on where you are) is over and we're now onto HPE itself. Neri compares HPE to FC Barcelona: "An organization with a storied history, engineering roots and deep technical expertise and with the inspiration to transform our capabilities."

The company is not looking back, however, but instead is focused on "where the world is going".

In line with the company's own branding and strategy of late, as well as the way the technology hype winds are blowing, this is of course AI.

Neri points out that while it took the telephone 75 years to reach 50 million users and mobile phones 12 years to do the same, for generative AI it took less than a year.

"This is not just rapid growth, it's a major paradigm shift, profound transformation unlike anything we have ever seen," he says. "AI is not just enhancing technology, it is enabling new worlds of interaction capabilities, redefining what is possible in our lifetimes. It's transforming every sector, every line of business, and creating opportunities we couldn't even imagine 18 months ago."

He touches on another key HPE talking point – one we may hear an increasing amount about over the coming years – which is "responsible" AI. This includes elements like privacy (more on that shortly...), as well as sustainable technology development, and "human focused".

Now we're onto HPE's involvement in high performance computing (HPC) and supercomputing. El Capitan gets a mention, but Neri also points out that the company has built the top three fastest supercomputers in the world, according to Top500.

"In football terms, I call that a hat trick," he says.

Time for some announcements. First up, HPE VM Essentials.

HPE announced it was developing its own virtual machines at Las Vegas, and this is an evolution of that. It incorporates technology from the company's most recent acquisition, Morpheus.

Neri makes a not-so-oblique reference to the controversy around VMware licensing changes since it was acquired by Broadcomm.

" I speak to customers, a lot – more than 50% of my time I spend with customers and partner –, and what they tell me is that the virtualization landscape is shifting due to recent changes in various contracts and licensing terms. Many customers tell me their costs have increased three to five times," he says.

"For that reason, many customers are re evaluating their options and looking to create the right mix of run times across the hybrid cloud operating model."

In response, the company is releasing HPE VM Essintials

"HPE VM Essentials provides a unified VM management experience, which means you can manage existing VMware war notes or the new HP VM essentials hypervisor with a simplified experience across both stacks," says Neri.

Feedback from the beta test has been very positive, says Neri, and it will be generally available as a standalone solution from next month.

Some new AI announcements as well. The company has taken the opportunity of being in Europe, where privacy and data security regulations tend to be stricter, to announce some new air-gapped versions of its products.

First up, HPE Private Cloud Enterprise Disconnected.

"Many of you have asked us for the same HPE Green Lake cloud native experience, but not connected to the Internet or outside the cloud," says Neri. "In response to that, today, we are introducing the option for disconnected management for HPE private cloud solutions for a fully air gaps management option for you, for the building or our disconnected management capability."

The second air-gapped product is HPE Alletra Storage MP Disconnected. This is an on-premises, fully disconnected block storage option.

"This new solution provides customers in highly regulated environments the ability to manage with stringent privacy and security requirements for the most efficient mission critical databases," said Neri.

There's a new member of the Alletra storage family as well, the X10,000 all-flash object storage device that has been optimized, Neri says, for high speed data lakes and rapid store for backup and recovery.

"It will offer up to six times faster object storage performance compared to our competitors," Neri claims

Neri is joined by two guests, HPE CTO Fidelma Russo and Neil MacDonald, EVP and GM of compute, HPC and AI.

This is a wide ranging conversation, starting off with the fact that AI isn't new – it's what MacDonald studied at university some years ago. However, newer forms of AI – even preceding generative AI – are quite different from the AI of old.

"In traditional terms, artificial intelligence didn't rely on massive amounts of data and computation. It was about hand crafted rules and expert knowledge being encoded," he says. "Since then, we saw a more modern subset of AI emerge with machine learning, and then later deep learning, which allowed computers to make predictions without being explicitly programmed with that knowledge."

MacDonald says that despite these differences, ML and deep learning are both about learning patterns, which is what classical AI has always been about.

"Generative AI is a key shift, because ... it's no longer about labeling things, but it's generating things," MacDonald says. "It's radically different."

Generative AI, he says, has "enormous potential to fundamentally transform human productivity".

While the three are very much talking up the technology, they do address one of the key problems with increased AI adoption – energy consumption.

This is something that was touched on in June at Discover Las Vegas, and which we've discussed in the ITPro Podcast as well.

Neri is now joined on stage by Rami Rahim, CEO of Juniper Networks.

HPE announced its intention to acquire Juniper Networks earlier this year and, for now, it's still a pending acquisition. In that respect there's not an awful lot of news to report, but the conversation did illuminate a bit more how Juniper's technologies will be used.

According to Rahim: "It's about more than just combining (our) strengths to create the best networking business on the planet.

"I think it's about transforming what networks can do for those that use them, from delivering better experienced with traditional applications to meeting the growing demands of AI workloads."

Rahim and Neri talk up Juniper Networks' heritage in AI, which goes back almost 10 years, Rahim says.

In short, Juniper's technology will build on what HPE has in place already with Aruba and help bolster the company's AI strategy, the CEOs say.

And with that the keynote is over, as is this live blog.

For more from HPE Discover Barcelona 2024 as it happens, and our archive content from June, catch up on all our HPE Discover coverage, click here.

Hewlett Packard Enterprise (HPE) has powered up the latest in its AI-capable supercomputers, Aurora.

Aurora is the second exascale system delivered by the company. Having reached 1.012 exaflops on 87% of the system, is the second fastest in the world after Frontier – also created by HPE – according to Top500. 

Built in collaboration with Intel for the United States Department of Energy’s Argonne National Laboratory, the system uses HPE Cray EX infrastructure and features 75,000 compute node endpoints, 2,400 storage and service network endpoints, and 5,600 switches. The 10,624 compute blades that make up Aurora also feature 21,248 Intel Xeon CPU Max Series processors and 63,744 Intel Data Center GPU Max units.

According to HPE, researchers using Aurora will be able to run generative AI models to “accelerate scientific discovery”. Early experiments run on the system include brain mapping, developing methods for detecting neutrino interactions, and fusion reactor simulations.

Rick Stevens, associate laboratory director and distinguished fellow at Argonne National Laboratory, described Aurora as a "first of its kind supercomputer" that will be a gamechanger for researchers. 

"Reaching this milestone with a second exascale system in the US is an incredibly significant achievement that will advance open science initiatives globally." he said.

For HPE, AI and HPC are inseparable 

High-performance computing (HPC) has been an important part of HPE’s portfolio since its acquisition of SGI almost 10 years ago and Cray in 2019. It was only last year, however, that the company really started to put it front and center in its public announcements.

This shift has been spurred at least in part by the rise of generative AI and the company’s move to embrace it. At HPE Discover 2023, held in June last year, CEO Antonio Neri revealed HPE’s new focus, telling attendees that it had moved from being the "edge to cloud" company to "edge to cloud to AI".

This direction of travel was underscored at the time by the release of HPE GreenLake for Large Language Models, an HPC-powered infrastructure offering that allows customers to train and deploy large-scale AI projects without having to invest in their own specialist hardware at extremely high cost.

Subsequently, it has built several AI-focused HPC offerings, including Isambard-AI, the UK’s most powerful supercomputer, Helios, Poland’s fastest system, Venado in New Mexico, and now Aurora.

The creation of a “reliable and practical” quantum supercomputer could be reached within the next decade, according to researchers at Microsoft. 

In an announcement outlining its quantum computing roadmap this week, the tech giant said it has achieved the “first milestone” towards developing next-generation computing capabilities. 

Chetan Nayak, technical fellow and corporate VP of quantum hardware at Microsoft, said the firm's ambitions rest on recent advances in the development of ‘topological qubits’. 

Microsoft announced last year it had reached a breakthrough with the development of these qubits, which offer far greater reliability and stability compared to traditional techniques. 

Harnessing topological qubits, Nayak said the company will be able to develop a quantum supercomputer capable of performing 1 million reliable quantum operations per second (rQOPS), which marks a step change compared to current capabilities. 

A peer-reviewed paper published by Microsoft researchers outlines these capabilities, noting that the breakthrough has significant long-term potential. 

Quantum development challenges

The announcement follows years of development in the quantum computing space for Microsoft, which Nayak noted has proved challenging in terms of ‘scaling up’ capabilities.

“A quantum supercomputer must be powered by reliable logical qubits, each of which is formed from many physical qubits. The more stable the physical qubit is, the easier it is to scale up because you need fewer of them,” he said. 

“Over the years, Microsoft researchers have fabricated a variety of qubits used in many of today’s NISQ computers, including spin, transmon, and gatemon qubits. However, we concluded that none of these qubits is perfectly suited to scale up.

“That’s why we set out to engineer a brand-new qubit with inherent stability at the hardware level. It has been an arduous development path in the near term because it required that we make a physics breakthrough that has eluded researchers for decades.”

Microsoft’s quantum supercomputer roadmap

Having established a breakthrough, Microsoft said the next phase of its roadmap will focus on several key areas, starting with the development of hardware-protected qubits. 

Hardware-protected qubits will have built-in error protection for computation processes, the company said, and will be smaller, faster, and more controllable for reliability. 

“This unique qubit will scale to support a reliable qubit, and will enable engineering of a quantum supercomputer,” Nayak wrote. “Each qubit operation will take less than one microsecond. This means problems can be solved in weeks, rather than decades or centuries.”

“Our qubits will be controlled by digital voltage pulses to ensure that a machine with millions of them doesn’t have an excessive error rate or require unattainable input/output bandwidth.”

High-quality hardware-protected qubits, which are the next phase of development on the roadmap, will then help reduce computational error rates. 

The development of a ‘multi-qubit system’ is also on the horizon, the company added. This will create a series of quantum algorithms that can be “executed when multiple qubits operate together as a programmable Quantum Processing Unit (QPU) in a full stack quantum machine”. 

Longer-term, the goal for Microsoft is to create what it describes as a “resilient quantum system” before finally reaching a fully-operational quantum supercomputer. 

This quantum system, the firm said, will be capable of solving “impactful problems even the most powerful classical supercomputers cannot”. 

HPE has launched a new artificial intelligence (AI) offering for its GreenLake as a service platform at its annual conference, HPE Discover.

Dubbed HPE GreenLake for Large Language Models (LLMs), the high-performance computing (HPC)-powered offering allows customers to privately train and deploy large-scale AI projects, which has previously been a challenge. 

Access to supercomputers outside of very specific industries is limited and often expensive, and unless you’re an AI company, using it for AI training hasn’t typically been a priority.

Justin Hotard, executive vice president and general manager for HPC and AI at HPE, told a pre-announcement press conference: “Supercomputers today are costly and complex to adopt and manage. Supercomputers also require unique data center capabilities for power and cooling. And until now, supercomputers have not been available on demand in a consumption model.”

HPE GreenLake for LLMs hopes to solve this, the company said. It will give organizations the opportunity to train their own LLMs using HPE’s Cray XD supercomputers and Nvidia H100 GPUs on demand, without needing to invest in additional infrastructure or skills.

HPE GreenLake for LLMS: Leaning on German engineering

Underpinning the GreenLake for LLMs proposition is technology developed by German startup Aleph Alpha. 

The Heidelberg-based company’s flagship Luminous software is a pre-trained LLM optimized for analyzing and processing data at scale.

Explaining the decision to work with Aleph Alpha, Hotard said Luminous was trained using HPE supercomputers and existing HPE software.

“[Luminous] has already been implemented by various organizations in healthcare and financial services, and in the legal profession as a digital assistant,” he added. 

“By enabling on-demand access to Luminous, enterprises can also take advantage of one of its multiple languages, which include English, French, German, Italian, and Spanish.”

HPE has also taken into account data privacy and data sovereignty concerns while developing the product. 

GreenLake for LLMs only uses the customer’s data for their language model training, eliminating fears about exposing trade secrets, for example.

The company foresees a relatively broad potential customer base. Hotard told ITPro that it will be primarily useful to enterprises, including larger public sector bodies, but that “there are certain startups that can take advantage of this to accelerate their offerings and services” – notably those in the AI space.

A true private cloud offering

Unusually for an HPE GreenLake product, this is not a hybrid or on-premises managed product, but is hosted entirely in a remote private cloud housed in a facility in Canada. 

This means that, for now at least, customers won’t be bringing any of the associated hardware into their data centers.

The company is also hopping aboard the eco-friendly data center train for this project, with the company planning for this initial instance to be 100% carbon neutral.

Hotard said: “The objective [for] all of our cloud deployments … is to provide 100% carbon neutral offering to our customers. The first one is using nearly 100% renewable power. And one of the benefits of liquid cooling is you can actually use the … wastewater, the heat comes out of the heat of water and reuse it.

“We have examples of that in other supercomputing installations and we're leveraging those learnings and expertise for this cloud deployment as well.”

Anyone getting excited to try HPE GreenLake for LLMs for themselves may wish to temper their expectations, though. 

While the firm is accepting orders now, you will effectively be joining a waiting list. Additional availability is expected by the end of the year, but will be limited to North America. 

Availability in Europe is expected to follow early next year, with a question mark over rollout in the rest of the world. Pricing is also currently unavailable.

Exascale computing represents the next frontier in high-performance computing (HPC). It delivers unprecedented processing power capable of executing up to one quintillion (10¹⁸) calculations per second, marking a significant leap forward and enabling breakthroughs in fields such as artificial intelligenc (AI), climate and weather modelling, drug discovery, and nuclear fusion research.

For years, nations have raced to develop sovereign exascale supercomputers, recognizing their strategic importance in scientific progress, economic competitiveness, and national security. The EU, UK, US, China, and Japan have all made significant advances, with multiple exascale machines now operational.

As the world faces various challenges such as climate change, harnessing the power of supercomputers is going to be an advantage that nations want, and are likely to be protective over once they have it.

Despite its immense power, exascale computing is not just about speed: it is about solving previously intractable problems. With the ability to process vast datasets and run complex simulations, exascale systems are expected to accelerate scientific discoveries, from modelling the universe’s origins to simulating new materials at the atomic level.

What exactly is exascale computing?

Exascale computing refers to a new generation of supercomputers capable of performing at least one exaFLOP, equivalent to one quintillion (10¹⁸) calculations per second. This represents a thousandfold increase over the previous HPC benchmark, petascale computing, which is still very powerful.

At its core, exascale computing is about unprecedented processing power, enabling highly complex simulations, real-time data analysis, and AI model training at a scale never seen before. These machines rely on thousands of interconnected processors and accelerators, such as GPUs and custom-built AI chips, to distribute workloads efficiently, and require vast memory bandwidth and energy-efficient architectures to operate.

The transition to exascale computing has been driven by growing global demand for faster and more detailed simulations in areas such as climate modelling, materials science, and medical research.

Traditional supercomputers, even at petascale levels, struggle with the increasing complexity of modern datasets and AI models. Exascale systems bridge this gap, allowing scientists and engineers to simulate entire planetary systems, predict weather patterns with greater accuracy, or model atomic interactions for next-generation medicines.

Beyond its scientific applications, exascale computing is also a matter of national strategy and security. Countries investing in exascale tech are using it to advance AI, develop next-generation encryption, and enhance defence capabilities.

Which countries have exascale capabilities?

The US remains the leader in exascale computing, with three operational systems. Frontier, launched in 2022, was the world’s first exascale supercomputer, delivering 1.1 exaFLOPS. In 2024, the Aurora system exceeded 1.012 exaFLOPS, followed by El Capitan in late 2024, which became the world’s fastest supercomputer, hitting 1.742 exaFLOPS.

These machines drive AI research, climate modelling, and national security initiatives.

Europe has also made strides, with Jupiter, the EU’s first exascale supercomputer, having gone live in September 2025. France is also building its own exascale supercomputer, with AMD and Atos subsidiary Eviden starting assembly on the Alice Recoque supercomputer in 2026 with an aim to solve complex scientific and industrial problems.

Meanwhile, China and Japan continue to advance their HPC capabilities, though China’s exascale progress remains largely undisclosed.

The UK, however, has struggled to keep pace. Despite announcing a £1.3 billion exascale project in 2023, the government withdrew funding in 2024, shifting priorities to other AI initiatives. As of 2025, the project was revived with £750 million in government backing – though with “exascale” notably missing from the announcement.

Industry applications of exascale computing

Exascale computing is revolutionizing scientific research, enabling simulations and analyses at an unprecedented scale and accuracy, while in climate science, researchers can now model entire planetary systems, predicting extreme weather events and long-term climate changes with greater precision.

Similarly, in astrophysics, exascale supercomputers are used to simulate the formation of galaxies and black holes, helping scientists better understand the universe’s origins.

The healthcare sector is another key beneficiary. Exascale computing accelerates drug discovery and genomic research, allowing scientists to simulate the behavior of proteins and molecular interactions in record time, which has major implications for cancer treatment, precision medicine, and the development of new vaccines.

AI-driven medical diagnostics also stand to benefit, as exascale systems can process vast amounts of patient data, leading to earlier disease detection and more personalized treatment plans.

In the energy sector, exascale computing is driving advances in nuclear fusion, renewable energy optimization, and oil and gas exploration. By running highly detailed simulations, researchers can improve the efficiency of wind farms, optimize energy grids, and develop next-generation nuclear reactors. Companies in the oil and gas industry also use exascale models to better predict underground reserves and reduce environmental impact.

How powerful is exascale computing?

Exascale computing represents a monumental leap in classical supercomputing, providing the raw processing power to perform one quintillion (10¹⁸) calculations per second. To put this into perspective, an exascale supercomputer can complete in one second what would take a human over 31 billion years to calculate manually.

Compared to petascale systems, the previous high-performance computing benchmark, exascale machines are around 1,000 times faster, making them capable of handling unprecedented levels of data analysis, simulation, and AI model training.

However, while exascale computing pushes classical computing to its limits, quantum computing takes an entirely different approach.

Instead of using traditional binary processing (0s and 1s), quantum computers leverage qubits, which can exist in multiple states simultaneously, allowing them to solve certain complex problems exponentially faster. While exascale supercomputers excel in areas like physics simulations, AI, and cryptography, quantum computing is expected to outperform classical systems in highly specialized tasks, such as molecular modelling, financial risk analysis, and encryption breaking.

Despite its potential, quantum computing is still in its infancy, with most systems limited by error rates and hardware constraints. Exascale computing, on the other hand, is already a proven technology, actively driving advancements in AI, medicine, climate science, and engineering.

In the foreseeable future, rather than replacing exascale computing, quantum and classical supercomputers are likely to work in tandem, combining their strengths to tackle some of the world’s most complex computational challenges.

Where next?

Exascale computing represents a major leap in high-performance computing, driving advances in climate modelling, medical research, and AI. With the ability to process one quintillion calculations per second, exascale systems are revolutionising scientific discovery.

As competition in supercomputing intensifies, exascale is shaping the future of AI, advanced simulations, and scientific breakthroughs. The energy demands and software challenges of these systems remain key hurdles, but continued innovation in hardware efficiency and scalable algorithms is pushing the industry forward.

Most modern hardware is launched with a certain level of hyperbole. Every new release – phones, laptops and even processors – it’s the smartest, the fastest, the greenest or simply the best yet. Lenovo, however, has just launched a workstation packed with so much computational power, it might actually be underselling it.

The ThinkStation PX, launched alongside the P5 and P7, is described as a ‘desktop data centre’. Its chassis is co-designed with UK luxury car manufacturer Aston Martin, and beyond some contractual obligation to market it with that partnership in mind, Lenovo has integrated a three-channel cooling system to allow for unobstructed airflow. Even though I’m a huge nerd for Bond films, I’m telling you to put the P series’ striking DBS-inspired 3D grill chassis to one side and dive straight into the inners of this workstation, where the real business interest lies.

Inside the Lenovo ThinkStation PX, P7 and P5

As great as the design might be, take a look at the power; oh so much power. The ThinkStation – on first impression – is more powerful than it probably needs to be. It’s rare to my mind that a computer is released with two of the highest-end processors from Nvidia and Intel.

Usually one is newer or stronger than the other, but the ThinkStation PX has the latest Intel 4th Gen Xeon Scalable processor with 120 CPU cores, no less, and support for up to four dual-slot Nvidia RTX 6000 Ada Generation GPUs – these have the memory necessary to work with massive datasets and workloads for things like data science and simulation. This comes alongside 2TB DDR5 RAM and room for seven 4TB NVMe and two 12TB SATA storage units. Imagine all of this on your desk.

Even the smaller models – the ThinkStation P7 and ThinkStation P5 – house processing chops that will render almost every other desktop manufacturer green with envy. The major differences between the three models are the physical size, and amount of power expressed through the components.

Contrasting with the higher end PX, the entry-level P5 comes with a Xeon W-2400 with up to 24 cores, alongside room for two Nvidia RTX A6000 GPUs. Storage can comprise up to three 4TB NVMe and up to three 12TB SATA, paired with 512GB DDR5 RAM. The mid-range P7, meanwhile, uses the Xeon W-3400 CPU with up to 56 cores, as dwell as room for three Nvidia RTX 6000 Ada GPUs. This is alongside up to 1TB DDR5 RAM and four 4TB NVMe and three 12TB SATA units for storage.

Just a week ago I was lamenting the lack of business-focused innovation from Lenovo. The firm rocked up to MWC 2023 with a laptop proof of concept where the screen could be made taller (rollable display meets rollable eyes). I didn’t see the business use case there, but I certainly see it here with the ThinkStation PX; it’s pure, uncut computational power.

Data centre power at your desk

What calls for such a monstrously powerful workstation? Our jobs, unfortunately. They’ve gotten too complex, too challenging, and very demanding. Indeed, some of us need artificial intelligence (AI) just to get by; think data scientists and engineers using massive datasets the humble human brain cannot even begin to fathom.

At MWC 2023, Huawei’s storage expert Dr Peter Zhou noted the amount of data businesses generate annually by 2030 will be measured in yottabytes. This explosion of data has happened across industries and fundamentally changed how we work, according to Lenovo. The company is confident in this assessment because it asked customers for feedback and they told them exactly what they wanted – more power.

I’m trying not to delve into the metaverse, Web3 or any virtual reality (VR) stuff here, but these are also reasons we need more juice, and the ThinkStation PX does seem to have a clear market identity for those types of use cases.

During its presentation, Lenovo demonstrated the ThinkStation PX’s digital twin – a virtual replication of the hardware the company used to tweak the design and map out its capabilities. It was hard not to be impressed as the firm spun the design around, detailed its clever cooling system and slid out its virtual storage compartments.

The whole unit is designed for data centre use. It has the flexibility to work on a desk at home, or in the office – or even in a rack at a data centre – with a seamless transition between each scenario, according to its maker. The unit slots into a standard rack – it comes with sliding rail kits – and has front access storage. This means it won’t look out of place in a data centre rack, but it might also catch a few glances in the office with its fancy red trim.

My god, I want one. I don’t even have that much computational need, I’m merely a hard, but this seems as close to the whole ‘personal supercomputer’ marketing ploy as we've ever seen. I just love the sound of it; data centre power on the desk of your average worker. There aren’t any pricing details yet, so it’s presumably going to cost you (or your employer) an arm and a leg, but surely that’s a worthwhile investment to future-proof against the mountains of data we’ll all soon be working with.