Enterprises can choose two powerful AI accelerators based on their unique AI development needs
Powered by NVIDIA superchips, the Dell Pro Max with GB10 and GB300 help accelerate AI innovation
Dell’s AI accelerators are a key component within the broader Dell AI Factory portfolio
AI workstations are rolling out worldwide in 2026, giving enterprises high-performance computing capabilities on each individual worker’s desk, offering desk-based AI development alongside traditional cloud-based development options.
These new AI-powered devices mark a foundational change in enterprise AI innovation, enabling them to supercharge experimentation and adoption of AI.
Among the providers rolling out AI accelerators is Dell Technologies. These systems form a key component of Dell Deskside Agentic AI, which, as the name suggests, brings high-performance local AI compute close to users at their desks while complementing on-device, data-center, and cloud resources.
The workstation series includes the Dell Pro Max with GB10 and Dell Pro Max with GB300 AI accelerators, which can be configured in tandem with Dell Pro Precision towers for increased compute capacity.
Both workstations offer distinct capabilities, however, and it’s worth noting that these are designed around a range of factors such as business needs, performance, and capacity.
Here’s what to expect with the Dell Pro Max GB10 and GB300.
Under the hood of the Dell Pro Max deskside agentic AI workstations
The Dell Pro Max workstation devices are powered by NVIDIA’s Grace Blackwell architecture. Blackwell-architecture GPUs boast 208 billion transistors and are manufactured using a custom-built TSMC 4NP process, per official NVIDIA materials.
Notably, the Blackwell architecture is designed specifically to contend with “data center-scale reasoning AI workflows” and demanding AI training and inference workloads, delivering substantial improvements in performance and energy efficiency over previous-generation platforms.
The two systems offer different software configurations. Dell Pro Max with GB10 runs NVIDIA DGX OS on Linux with the NVIDIA AI software stack. This is a unified AI development solution designed for use in AI factories, which NVIDIA says offers “unparalleled performance, scalability, and innovation.”
The Dell Pro Max with GB300 is available with Ubuntu 24.04 LTS with NVIDIA AI Developer Tools.
While both workstations harness the same underlying features from NVIDIA, there are differences in terms of capabilities.
Dell Pro Max with GB10
The Dell Pro Max with GB10 comes in a small form factor, but it punches well above its weight. Dell specifically highlights this device as great for individual AI development work such as prototyping, fine-tuning, RAG and local inference.
Powered by the NVIDIA GB10 Grace Blackwell Superchip, it can handle LLMs up to 200 billion parameters.
On the agentic AI front, the device is capable of powering eight agents concurrently. Elsewhere, the device boasts up to 128GB LPDDR5x unified memory, and offers one petaflop (1000 teraflops) of FP4 computing power.
As with the Dell Pro Max with GB300, the workstation also comes equipped with NVIDIA’S DGX OS software stack platform.
Dell Pro Max with GB300
The Dell Pro Max with GB300 is the larger of the two workstations, designed for bigger workloads, larger AI models, and capable of more complex agentic orchestration tasks.
The Dell Pro Max with GB300 is powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip. It delivers 20 petaFLOPS of FP4 AI performance and includes 748GB of coherent memory, comprising 252GB of HBM3e GPU memory and 496GB of LPDDR5x CPU memory.
As noted, these devices are designed for more complex workloads and tasks. The Dell Pro Max with GB300 can support models containing up to one trillion parameters and as many as 150 concurrent agents.
These capabilities make the device a powerful piece of equipment to help supercharge agent prototyping, fine-tune LLMs, and optimize inference processes—and all on a single desk.
The GB300 also includes Dell’s revolutionary MaxCool thermal technology. This is a parallel liquid-cooling architecture aimed at maintaining performance without throttling, according to Dell.
Official materials from Dell note this helps the device sustain an optimal temperature by removing heat up to five times more efficiently.*
Swipe to scroll horizontally
Header Cell - Column 0
Dell Pro Max with GB10
Dell Pro Max with GB300
Best for
Individual prototyping, fine-tuning, RAG and local inference
Large-model development, post-training and multi-agent orchestration
AI performance
1 petaFLOP FP4
20 petaFLOPS FP4
Memory
128GB unified memory
748GB coherent memory
Model and agent scale
30–200B parameters; up to 8 agents
Up to 1T parameters; up to 150 agents
Software options
NVIDIA DGX OS
NVIDIA DGX OS or Ubuntu Pro
Potential cost savings*
Up to 28% for low-complexity or 76% for medium-complexity workloads
Up to 56%, 86% or 87% for low-, medium- or high-complexity workloads
Dell Pro Precision towers
The Dell Pro Max GB10 and GB300 workstations are designed specifically to work in tandem with Dell Pro Precision tower workstations. There are a number of key advantages here for enterprises, particularly in scalability and compute capacity.
Dell Pro Precision T2, T4, and T6 form complementary tiers within the broader Dell Deskside Agentic AI portfolio with scalable performance options. They provide a pathway from individual experimentation on Dell Pro Max with GB10 to higher-capacity, team-level development and multi-agent orchestration.
Dell Pro Max with GB10 can also operate as a connected AI accelerator alongside an existing workstation. In this configuration, developers can offload AI-specific workloads—such as fine-tuning, retrieval-augmented generation, and inference—from their primary workstation to GB10, allowing design, coding, and content-creation work to continue without being slowed by demanding AI tasks.
This offers huge advantages for enterprises by reducing the need to balance performance and compute capabilities with the pace of innovation.
Notably, two Dell Pro Max with GB10 systems can be connected using ConnectX-7 networking to tackle models containing up to 400 billion parameters. This, according to official Dell materials, further enables users to scale capacity and performance.
Choosing the right option
For enterprises, choosing between the Dell Pro Max with GB10 or GB300 is essentially down to whether each device meets their specific needs.
The larger GB300 range targets larger enterprises and workgroups, enabling them to tackle frontier-scale inference tasks, post-training processes, and run larger arrays of autonomous agents.
For smaller teams and individual developers, the Dell Pro Max with GB10 still delivers significant performance capabilities, albeit on a smaller scale for agile teams and individual developers.
The launch of both devices comes at a critical time for enterprises of all sizes, however. AI costs have risen significantly over the last 12-18 months, particularly since the advent of agentic AI.
Analysis from Signal65, for example, found that agentic workloads use between four and fifteen times more tokens compared to traditional chat-based interactions. The consultancy modeled persistent agent deployments over two years and compared the cost of on-premises Dell infrastructure with public cloud APIs.
Benchmark testing conducted by the consultancy found Dell Pro Max with GB300 users could see up to 87% lower costs for a high-complexity, AI-assisted software-development workload supporting approximately 60 agents.
Dell Pro Max with GB10 delivered up to 28% lower costs for a low-complexity knowledge-worker workload supporting eight agents. It also delivered up to 76% lower costs for a medium-complexity sales-agent workload supporting four agents. Remember that actual savings will vary according to workload complexity and other modeling assumptions.
This is where both Dell Pro Max options represent a relative game-changer for enterprises of all sizes.
A comprehensive ecosystem
In addition to the performance and cost efficiency of the Dell Pro Max range, enterprises also benefit from the broader Dell and NVIDIA ecosystems.
Both devices are designed to integrate seamlessly with the Dell AI Factory architecture, for example, positioning the Pro Max series as another key component within this end-to-end portfolio.
Moreover, with access to the NVIDIA NemoClaw and OpenShell software stack, along with integrated solutions such as the CrowdStrike Falcon platform, enterprises can drive AI innovation safely and securely.
If you think Dell Pro Max with GB10 or Dell Pro Max with GB300 are the right solutions for your AI development workflows, find out more on the Dell website:
*Based on internal analysis and testing compared with a reference cooling solution. February 2026.
TOPICS
Get the ITPro daily newsletter
Sign up today and you will receive a free copy of our Future Focus 2026 report - the leading resource for IT decision-maker insight on priorities and investment areas in AI, security and more.
5 reasons to add a deskside AI accelerator like the Dell Pro Max with GB10
How to catch fake North Korean IT workers
Which form factor? When to choose the Dell Pro 7 in 13in, 14in, or 2-in-1 design
The Dell XPS 14 is back with a vengeance – but is this high-end laptop worth the premium price tag?
How business leaders are using the Dell Pro 7 and Dell Pro 5
How small businesses can maximize productivity with Dell Pro laptops – even without an IT team
What IT leaders need to know about future-proofing with Dell Pro laptops
How Dell Pro 7 processors and memory are optimized for AI applications
The memory shortage is hitting PC sales hard, but vendor revenues are still growing at the expense of consumers
