A trillion-parameter AI supercomputer on every enterprise desk

- NVIDIA has positioned the DGX Station for Windows as the world’s most powerful deskside AI supercomputer

- DGX Station brings frontier AI agents to Windows

- DGX Station will support NVIDIA OpenShell on Windows, built on new Windows security and containment primitives 

 

NVIDIA has announced NVIDIA DGX Station for Windows, billing it "the world’s most powerful deskside AI supercomputer designed to build, run and connect always-on AI agents to Windows applications and workflows". The system is capable of running frontier AI models of up to 1 trillion parameters locally.

Historically, heavy-duty enterprise AI workloads — training, fine-tuning, large-scale inference and multi-agent development — have required powerful AI systems in the data centre that run on Linux, while the vast majority of Fortune 500 companies use Windows for everyday productivity, creative, design and engineering applications, NVIDIA observed.

Building on the NVIDIA DGX Station system design, DGX Station for Windows bridges this gap as the first deskside AI supercomputer to bring NVIDIA GB300 Grace Blackwell-class AI infrastructure directly into the Windows ecosystem — providing the compute needed to build, run and connect powerful AI agents to the applications and infrastructure Windows users already harness.

“As enterprises scale AI agents across their organisations, they need AI infrastructure that can connect directly to the applications and workflows that power their business,” said Chris Marriott, VP of enterprise platforms at NVIDIA. 

“DGX Station delivers supercomputing-class AI directly into Windows, where millions already design, engineer, research and create every day.”

“For decades, Microsoft and NVIDIA have partnered to advance the most powerful computing platforms in the world,” said Pavan Davuluri, Executive VP of Windows + Devices at Microsoft on May 31, the date of the launch. 

“Today, we’re taking that collaboration to the next level, scaling the full power of Windows from thin-and-light PCs to data-centre-class workstations with DGX Station powered by GB300. This unlocks a new class of AI performance on Windows, the platform enterprises trust for security, manageability and compatibility.”

The DGX Station design is powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop superchip, connecting an NVIDIA Blackwell Ultra GPU to a 72-core NVIDIA Grace CPU via the NVIDIA NVLink-C2C interconnect for ideal system communication and performance. It features up to 748 GB of coherent memory and up to 20 petaflops of FP4 performance, and can be paired with an NVIDIA RTX PRO 6000 Blackwell Workstation GPU for frontier AI compute with ray-traced visualisation and simulation.

In addition, DGX Station for Windows features the NVIDIA ConnectX-8 SuperNIC, optimised to supercharge hyperscale AI computing workloads. With support for networking at up to 800 Gbps, the ConnectX-8 SuperNIC enables fast network data transfers for AI workloads and high-speed connectivity of multiple DGX Station systems for even larger workloads.  

Always-on AI agents

Enterprise AI is evolving from relatively simple chatbot interactions to agentic inferencing that operates continuously, reasons in real time and connects directly to enterprise applications and workflows, NVIDIA said. Developed in collaboration with Microsoft, DGX Station for Windows serves as dedicated agent infrastructure — enabling enterprises to build and run agents with frontier intelligence locally, supporting AI models of up to 1 trillion parameters. DGX Station can also run agents at scale, with hundreds of agents executing on tasks simultaneously.

For example, powerful AI agents can be built and connected to 3D design and engineering applications, giving developers, designers and engineers intelligent assistants that understand their tools and processes to automate repetitive tasks and accelerate productivity.

For enterprise IT teams, DGX Station for Windows brings a secure, managed platform to GB300 deployments — extending the same Windows security, compliance and fleet management infrastructure organisations already rely on. AI agents deploy and operate within this managed environment, governed through familiar Microsoft tools.

Linux workloads get the same level of manageability support through Windows Subsystem for Linux. Enterprise-grade features, including deployment and system updates, help organisations maintain security, compliance and operational readiness across their fleets. 

Autonomous agents need to be developed and deployed in a secure runtime that governs how they act, how they operate tools and how they interact with the rest of the system. DGX Station is an ideal platform for Windows users to build and run always-on, autonomous agents locally and securely before scaling to data centre AI factories.

NVIDIA OpenShell is an open source, secure-by-design runtime for autonomous agents. Building on the new Windows security and containment primitives, it creates an individual, isolated sandbox for each agent and separates application-layer operations from infrastructure-layer policy enforcement.

This means security and privacy policies are out of the agent’s reach and applied at the system level. Instead of relying on behavioural system prompts, OpenShell uses the new Windows security and containment primitives with the aim to enforce constraints on the environment the agent runs in, so it cannot override policies, or leak credentials or private data.

DGX Station for Windows is built for the full spectrum of enterprise AI workflows — from autonomous agent deployment and frontier model development to high-throughput inference, data science and physical AI, all with the full Microsoft and Windows enterprise manageability stack. 

- AI agents: Build and run multiple frontier agents in parallel, and connect them directly to enterprise applications and workflows.

- AI development: Pretrain, fine-tune and iterate on large AI models within a Windows environment, with access to Linux AI toolchains via Windows Subsystem for Linux. 

- Data science: Ingest large datasets into up to 748 GB of coherent memory, eliminating data movement bottlenecks and accelerating every step from data preparation to machine learning and analytics. 

- AI inference: Run high-throughput inference on AI models, and run large AI models of up to 1 trillion parameters. 

- Physical AI: Pair the GB300 superchip with an additional NVIDIA RTX PRO Blackwell Workstation GPU to combine frontier AI compute with ray-traced visualisation and simulation in a single deskside system — delivering the performance needed for agents to perceive, simulate and interact in virtual-to-physical environments. 

Details

DGX Station for Windows is expected to be available globally from ASUS, Dell Technologies, GIGABYTE, HP, MSI and Supermicro in Q426.

Explore 

NVIDIA and Microsoft’s collaboration to deliver agents in the Windows experience extends from frontier to personal agents with NVIDIA RTX Spark, bringing the full spectrum of NVIDIA AI to slim laptops and small desktop PCs. Read the post about NVIDIA launching NVIDIA RTX Spark superchip for personal AI devices. 

Hashtags: #Computex, #Computex2026, #GTC, #GTCTaipei, #GTCTaipei2026