NVIDIA RTX Spark and DGX Station for Windows: What Enterprise IT Teams Need to Know Before Fall 2026

On June 1, 2026, at Computex in Taipei, NVIDIA and Microsoft made the most consequential joint PC-platform announcement in years. Two new products, the RTX Spark superchip and the DGX Station for Windows, reframe the Windows PC from a thin client for cloud AI into a first-class local compute tier for running AI agents on-premises. This is not a gaming story. It is an enterprise infrastructure story with direct implications for hardware budgets, software procurement, IT management, and data governance across every industry.

What Was Actually Announced

NVIDIA's RTX Spark is a new superchip that combines up to 20 Arm CPU cores (co-designed with MediaTek and connected via NVLink C2C), a Blackwell GPU with 6,144 CUDA cores, 128GB of LPDDR5X unified memory, and up to 300 GB/s of memory bandwidth. It delivers up to 1 petaflop of AI compute, enough to run capable on-device AI agents without a cloud connection.

The enterprise product is larger in every sense. The DGX Station for Windows is billed as the world's most powerful deskside AI supercomputer, capable of running frontier models of up to 1 trillion parameters locally. It offers up to 748GB of coherent memory and up to 20 petaflops of FP4 performance. It can also be paired with an NVIDIA RTX PRO 6000 Blackwell Workstation GPU for ray-traced visualization and simulation workloads.

Microsoft matched the hardware with deep platform work. Windows now includes workload profile scheduling (WPS) optimized for RTX Spark, so the OS scheduler can efficiently distribute tasks across all 20 cores. Microsoft and NVIDIA also unlocked TensorRT natively in Windows through Windows ML, giving AI developers direct access to GPU acceleration without leaving the Windows environment.

Why the Timing Matters

Historically, heavy AI workloads, including large-scale inference, fine-tuning, and multi-agent development, have lived on Linux-based data center hardware. The vast majority of Fortune 500 companies run Windows for everyday work. That split has been a real friction point: Windows IT teams could not easily touch the AI infrastructure their data science colleagues were running on separate, Linux-only systems.

DGX Station for Windows addresses that directly. It brings GB300-class compute into a managed Windows environment, meaning the machine slots into existing MDM, endpoint management, and audit tooling rather than sitting as a separate island.

The data sovereignty angle matters just as much. For regulated industries, including healthcare, finance, legal, and defence, local data residency is a compliance requirement, not a preference. A deskside box that runs trillion-parameter models without sending data to a cloud endpoint removes a blocker that has stalled AI rollouts in those sectors for the past two years.

The Broader Platform Play

NVIDIA and Microsoft are pitching a continuous stack, not isolated hardware. The DGX Station for Windows ships pre-loaded with Windows Server 2025 and Azure Arc-enabled agents, so IT administrators can manage it as just another Azure resource. A "Direct Connect" feature announced at Microsoft Build 2026 establishes a private, encrypted tunnel between a Windows 11 AI PC and a DGX Station over a local network, making local AI offload as straightforward as calling a REST API.

Security is built into the architecture via NVIDIA OpenShell, a runtime layer that provides agent containment so a rogue agent process cannot reach data it has no business accessing. New Windows OS security primitives underpin the whole agent platform.

On the software side, NVIDIA AgentKit enters early access in July 2026 with pre-built connectors for ServiceNow, SAP, and Salesforce. Adobe is also rearchitecting Photoshop and Premiere specifically for RTX Spark, targeting 2x faster AI and graphics performance. NVIDIA publicly committed to a multi-generation Spark roadmap at Computex, signalling this is a long-term platform investment, not a one-off product.

When and From Whom

RTX Spark laptops and compact desktops will be available this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE models to follow. Commercial DGX Station for Windows versions from Dell Technologies, ASUS, GIGABYTE, HP, MSI, and Supermicro are targeted for Q4 2026.

That timeline is concrete enough to include in a current budget cycle.

What IT Teams Should Do Now

Start the hardware refresh plan now. RTX Spark devices arrive Q4 2026. Procurement templates, MDM provisioning workflows, and Windows on Arm compatibility checks for internal apps need to be ready before hardware lands on desks.

Evaluate DGX Station for Windows as a regulated-data compute tier. Finance, legal, and healthcare teams should identify AI workloads that currently cannot move to the cloud for compliance reasons and map them to a DGX Station use case. The GPU cloud was the only viable option when models were too large to run locally. That constraint is lifting.

Review your agent security posture before agents deploy. The new Windows containment primitives and OpenShell are a framework, not a guarantee. IT security teams need to define agent permission policies, audit logging requirements, and data access boundaries now, not after rollout.

Watch the ISV ecosystem over the next 90 days. Analyst firm Gartner noted that "the combined Microsoft-Nvidia offering lowers the barrier for enterprise agentic AI, but successful adoption will depend on how quickly the ecosystem of ISVs and system integrators can retool." Track which of your existing software vendors commit to Foundry and AgentKit connectors.

Budget for the memory premium. Server DRAM prices surged by nearly 95% in early 2026 on AI infrastructure demand. RTX Spark and DGX Station use unified memory architectures that are more efficient, but Q4 hardware pricing will still reflect a market under supply pressure. Build that into your models now.

The bottom line is straightforward. NVIDIA and Microsoft have given enterprise IT a credible, managed, on-premises path to serious AI workloads, backed by the same OEMs your procurement team already calls. The planning window is open today. Teams that wait until devices arrive will lose six months.

How 247techify can help

At 247techify, we help businesses plan and execute technology infrastructure upgrades, covering AI-ready hardware evaluations, Windows endpoint management, and on-premises compute strategy. If your team is working out where RTX Spark, DGX Station, or local AI infrastructure fits your roadmap, get in touch and we will work through it with you.