Microsoft and Nvidia Reveal Surface Laptop Ultra and RTX Spark AI Hardware

Microsoft and Nvidia unveiled the Surface Laptop Ultra and new RTX Spark developer hardware during a San Francisco event on October 7, 2026. Starting at $2,599 with shipments beginning October 16, the machines introduce hybrid AI capabilities, local model orchestration, and Microsoft Execution Containers for agent security on Windows 11.

Microsoft and Nvidia Reveal Surface Laptop Ultra and RTX Spark Chips in San Francisco

Microsoft CEO Satya Nadella and Nvidia CEO Jensen Huang outlined a co-engineered platform designed to run heavy artificial intelligence workloads away from the cloud and on local devices. Huang traced the technological lineage back to the release of Windows 3.1 in late 1992, noting that Nvidia owes its initial creation to the personal computer ecosystem.

At the Windows AI and Surface gathering hosted at San Francisco’s Dogpatch Studios, Nvidia chief executive and founder Jensen Huang mentioned to former White House artificial intelligence policy advisor Sriram Krishnan that GeForce owes its existence entirely to Windows. Huang also noted that Windows has been with him 34 years and looks like it will be with him for 34 more, while Satya Nadella added agreement before they concluded their discussion. During the presentation, Nvidia chief executive Jensen Huang revealed that the computer’s Blackwell RTX Spark graphics processor delivers computational capacity equal to Nvidia’s $250 million DGX-1 data center server from 2016. Pavan Davuluri, executive vice president of Windows + Devices, took the stage to explain that Microsoft is building primitives directly into Windows so that agents can be secured, observed, and governed using MXC, Microsoft Security, and Agent 365.

Windows secure platform for agents

The centerpiece of the hardware rollout is the Surface Laptop Ultra, a 15-inch portable machine powered by an Arm-based Nvidia RTX Spark processor. The device supports up to 128GB of unified LPDDR5X memory and a Blackwell graphics processor containing 6,144 cores paired with a 20-core Grace CPU. Pavan Davuluri, Microsoft’s executive vice president of Windows and devices, demonstrated the hardware running GitHub Copilot alongside demanding applications like the video game Gears of War: E Day and the DaVinci Resolve video editing platform simultaneously.

Pricing Tiers and Enterprise Availability for Surface Laptop Ultra and RTX Spark Dev Box

Microsoft’s entry-level Surface Laptop Ultra begins at $2,599, while configurations climb to $5,900 for high-end builds featuring 128GB of memory and one terabyte of storage. Preorders opened immediately following the event, with general shipments scheduled to begin on October 16.

GitHub Copilot on Windows

Max Weinbach, an analyst at Creative Strategies, wrote on X after the announcement that Microsoft might get him back to Windows.

For professional developers and engineering teams, Microsoft also debuted the Surface RTX Spark Dev Box starting at $6,000. Preorders for companion RTX Spark laptops from manufacturing partners including Dell and HP went live for an October 16 release, with compact desktop configurations following in November. Tom Mainelli, IDC’s group vice president of device and consumer research, observed that cost remains a central consideration for buyers balancing local capabilities against cloud subscriptions.

Microsoft and Nvidia Reveal Surface Laptop Ultra and RTX Spark AI Hardware
Photo: NVIDIA Blog

“Ultimately, the cost is a big driver. But you’re also going to get to a point where you want to make decisions about what data goes to the cloud, what data stays locally, and then you will be thinking about the orchestration layers.”

Tom Mainelli, IDC’s group vice president of device and consumer research, via Yahoo Finance

Microsoft Execution Containers and Hybrid Intelligence Updates for Windows 11

As autonomous AI agents gain popularity, security concerns have intensified around software gaining unauthorized access to private system files and enterprise networks.

The operating system updates also bring intelligent routing capabilities across local silicon and cloud data centers. GitHub Copilot will coordinate requests automatically behind the scenes, determining whether an active task requires cloud scale or local execution. Developers can manually select local models through Windows ML or connect to OpenAI-compatible local endpoints, including Meta’s Nemotron model and open-source software such as OpenClaw.

New Get Started experience for agents on Windows

Nvidia DGX Station for Windows and Enterprise Hardware

Powered by the Nvidia GB300 Grace Blackwell Ultra Desktop Superchip, the deskside supercomputer delivers up to 20 petaflops of FP4 AI compute and supports large artificial intelligence models scaling up to one trillion parameters with coherent memory access.

By handling inference workloads locally rather than routing every query to external servers, corporate engineering groups can preserve sensitive intellectual property within secure internal boundaries. Manufacturers including Dell and HP announced enterprise workstations featuring the GB300 architecture for release later in the year, integrating directly with existing Windows management tooling used across corporate fleets.

New Windows Search Experience

Related reading


Discover more from Archyworldys

Subscribe to get the latest posts sent to your email.