AI · Oct 7, 2026
NVIDIA, Microsoft Kick Off a New Beginning for Windows PCs With RTX Spark and AI AgentsMicrosoft releases new Nvidia-chip AI PCs with revamped Windows 11
Oct 7, 2026, 1:22 PM · TechCrunch

The pricey Nvidia-powered Surface machines are the headline, but the bigger move is a Windows feature to sandbox AI agents, and that one ships to everyone.
Why it matters
At a San Francisco event during the city's Tech Week, Microsoft gave specs and prices for its first PCs built on Nvidia's RTX Spark chip, months after Nvidia announced the partnerships in June. The Surface Laptop Ultra starts at $2,600, or $3,700 for a more powerful chip, and runs up to $5,900; Microsoft says that top configuration is already out of stock. A Surface RTX Spark Dev Box workstation starts at $6,000 and comes loaded with developer tools including VS Code, GitHub Copilot CLI, WSL and PowerShell 7.
The machines are built to run AI models locally, and they ship with a revamped Windows 11 that includes Execution Containers, a feature Microsoft says makes it easier to sandbox AI agents. Satya Nadella said that feature will come to all Windows 11 users.
From the desk
We think Execution Containers is the most important line in this story. Agents that can click, type and run code on a personal computer are only as safe as the walls around them. Building a sandbox into the operating system, rather than leaving each app to improvise, is the kind of unglamorous plumbing that makes local agents something a careful person or IT department could actually allow. If it works as described, it is a genuine step forward for everyone on Windows, not just buyers of a $6,000 box.
Nadella's framing is worth noting too. He argued that a model alone does little, and that the real work is the orchestration layer around it, with memory, context and the ability to act. Microsoft wants Windows to be that layer for anyone's agent. That is a reasonable bet, and it puts Windows in direct competition with the browser and with cloud agent platforms for where agents live.
The hardware story is narrower. These are expensive machines aimed squarely at developers, and Microsoft is openly courting Mac users with up to $1,000 off for trading in a MacBook Pro. Local inference is good for privacy, latency and cost once the hardware is paid for. But at these prices, running models on-device for free is a benefit mostly for people who can already afford it.
The risk we would name is the one that comes with any agent platform at operating-system scale. A sandbox shifts the question from whether an agent can do damage to how good the boundaries are. Details on what Execution Containers isolate, and what agents can still reach, were not in this report. I'm watching for that documentation, and for how security researchers test it once it reaches ordinary Windows 11 machines.
Context
Nvidia said in June it had agreements with Microsoft and other PC makers to build AI- and agent-ready Windows PCs on RTX Spark. Dell also detailed its Dell XPS 16 Creator Edition on the same day, priced at $3,800 and available for preorder at Best Buy for later-October delivery.
Who feels it
- Developers
- High-end local hardware plus built-in agent sandboxing makes Windows a more serious option for building and testing agents, at a steep price.
- Everyday Windows 11 users
- Execution Containers could make running AI agents on a personal PC safer, depending on how strong the isolation proves to be.
- Enterprise IT
- An OS-level sandbox gives administrators a clearer control point for agent software on employee machines.
- Apple
- Microsoft's MacBook Pro trade-in offer targets the developer audience that has favored Macs for local AI work.
What to watch
- When Execution Containers reaches general Windows 11 users and what it isolates
- Independent security testing of the agent sandbox
- Real-world local inference performance of RTX Spark machines
- Other PC makers' RTX Spark pricing and availability