Microsoft prices Surface Laptop Ultra from $2,600, debuts RTX Spark workstation

Microsoft put a number on its first Nvidia RTX Spark-powered PCs on Wednesday, pricing the new Surface Laptop Ultra from $2,600 to $5,900 and its Surface RTX Spark Dev Box workstation at $6,000. The company unveiled both machines at an event in San Francisco during the city's Tech Week, six months after Nvidia announced chip supply agreements with Microsoft and other PC makers for AI- and agent-ready Windows hardware.

What is the Surface Laptop Ultra, and what does it cost?

The Surface Laptop Ultra ships in two base configurations:

• A model starting at $2,600
• A more powerful chip variant starting at $3,700

Buyers can configure either with extra memory and storage up to a top-end price of $5,900. Microsoft said that highest-end configuration is already out of stock.

The pitch is local AI. Microsoft says the laptop can run AI models on-device using its CPU, GPU, and unified memory, without paying for cloud inference. The hardware includes cooling changes Microsoft considers necessary to sustain those workloads.

What does the $6,000 workstation offer developers?

The Surface RTX Spark Dev Box is a developer-class machine built around the same Nvidia chip. It carries Microsoft dev tools out of the box:

• Visual Studio Code
• GitHub Copilot CLI
• Windows Subsystem for Linux (WSL)
• PowerShell 7

The Dev Box also targets on-device model execution. Microsoft is chasing the developer cohort that has drifted toward Apple silicon, a migration visible in Silicon Valley offices and at AI meetups. To sweeten the trade, Microsoft will take up to $1,000 off when they hand over a MacBook Pro.

What is new in Windows 11?

The PCs ship with a revamped Windows 11 that adds a feature Microsoft calls Execution Containers. Microsoft says the feature sandboxes AI agents, isolating their actions from the rest of the system.

Microsoft CEO Satya Nadella said Execution Containers will roll out to all Windows 11 users, not just buyers of the new Surfaces. That widens the sandbox story from a hardware pitch to a platform play. Every Windows user gets a container model for running model-driven software.

What did Nadella say about agents?

Speaking at the event, Nadella framed the company's strategy around orchestration, not raw model access:

"One of the things we realized in the last 3-4 years is that just having a model doesn't do much for anything. You really do need to orchestrate, and you need to have memory outside of the model. You need to have this harnessed layer that is able to take multiple models plus context, plus memory, and the action space."

Nadella added that Microsoft is enabling that orchestration "not just for our apps but apps for anyone's agent. That's what the Windows platform is all about going forward."

The phrasing matters. Microsoft is positioning Windows as the substrate for third-party AI agents, not only for Copilot. Execution Containers give the company a mechanism for enforcing policy, security, and resource limits on those agents, regardless of who builds them.

Who else is shipping RTX Spark hardware?

Dell used the same Wednesday to disclose specs and pricing for its previously announced Dell XPS 16 Creator Edition. The machine is available for pre-order at Best Buy, with shipments scheduled for a later October delivery. Dell priced the Creator Edition at $3,800.

The two announcements together mark the first wave of retail RTX Spark PCs. Nvidia announced chip supply agreements with Microsoft and a roster of PC makers back in June, but machine-level details were thin at the time. Wednesday's event filled in the blanks for two of the bigger names in that group.

Why does the hardware matter beyond the spec sheet?

The GPUs on these machines push enough performance to host modern open-weight models locally. That shifts inference cost from a recurring cloud bill to a one-time hardware purchase. For developers iterating on agents, that removes a credit-card gate between prompt and prototype.

It also puts pressure on Apple's developer base. Microsoft is naming the audience directly with the MacBook Pro trade-in. The Cupertino hardware has become a default in Silicon Valley developer culture, in part because of its memory bandwidth and battery life. Microsoft is now matching the memory architecture claim and bundling the dev tooling it already controls.

Microsoft also positioned the machines for content creators, video processing, and gaming. The RTX Spark silicon doubles as a competent creator GPU, which broadens the addressable market beyond AI developers and reduces the SKU's dependence on a single buyer profile.

What is the competitive context?

The launch lands in a year when every major PC maker has raced to ship AI PCs with neural processing units. Most of those machines target lightweight on-device assistants. The RTX Spark line targets heavier workloads: large language models, multimodal pipelines, and agent runtimes that would otherwise consume cloud credits.

Microsoft's bet is that the developer audience will pay a premium for a workstation that can host a model locally, sidestep cloud costs, and integrate cleanly with Windows. The $6,000 Dev Box is the cleanest expression of that bet. The Surface Laptop Ultra, starting at $2,600, is the volume play.

Execution Containers are the long game. Microsoft is laying the platform groundwork before the agent ecosystem matures. A sandboxed, orchestrated agent runtime is the kind of feature that looks pedestrian in 2025 and indispensable once third-party agents hit the Windows desktop in volume.

Dell joining the RTX Spark lineup the same day signals that Microsoft's positioning is not a one-off Surface experiment. Two of the largest Windows OEMs are now selling Nvidia RTX Spark machines at price points from $2,600 to $6,000, with a Windows-level sandbox feature that travels to every Windows 11 user, not just buyers of these PCs. The agent platform fight is moving from cloud APIs to local silicon, and Microsoft is laying claim to the orchestration layer before the developer community settles on a default.