- The Surface Laptop Ultra is the first consumer device featuring the Nvidia RTX Spark SoC, starting at $2,599.
- It utilizes a unified memory architecture (up to 128GB), allowing the GPU to access system memory directly for large AI models, bypassing traditional VRAM limits.
- Targeted explicitly at developers, it supports local agentic AI workflows and AAA gaming performance.
AI CURATED HOT STORY · 10/8/2026
Microsoft debuts Surface AI PCs with Nvidia RTX Spark: Blackwell GPU and 128GB unified memory enable free local LLM execution
Microsoft launches Surface Laptop Ultra and Dev Box powered by Nvidia RTX Spark SoCs, marking a new era of local large model inference in consumer AI PCs. Starting at $2,600 with top configurations reaching $5,900, these devices allow free local execution of large AI models without cloud subscriptions. High-end models sold out immediately, indicating strong market demand.
Technically, the Surface Laptop Ultra is the first consumer device featuring RTX Spark, integrating a Blackwell architecture GPU with up to 128GB LPDDR5x unified memory. This design breaks traditional VRAM limits, allowing direct GPU access to system memory for handling large parameter models, significantly optimizing local AI development and high-performance gaming workflows. The Dev Box deeply integrates Microsoft’s developer stack, including VS Code, GitHub Copilot CLI, and WSL, providing end-to-end support for building local agentic applications.
This move reshapes expectations for local compute costs and solves VRAM bottlenecks for local LLM deployment through hardware-level unified memory design. With improved pre-installed toolchains, Microsoft is accelerating the shift of AI capabilities from cloud to edge, laying the hardware foundation for independent developers and enterprises to build private, low-latency agentic ecosystems.
Coverage timeline
Total 2 reports · Latest first- Ars Technica AIT2·68 pts
- TechCrunch AIT2·68 pts
- Both Surface Laptop Ultra and Dev Box feature Nvidia RTX Spark chips, emphasizing local, cost-free AI model execution.
- Pricing ranges from $2,600 to $5,900, with top-tier models already out of stock, indicating strong initial demand or supply constraints.
- The Dev Box deeply integrates Microsoft's development stack (VS Code, GitHub Copilot CLI, WSL), optimized for building local agentic applications.