Microsoft's $2,599 Surface Laptop Ultra Bets on Nvidia RTX Spark
Microsoft's Surface Laptop Ultra starts at $2,599, ships October 16, and runs Nvidia's RTX Spark SoC with up to 128GB of unified memory for local AI work.

Updated
Why it matters
- Surface Laptop Ultra starts at $2,599 and ships October 16
- It is Microsoft's first PC built on Nvidia's RTX Spark SoC
- Configurations offer a 5120-core or 6144-core Blackwell GPU with up to 128GB LPDDR5x unified memory
- Microsoft demonstrated Gears of War: E-Day running on the system without a traditional gaming GPU
- The device was announced at Microsoft's first live event in two years, alongside Windows 11 changes
Microsoft's new Surface Laptop Ultra starts at $2,599, ships October 16, and will be among the first PCs built on Nvidia's RTX Spark SoC. The company announced the device today at its first live event in two years, alongside a host of coming changes to Windows 11 and a vision of local-AI and agentic workflows reshaping personal computing for developers and home enthusiasts.
The announcement matters because it marks a concrete step in the shift toward unified-memory, AI-capable personal computers — machines designed to run large models and agentic workflows locally rather than relying entirely on the cloud. Microsoft is positioning the Ultra squarely at the developer community, while arguing the same hardware can serve mainstream users.
What hardware did Microsoft announce?
The Surface Laptop Ultra was originally teased in May 2026. It is Microsoft's first offering to take advantage of the new Nvidia RTX Spark system-on-a-chip.
Key specifications from the announcement:
- Starting price: $2,599
- Ship date: October 16, with pre-orders open now
- GPU: either a 5120-core or 6144-core Blackwell GPU depending on configuration
- Memory: up to 128GB of LPDDR5x unified memory
- Multiple SoC and memory configurations available
The unified memory architecture is the central design decision. By leaning on unified memory allocation rather than the physical VRAM limits of a traditional GPU, the laptop can assign its resources dynamically across workloads.
Why does unified memory matter for AI work?
Traditional discrete GPUs are constrained by their own onboard VRAM, which limits the size of AI models that can run locally. The Ultra sidesteps that ceiling by drawing on up to 128GB of shared LPDDR5x memory.
Microsoft said this flexibility lets the machine handle modern gaming tasks alongside local AI development and deployment — two workloads that historically demanded separate hardware.
The stakes are practical. Local AI development and deployment have grown into a significant use case for developers, and hardware that can serve both inference and everyday computing tasks reduces the need for dedicated machines or cloud dependencies.
Can it really run AAA games?
Microsoft used the presentation to demonstrate graphical performance as proof of the laptop's capability. The company showed the new Gears of War: E-Day running on the system.
The demonstration carried a clear message: even without a traditional gaming GPU, AAA titles are on the table. Microsoft highlighted this gaming showcase to show just how capable the laptop can be, even though the device is clearly focused on the developer crowd.
What else came out of the event?
Beyond the hardware, Microsoft outlined a host of changes coming to Windows 11. The company also shared its vision of how local-AI and agentic workflows could reshape personal computing for the developer community and home enthusiasts.
The event breaks a two-year gap in live Microsoft presentations, and its emphasis — AI-first silicon, unified memory, and agentic workflows — signals where the company expects personal computing to move next.
With shipping set for October 16 and pre-orders open today, the Surface Laptop Ultra will be an early test of whether the RTX Spark platform can make local AI development a mainstream buying decision at a $2,599 entry price.
Original: microsoft.com
More from Elena Vasquez
Show full bio
Market editor covering media and advertising at AI In Context.
174 articles
Related articles
- Microsoft Gathers Nadella and NVIDIA's Jensen Huang for AI PC Reset
- Microsoft Quietly Retires the Copilot+ PC Brand From New Surface Lineup
- Nvidia Raises 7-Year-Old Shield TV Pro Price by $100, Blames AI Memory Costs
- AWS and OpenAI Sign $38 Billion Multi-Year Compute Deal
- OpenAI Commits to 6 Gigawatts of AMD GPUs in Multi-Year Deal