Microsoft's New Laptop Can Run Powerful AI Right on Your Desk

Microsoft announced the Surface Laptop Ultra, a new Windows laptop that runs on an Nvidia chip called RTX Spark, at an event in San Francisco on October 7. The cheapest version costs $2,599 and goes on sale October 16. The Verge
Microsoft's CEO and Nvidia CEO Jensen Huang were set to show it off together. Reuters Nvidia supplies the computing chip. Microsoft supplies Windows and the laptop design.
The $2,599 model has an 8-core processor, 24GB of memory and 512GB of storage. The Verge Microsoft says it can do up to one petaflop of AI work, a measure of AI speed, and describes the chip as combining an ultra-efficient CPU with a powerful RTX GPU. Microsoft Other versions offer up to 128GB of unified memory, where the processor and graphics share one large memory space. Microsoft
Early reports pointed to higher-spec options in the same body, with 18 cores and 5,120 graphics cores, or 20 cores and 6,144 graphics cores. Windows Central Nvidia has said these laptops can be as thin as 14 millimeters and as light as three pounds, with 14- to 16-inch screens. Nvidia
The old Surface charging port is gone. The Laptop Ultra uses magnetic USB-C charging instead. The Verge For offices that manage many laptops and for firms that make chargers, that removes a special Microsoft plug and moves charging to standard USB-C.
Microsoft also showed the Surface RTX Spark Dev Box, a small desktop computer for software makers. It can be preordered for $5,999 and ships in November. The Verge Microsoft says it also offers up to one petaflop of AI work and has 128GB of shared memory. Microsoft Nvidia also sells RTX Spark on its own as a small, very efficient desktop. Nvidia
For software, Microsoft showed Windows Hybrid Intelligence, which lets the Copilot assistant use files on your PC to do things for you, plus a new Windows search that lets you take quick actions right from the search bar. The Verge Nvidia says RTX Spark will power Windows PCs made for personal agents, small AI helpers that do tasks for you. Nvidia
Stepping back, what matters here is that a lot of shared memory and fast AI processing now fit inside a normal laptop. That makes it easier to run big AI tools on the computer itself and avoid slowdowns from moving data between separate memory spaces. Developers can test new AI helpers on the laptop or the desk box using the same files, search and Copilot they will use later. In my view, there is reason for hope. If the drivers, Nvidia's developer tools and support for Windows apps on these chips work well day to day, more useful assistants can run without checking in with the cloud every time.


