The Pulse
Microsoft and NVIDIA Put RTX Spark PCs on Preorder
Microsoft and NVIDIA opened preorders October 7 for RTX Spark Windows PCs, including the $2,599 Surface Laptop Ultra, with laptop shipments starting October 16. The companies also announced Windows security controls for AI agents and previe
AI.info Team · Written with AI ·
Microsoft says Windows PCs run 2 trillion local inferences a month
Copilot+ PCs already handle more than 2 trillion AI inferences locally each month, Microsoft said October 7 as it introduced a new tier of Windows machines built around NVIDIA’s RTX Spark chip. The new laptops and compact desktops are aimed at developers, creators and AI users who want to run larger models and agents on their own hardware, while retaining access to cloud computing when a task needs it.
Microsoft and NVIDIA opened preorders for RTX Spark PCs on October 7. Laptop shipments are scheduled to begin October 16, while compact desktops are due in November, according to NVIDIA. Microsoft’s new Surface Laptop Ultra starts at $2,599; the Surface RTX Spark Dev Box is listed at $5,999 and is scheduled to ship in November in the United States.
RTX Spark puts up to 128GB of memory in a laptop
The RTX Spark combines an NVIDIA Blackwell graphics processor with up to 6,144 cores and an NVIDIA Grace CPU with up to 20 cores. NVIDIA lists up to 128GB of unified memory and one petaflop of FP4 AI performance for the platform. Microsoft says the Surface Laptop Ultra can run AI models exceeding 120 billion parameters locally, though the amount of memory available to a model depends on the device configuration and workload.
The launch spans products from Acer, ASUS, Dell, HP, Lenovo, Microsoft, MSI and Gigabyte. Microsoft’s announced lineup includes the ASUS ProArt P14 and P16, Dell XPS 16 Creator Edition, HP OmniBook Ultra 16, Lenovo Yoga 9n 2-in-1 and MSI Prestige N16 Flip AI+, alongside its Surface devices. The companies are positioning the machines for AI work as well as creative software and gaming, rather than as dedicated model servers.
Windows adds controls for agents running in the background
Microsoft also announced general availability of Microsoft Execution Containers, which give Windows a way to contain agents running in the background under operating-system control. The company says its broader approach combines containment with identity and management controls, so organizations can restrict what agents access and identify which agent performed an action.
“We needed to make the desktop the most secure place for agents to execute,”
Satya Nadella, Microsoft CEO
Windows is also being prepared to route work between local models and cloud services. Microsoft says Copilot will be able to use local context and take actions with a user’s permission; those hybrid-intelligence features are expected to begin rolling out to Copilot+ PCs in the coming months. The company has not given a specific release date for those features.
DGX Station brings larger workloads to Windows
For heavier development and enterprise work, NVIDIA and Microsoft previewed DGX Station for Windows, a deskside system built around NVIDIA’s GB300 Grace Blackwell Ultra Desktop Superchip. NVIDIA lists up to 748GB of coherent memory and up to 20 petaflops of FP4 AI compute, enough, the companies say, to run models at up to a trillion parameters locally. Microsoft says teams could use the system to run multiple agents and keep sensitive data within their own environment.
The announcements extend Microsoft’s pitch beyond laptop hardware: its Windows post says Copilot+ PCs already process billions of local inferences, while RTX Spark targets workloads requiring substantially more memory and compute. The distinction matters for buyers deciding whether a local model can handle their work, whether an agent should be permitted to act on files or applications, and when a cloud service is still needed. Microsoft’s new Copilot features are not yet available, while the first RTX Spark laptops are due to start shipping October 16.