Microsoft and Nvidia’s RTX Spark Event: What to Watch on Oct. 7

Microsoft has already told us what its next big laptop can do. It just hasn’t said what it costs. On Wednesday, Oct. 7, at 10 a.m. PT (1 p.m. ET), Satya Nadella, Windows and Surface chief Pavan Davuluri and Nvidia’s Jensen Huang share a stage in San Francisco for an event Microsoft says is about “how local AI will shape the next chapter of the PC”, Windows Central reports. That is a big billing for a hardware show, and the guest list explains it.

Per Thurrott, Microsoft’s invitation promises “the latest innovations across Windows, NVIDIA RTX Spark, Surface and the broader PC ecosystem.” Nothing below is confirmed until it happens on stage. Here’s what has been reported, and how we’ll judge it.

The laptop Microsoft already announced

The expected headliner is the Surface Laptop Ultra, which Microsoft introduced on May 31 without a price. On paper it’s the most ambitious Surface in years: a 15-inch mini-LED touchscreen rated at up to 2,000 nits peak HDR brightness, up to 128GB of unified memory, an Nvidia Blackwell RTX GPU with full CUDA support, and what Microsoft calls “1 petaflop of AI compute,” enough, it says, to run models of up to 120 billion parameters locally. The company’s line: “Engineered with NVIDIA from the silicon up, optimized for RTX Spark, built on Windows.” Availability was promised for “later this year.”

The details that matter most are still missing. TweakTown notes the machine is listed as pre-release pending regulatory approval, with no pricing. Notebookcheck expects Microsoft to also show a Surface RTX Spark Dev Box, a desktop pitched against Apple’s Mac Studio, and says both were promised “before the start of 2027 in some capacity.”

What RTX Spark actually is

RTX Spark, also called N1X, is Nvidia’s Arm-based PC chip, pairing Grace CPU cores with a Blackwell GPU. Tom’s Hardware reported on Sept. 3 that it launches in laptops in October, with desktops to follow, in two versions:

• 20 CPU cores, 6,144 CUDA cores and 24GB to 128GB of unified memory, for laptops and mini PCs
• 18 CPU cores, 5,120 CUDA cores and 24GB to 32GB, for laptops only

Lenovo’s Yoga 9n and an Acer mini desktop were shown at IFA, with Asus, Dell, MSI, HP and Microsoft to follow. Tom’s Hardware also lists what Nvidia hasn’t said: power consumption, pricing, exact availability and performance beyond core counts.

On the software side, Microsoft says Windows 11 version 26H2 will be generally available in the second half of 2026, delivered as a small enablement package to PCs on 24H2 or 25H2. Thurrott suggests the event could announce its public rollout. TweakTown reports Windows will get a scheduler for spreading work across RTX Spark’s 20 CPU cores and tuned Prism emulation for older x86 apps. Treat both as reported until Microsoft demos them.

Is this the first real AI PC?

Our buying rule for AI PCs is simple: RAM before TOPS. A bigger neural-processor number gets you a faster small model. It doesn’t let you run a bigger one. To run a large language model on your own machine, its weights have to sit in memory, and the GPU has to be able to read them quickly.

By that test, the top RTX Spark configuration is the first Windows design that looks built for the job. Up to 128GB of memory a GPU can address, plus CUDA, which is what most local AI software is written for first. Microsoft’s 120-billion-parameter claim is a memory claim, and an honest one in principle.

The fine print decides whether it’s real for you. Three things to watch on Wednesday:

• The price of memory. The cheaper RTX Spark version stops at 32GB, and both start at 24GB. If 128GB arrives only at the top of the range, most buyers get a fast laptop, not a local-AI machine.
• Memory bandwidth. It sets how fast a model can talk back, and neither company has published it.
• The petaflop’s fine print. The announcement we read doesn’t say at what numerical precision that figure is measured. Low-precision peak numbers flatter every chip.

Our read: the first RTX Spark laptops are likely to be judged as Windows-on-Arm laptops before they’re judged as AI machines. App compatibility, battery life and price will decide the reviews. That is where Qualcomm’s Snapdragon machines have spent two years fighting.

Microsoft first showed this laptop more than four months ago. The one number it has held back is the price.