Jensen Huang’s remark was a tribute to the platform that gave Nvidia an early market: Windows PCs. He was not saying Microsoft founded Nvidia. At an October 7, 2026 event, Huang and Microsoft CEO Satya Nadella connected that history to a new joint effort to run AI workloads on Windows devices rather than relying on the cloud for every task.
27
31
8
What Huang meant by “Nvidia was founded because of Windows”
Huang recalled Nvidia’s origins taking shape in late 1992, when Windows 3.1 was emerging; Nvidia was founded in 1993. The point was that Windows helped make the PC graphics market Nvidia set out to serve possible.
29
31
The companies’ story then moved from Windows PCs and graphics toward more capable GPUs. Windows 95 and DirectX feature in Huang’s account of that development; later, CUDA helped make Nvidia GPUs useful for computation beyond graphics, a capability that became important to AI. The history is not a claim that Windows alone created Nvidia’s later business. It is Huang’s shorthand for how a software platform helped establish the market and use case from which Nvidia grew.
29
31
The market-cap joke—and what it does and doesn’t say
After Huang credited Windows, Nadella joked about transferring some market capitalization “back” to Microsoft. The punchline was that Nvidia had grown into a company of extraordinary value after beginning in a market Windows helped enable. It was banter, not a literal proposal or a way to measure how much value Windows created.
27
31
Company valuations change over time, and the cited event coverage does not establish a single comparable market-cap snapshot. The joke is best read as a playful acknowledgment of how dramatically the companies’ relative fortunes had shifted—not as a precise financial comparison.
27
28
What Microsoft and Nvidia announced for AI PCs
The event turned the historical connection into a current product strategy: Microsoft and Nvidia said they are co-engineering hardware and software for AI agents on Windows PCs.
8
The announcements included:
- RTX Spark-powered PCs: Microsoft’s Surface Laptop Ultra is specified with an Nvidia Blackwell GPU with up to 6,144 cores, a Grace CPU with up to 20 cores, and up to 128 GB of unified memory. Nvidia describes RTX Spark as offering up to 1 petaflop of AI compute.
9
40
- A new Surface laptop: Microsoft opened preorders for the Surface Laptop Ultra at $2,599 and announced availability for October 16, 2026.
7
9
- Agent safeguards: Microsoft Execution Containers are intended to contain agent workloads and help prevent agents from accessing data without permission.
4
- More form factors: Nvidia said RTX Spark systems would extend to compact desktops, with units expected to be available in November. It also previewed DGX Station for Windows, aimed at enterprise use.
8
16
What local AI could change—and what remains unproven
Microsoft also announced an AI coding model that can run directly on PCs. Paired with hardware designed for on-device workloads, that makes it possible for some coding and agent tasks to run locally instead of being sent to cloud services.
4
7
Local processing does not mean every AI task will move off the cloud, or that every announced PC will handle every agent workload well. The October 7 event presented a product direction and availability plans; real-world performance and usefulness will depend on the devices, software, and tasks people actually use.
4
8
9
That is the connection between Huang’s comment and the new partnership: Windows helped provide Nvidia with an early platform for PC graphics, and the companies now want Windows PCs to become a platform for running more AI locally.