Microsoft Surface & Windows October 2026 Event Live: Key Announcements & Other Details

As announced last month, Microsoft has kicked off the biggest Surface and Windows event of 2026 in San Francisco, and if you are eager to learn more about what the company has announced, keep refreshing the tab. We’ll cover each and every update from the live event.

Surface Laptop Ultra, RTX Spark and Windows AI

The biggest focus is expected to be the Surface Laptop Ultra, which Microsoft has already teased as a high-end machine powered by NVIDIA’s RTX Spark platform. The laptop is aimed at developers and creative professionals, with reports pointing to configurations offering up to 20 CPU cores, 6,144 GPU cores and 128GB of unified memory.

Pricing is likely to be one of the biggest announcements. Reports suggest the entry-level Surface Laptop Ultra could start well above $2,000, although Microsoft has yet to confirm the final US pricing. Exact availability and shipping dates are also expected to be clarified during the event.

NVIDIA CEO Jensen Huang is also joining Microsoft CEO Satya Nadella on stage, putting RTX Spark firmly in the spotlight. Microsoft is expected to demonstrate how the new hardware can run large AI models locally, reducing the need to send demanding workloads to the cloud.

[LIVE UPDATES TO FOLLOW]

10:25 AM PT: The event has shifted into a live demonstration of an AI agent handling code cleanup. The demo shows the agent working through a development task, making this part of the presentation feel more like Microsoft Build than a traditional Windows and Surface event.

10:23 AM PT: Microsoft has now moved into a deeper look at Windows ML, describing it as a scalable way to deploy AI runtimes across CPUs, GPUs and NPUs. Microsoft also confirmed that llama.cpp is coming to Windows ML, giving developers a way to experiment with newer AI models locally on Windows.

10:21 AM PT: NVIDIA is also bringing more AI models to local hardware. During the event, Microsoft revealed that NVIDIA’s Nemotron will get a new model on October 15, while DeepSeek V5 Flash will reportedly be capable of running locally on systems with around 60GB of memory.

10:19 AM PT: GitHub’s HydraFusion is also being extended to AI models running locally on Windows, giving developers another way to route workloads across different models. The move fits into Microsoft’s broader push toward hybrid intelligence, where AI workloads can run either on the device or in the cloud.

10:16 AM PT: Pavan Davuluri says this approach is necessary because users’ appetite for intelligence is effectively unlimited. Microsoft is also discussing the security challenges that come with AI agents, with Davuluri noting that partners including OpenAI and Anthropic are already building on the company’s security tools.

10: 16 AM PT: Microsoft is launching End-to-end platform for securing agents.

10:12 AM PT: Microsoft confirms Execution Containers are now moving from preview to general availability.

10:10 AM PT: Windows and Devices chief Pavan Davuluri is here, and he is talking about secure platform for agents on three principles: Containment, Identity, Manageability.

10:08 AM PT: Nadella is highlighting how Microsoft wants to bring Agent 365, Microsoft IQ, and Copilot together as part of the core Windows architecture, pointing to a future where AI agents are built more deeply into the operating system rather than treated as separate apps.

10:06 AM PT: Microsoft CEO Satya Nadella has taken the stage and says, “My life was shaped by one such event back in 1991,” hinting at the significance of today’s event, which many of you might also be thinking right now.

This is a developing article…