Microsoft and NVIDIA detail Ignite’s AI sessions ahead of November 17 keynote
The confirmed program covers AI infrastructure, agent development and physical AI. Remote attendees can watch several sessions, but the hands-on lab will not be streamed or recorded.
Microsoft and NVIDIA’s Ignite program shows how their Azure and AI stacks are being positioned across accelerated computing, agent development, and cloud-to-edge deployment—but the agenda is technical, not a product-launch announcement. Three joint sessions are livestreamed, while the Nemotron–Microsoft Foundry lab is limited to onsite attendees and overlaps the first talk. Remote viewers can watch the physical-AI session on demand; Microsoft cautions that session times may change.
01
NVIDIA’s Dave Salvator leads the accelerated-computing session Tuesday, November 17, from 3:15–4:00 p.m. Pacific.
02
The November 18 agent session combines Microsoft Foundry, NVIDIA Nemotron models, and NVIDIA infrastructure on Azure.
03
The onsite-only lab runs November 17 from 2:30–3:45 p.m., overlapping the accelerated-computing session from 3:15 p.m.
Microsoft Ignite 2026 runs November 17–20 at San Francisco’s Moscone Center and online, opening with Microsoft’s keynote on November 17 at 10:00 a.m. Pacific Time. To watch online, follow the attendance-registration link in Microsoft’s official event guide. The NVIDIA program that follows includes streamed AI sessions, plus hands-on activities reserved for people attending in person.
Microsoft’s November 12 preview directs viewers to registration for online attendance access. Sessions labeled “In person + online” are open to remote attendees, while recorded sessions can be watched on demand. Check that attendance label when building a schedule: an in-person session may have a recording afterward without being available to watch live.
Three online sessions to put on the calendar
The confirmed joint program spans the computing behind AI, software agents that carry out tasks, and systems running across cloud and local infrastructure. These are scheduled technical sessions, not confirmations of new product releases. For online viewers, the following three talks offer distinct routes into that program. All times below are Pacific Time.
Tuesday, November 17, 3:15–4:00 p.m.: Power accelerated computing on Azure with NVIDIA. NVIDIA’s Dave Salvator will cover AI, data science and simulation workloads using the company’s GPUs, software and services. This is the session focused on the computing platform and the different kinds of work it supports. Attendance is available in person and online.
Wednesday, November 18, 11:30 a.m.–12:15 p.m.: Architect reliable and secure agentic AI at scale. The session will show how Microsoft Foundry, NVIDIA Nemotron models and NVIDIA infrastructure on Azure fit together for building AI agents. It is available in person and online, separately from the hands-on lab.
Thursday, November 19, 2:45–3:30 p.m.: Scaling intelligence cloud to edge with agentic AI factories. NVIDIA’s Dion Harris will discuss infrastructure co-engineered for Azure datacenters and Azure Local. The agenda also includes efficiency measured in tokens per watt and agent security using NVIDIA open models and Open Agent Safety. Attendance is available in person and online.
Physical AI has a different viewing arrangement
Industries reimagined with physical AI is scheduled for Thursday, November 19, from 10:15–11:00 a.m., with speakers from Microsoft and NVIDIA. That session is in person only, but it will be recorded. Remote viewers interested in this part of the program therefore have an on-demand option rather than live online attendance.
For visitors at Moscone Center, NVIDIA will also occupy booth 5644 and plans 30 in-person theater sessions from November 17–19. That show-floor program is another part of the joint presence beyond the scheduled breakout talks and lab.
Before committing to a session, check the latest listing through Microsoft’s schedule and add it to your event calendar. Microsoft says session times are subject to change.
Reader comments
Newest comments first. Replies stay oldest first.