NVIDIA Makes GTC 2026 Highlights Available On Demand; 2027 Event Set for San Jose

NVIDIA said highlights and sessions from GTC 2026 are available to watch on demand. GTC brings together developers, researchers and business leaders at conferences and regional events around the world. The next event is scheduled for March 15-18, 2027, in San Jose.

NVIDIA has included physical AI, AI factories and inference among GTC’s themes. This indicates the conference’s agenda, not the performance or deployment record of any individual robot. To assess technology trends, readers need to distinguish the specifications and implementation conditions presented in each on-demand session from the themes listed by the conference.

One notable aspect of GTC’s announcement is that it lists physical AI alongside AI factories, agentic AI and inference, rather than focusing solely on physical AI. This is not information about an individual robot announcement or performance evaluation; it shows the breadth of technology areas NVIDIA will cover at the conference. The announcement also mentions hands-on training, but its page does not specify what it will involve or the conditions for participating. Therefore, the page cannot be read as evidence of the implementation status or real-world performance of individual products. In our coverage, [Hugging Face Releases LeRobot v0.7.0 Robotics Learning Platform, Adding Interactive Control and Large-Scale Data Training](/en/news/hugging-face-releases-lerobot-v0-7-0-robotics-training-platform-adding-4e8d3f) covered robotics learning platforms, including world action models from NVIDIA and others. We also covered the company’s explanation of its AI computing infrastructure in NVIDIA Explains Full-Stack Design to Address AI Memory Constraints. The latest GTC announcement does not add technical details to those reports, but it connects editorially by listing robotics learning, computing infrastructure and inference among the conference’s areas of focus — different ways to examine AI operating in the physical world, from software and computing to physical machines. The announcement does not say which sessions will link these themes. To understand the industry landscape, it is important to distinguish conference themes from evidence of technologies or products actually being offered. For example, listing “physical AI” as a theme does not establish what tasks robots can perform, what computing resources they use, how training data is handled or whether products are being mass-produced. Likewise, listing AI factories and inference does not establish how the two are connected, their processing capacity or their terms of use. The page shows only that NVIDIA has made these topics part of GTC’s agenda. What remains to be seen is the specific content of each on-demand session, whether products, specifications or performance figures are presented, and how far discussions of physical AI address real-world deployment and operating conditions. The page does not list the number of sessions, the content of individual presentations or results from real-world testing. It gives the dates for the 2027 event, but its program and participation requirements cannot be confirmed from this announcement alone. Technology trends should be assessed separately using individual materials as they become available.

NVIDIA’s announcement lists physical AI, AI factories, agentic AI and inference among the session topics. It also says the event will offer hands-on training and opportunities to connect with experts and other attendees.