GTC 2026 Preview: Agents and Physical AI, Not Chat Elo
NVIDIA scheduled GTC 2026 for March 16–19 in San Jose, with agentic systems, physical AI and AI-factory infrastructure among its announced themes. This March 10 preview distinguishes the advance program from later product launches.
NVIDIA’s advance program for GTC 2026, March 16–19 in San Jose, centers on the full AI stack: infrastructure, open models, agentic systems and physical AI. For a March 10 preview, those are announced themes, not completed launches or measured deployment outcomes.
In its March 3 announcement, NVIDIA scheduled Jensen Huang’s keynote for March 16 at 11 a.m. Pacific. It described sessions on AI factories, large-scale inference, robotics, digital twins, scientific computing and enterprise deployments, alongside developer training and infrastructure workshops.
The program places applications and the infrastructure behind them in the same conference. That makes it relevant to readers following production agents and robotics as well as the GPU roadmap, without establishing which products will ship or how customers will adopt them.
Claims vs checks#
Dates, venue, keynote time and program themes come from NVIDIA’s pre-event announcement. The company’s attendance expectations are projections; they do not substantiate the earlier assertion of record actual attendance.
Limits#
A conference agenda establishes planned topics, not product availability or factory deployment rates. Later launch coverage should be dated separately from this preview.
Correction (October 14, 2026): Restored pre-event chronology and removed retrospective NemoClaw and record-attendance claims from this March 10 preview.