Nvidia’s GPU Technology Conference officially opened today in San Jose, California, with Jensen Huang delivering the keynote address at 11am Pacific from SAP Center, home of the San Jose Sharks, in front of a crowd drawn from 190 countries.
More than 30,000 attendees registered for this year’s event — developers, researchers, enterprise buyers, and AI-native founders — making it comfortably the largest gathering in GTC’s history and arguably the most closely watched Tech conference of 2026.
“GTC is the epicenter of the AI industrial era,” Huang said ahead of the event. “AI is no longer a single breakthrough or application — it is essential infrastructure. Every company will use it. Every nation will build it.”
The pre-show, which began at 8am Pacific before Huang took the stage, featured Perplexity CEO Aravind Srinivas, LangChain CEO Harrison Chase, Mistral AI CEO Arthur Mensch, and several other prominent figures from the AI ecosystem.
Nvidia enters GTC 2026 as the world’s most valuable company, having accumulated that position on the back of GPU demand from data centre operators building the infrastructure to train and run large language models at commercial scale.
In an interview earlier this year with South Korea’s Korea Economic Daily, Huang was quoted as saying Nvidia would unveil “a chip that will surprise the world” at the conference — a comment that generated weeks of speculation about what specific hardware or platform the remark referred to.
That chip comment has to be understood in context: Huang also acknowledged that “pushing performance further is getting harder as technologies approach physical limits,” which means any surprise announcement is more likely to be a system-level breakthrough than raw transistor density alone.
A pre-conference blog post by Huang titled “AI is a 5 Layer Cake” outlined his thesis that AI infrastructure depends on five simultaneous layers — energy, chips, networks, models, and applications — each of which needs to scale together or the whole system bottlenecks.
Nvidia has reportedly been investing up to $26 billion in open-source AI model development, and is expected to announce NemoClaw, an open-source enterprise AI agent platform, designed not to compete with frontier labs but to keep developers building inside Nvidia’s software ecosystem.
The conference follows what has been a volatile few weeks for Nvidia’s stock, which fell alongside the broader tech sector during the Iran-driven market sell-off but remains broadly supported by analyst price targets well above current levels, given continuing demand visibility from hyperscaler customers.
Nvidia will hold a financial analyst Q&A session tomorrow at 9am Pacific, giving institutional investors a more structured opportunity to press management on guidance, export restrictions to China, and the competitive landscape for GPU alternatives from AMD and Intel.
Whatever Huang announces from the SAP Center stage today is likely to shape the narrative around AI infrastructure investment for the next several months, at a moment when the broader market is desperately searching for a positive story that isn’t tied to the price of oil.