Nvidia GTC 2024: What to watch from the AI giant’s technology conference

Nvidia is arguably the most important company in the tech sector these days, with each swing in its stock driving the Nasdaq Index higher and lower and a market cap that has more than tripled in the past year. So when it speaks, a lot of people listen. This week, Nvidia should have a lot to say. The company’s GTC conference is one of the hottest tech gatherings of the year. Historically, Nvidia has used it as a launch pad for some of its biggest products, including the Hopper graphics architecture and the H100 graphics processing unit, which has become the de facto choice for companies building artificial intelligence models. This is the first time Nvidia has held an in-person GTC event since 2019—and a lot has changed in that time. This year, it’s expected to attract 16,000 people in person with several hundred thousand watching online. The keynote address by CEO Jensen Huang will take place Monday at 4 p.m. ET, with sessions from OpenAI, Microsoft, Amazon, and Google over the course of the next three days. If you’d like to watch live, plenty of sites will be carrying it live, but you can view it directly on Nvidia’s website. While surprises at GTC are certain, there are a few things you can expect to hear more about this week at the so-called AI Woodstock. New tech Nvidia is widely expected to debut the successors to both Hopper and the H100. The new architecture, code-named Blackwell and its accompanying GPU, called the B100 during its development, should offer better performance for large language models. Shipments of the new tech aren’t expected to start in the near future, but this will be a chance for Nvidia to show partners how it plans to drive the next wave of AI. Some also expect that the company could debut a new AI chip designed for China that puts it in compliance with U.S. export rules. It’s telling that Nvidia itself is calling Huang’s keynote a “transformative moment in AI” in previews. Sovereign AI As one of the leading proponents of sovereign AI, which it defines as “a nation’s capabilities to produce artificial intelligence using its own infrastructure, data, workforce and business networks,” Nvidia is likely to at least touch on the subject this week. At its heart, the concept revolves around containing large language models within the borders of a country or a company for safety purposes. It’s an especially important topic given ongoing concerns about national security risks with data sharing, such as TikTok’s ties to the Chinese government. Updates from the giants of AI More than 1,000 organizations will take part in more than 900 sessions at GTC—and the exhibition floor will showcase more than 300 exhibits. Among the featured speakers are Brad Lightcap, chief operating officer at OpenAI; Sébastien Bubeck, vice president of Microsoft GenAI; and Vincent Vanhoucke, distinguished scientist and senior director of robotics, Google DeepMind. That could shed some light on everything from regulation of AI to competition in the AI space, which could act as a guide for investors looking to find an on-ramp to the AI explosion that’s not prohibitively expensive. Professional certification With demand for AI workers at all-time highs, Nvidia will offer a new professional certification in generative AI that could help fill some of the gaps. Two certifications will be offered, one in large language models and one in multimodal workflow skills. In-person attendees will have the opportunity to access training to prepare for the certification exam at the conference.