vidia GTC 2026, the first marquee AI event of 2026 has arrived. And with the "Moore's Law of AI" doubling every 4 months right now, we should expect plenty of news and announcements this week.
The Deep View will be on-the-ground at Nvidia GTC in San Jose, covering the most important developments in real time. That includes Nat Rubio-Licht, Faris Kojok, and yours truly.
These are the big trends we're tracking:
- Nvidia's chip strategy for inference: Nvidia has owned the lion's share of the AI market because of its technology advantage in GPUs. However, the market has been dominated by model training so far, and it's about to shift to inference and the compute needed to run AI day-to-day as the number of people and organizations running AI grows dramatically. When it comes to inference, Nvidia has less of an advantage, and we're seeing companies like Cerebras swoop in and take market share because they can run inference faster and less expensively. Nvidia made its $20B Groq deal to tackle inference. There are reports of a big inference announcement coming at GTC.
- Robots and physical AI: Nvidia loves robotics. Jensen Huang is a wonderful storyteller and robots are a tangible, physical manifestation of the current advances in AI. I'm sure Nvidia will trot robots on stage during the main keynote on Monday, but will we learn more about practical advances of robots of various shapes and sizes, consumer and enterprise? Robots, especially humanoid robots, are advancing much faster in China right now. Can Nvidia and its partners offer a counterpoint? Also, keep an eye on announcements for autonomous vehicles, another form of physical AI. The economics are getting a lot better, opening up new possibilities. Nat will be tracking developments across these topics throughout the week.
- Open models: Nvidia released its latest 120B-parameter model, Nemotron 3 Super, ahead of GTC and is also promising that Nemotron 4 Ultra, with four times as many paramters is coming soon. Nvidia is quickly becoming a leader in this space with models that are more open and are outperforming competitors. It's worth following anything they announce about new models of any type, as it's likely to have downstream effects on making AI more accessible and customizable for enterprises and lowering inference costs.
Our Deeper View
Nvidia GTC will be a harbinger of what to expect across the AI space during the rest of 2026. Expect to hear a lot about AI agents, world models, robots, self-driving cars, and AI inference. Making the unit costs work for AI inference is one of the critical factors facing the industry, especially enterprises deploying production-ready AI projects that demand strong ROI and snappy performance. As we discussed in a recent episode of The Deep View: Conversations podcast, AI inference is shaping up to become one of the largest markets in the history of the world. And it needs to get much more efficient and much less expensive to fully realize that potential. You can follow my Nvidia GTC updates in real time on X/Twitter at x.com/jasonhiner.