Just stepped out of Jensen Huang’s keynote and my head’s still spinning. Three things stuck:
- New Rubin chips are twice as fast, no extra power, no new AC units—your server room stays quiet.
- Cars that show their “work” when they decide to swerve—Mercedes is first, U.S. roads this year.
- Robots that learn like ChatGPT learns text—Atlas the humanoid already trained; factories next.
Bottom line: AI just got cheaper, cooler, and way more physical.
Two pocket-sized R2R2-style bots rolled on stage to say hi. They’re part of NVIDIA’s new open-model push—basically starter brains for any company that wants a lovable, roaming assistant without starting from scratch.