The key claim: "we've achieved AGI." Sounds like a perfect headline (which is why I put it in the headline..), but in context it's more of a marker phrase. Huang means that current systems already solve tasks that used to be considered criteria for strong AI — and the definition shifts every time another milestone is passed.
Beyond the headline: why a $500K developer should spend $250K on AI tokens (and Huang is serious), how scaling laws hit walls not in algorithms but in memory, electricity, and supply chains, why NVIDIA is thinking about data centers in space, and why Doom is the most influential video game in history.
Link to the interview