Just caught Jensen Huang's latest comments on where AI is heading, and honestly, it reframes a lot about Nvidia's next chapter.

So here's the thing - everyone knows Nvidia basically owns the GPU market right now. They've built this insane moat through constant innovation, annual chip updates, and an entire ecosystem around their hardware. Training models? That's Nvidia's bread and butter. But the story isn't just about chips anymore.

What caught my attention is what Huang flagged about six months back - there's been this inflection point in AI that most people are only now catching onto. We're moving past the training phase into something way more interesting: agentic AI. These systems aren't just sitting there processing data anymore. They're actively solving real problems, and that's a whole different beast.

Huang put it simply - the agents are genuinely smart and they're tackling actual use cases. That means continuous GPU demand as these systems run inference and operate at scale. But here's where it gets bigger: he's also talking about physical AI, bringing these intelligent agents into robotics and real-world applications. He literally called it a "giant opportunity."

What this tells me is that the revenue runway for Nvidia extends way beyond what people might think. Sure, the training boom was massive, but we're probably in the early innings of what comes next. The company's forecasting $78 billion in revenue with 77% growth year-over-year, and if this agentic and physical AI thesis plays out, that could just be the beginning.

Now, will the stock moon tomorrow? Probably not - macro conditions and market sentiment matter. But if you're thinking long-term about where AI actually goes as a technology, Nvidia's positioning looks pretty solid. The infrastructure play isn't over; it's evolving.