Everyone watched the Dwarkesh interview with Jensen Huang looking for drama. The China fight. The Anthropic regret. The CUDA moat.

But if you watch it a second time as a map, Jensen basically showed you where the next decade of AI value gets created.

He didn't mean to. He was just answering questions. But when a guy who runs the most important company in AI talks for an hour about his regrets, his bottlenecks, and where he's personally putting money.. you're getting a preview.

Most people missed it because they were busy clipping the best burns.

Remember when the news was about what happened, not how to feel about it? 1440's Daily Digest is bringing that back. Every morning, they sift through 100+ sources to deliver a concise, unbiased briefing — no pundits, no paywalls, no politics. Just the facts, all in five minutes. For free.

Read the newsletter trusted by 4.5 million fact-seekers.

This was the moment that changed how I read the entire interview.

Jensen admitted Nvidia missed the chance to invest in Anthropic early. And the reason he regrets it isn't what you'd think.

It's not that he wishes he owned the equity.

It's that by failing to invest, he forced Anthropic into Google and AWS's arms. Who put in billions. And in return, Anthropic used their compute. TPUs. Trainium. Not Nvidia.

Jensen said he didn't fully understand at the time that a lab like Anthropic couldn't be funded through normal VCs. What they were trying to build was just too capital-intensive. By the time he figured that out, the deals were done.

"It's still okay to have regrets," he said.

Now think about what that actually means.

The guy selling the picks and shovels.. the one collecting 70% margins on every AI chip shipped.. realized that in this new era, selling chips isn't enough. You have to bankroll the buyer too.

The hardware layer and the foundation model layer are vertically integrating. And the price of staying in the game is no longer just great hardware. It's great hardware plus billions in equity checks.

Jensen figured this out late. He's now reportedly backing OpenAI and Anthropic in funding rounds valuing them in the tens of billions. Not because he loves being a VC. Because that's what it now costs to make sure the world's most important AI labs run on your stack.

And if you're reading this as a signal, here it is.

The race isn't just about who has the best chips anymore. It's about who has the balance sheet to keep the best customers locked in. That's a very different game. One where only two or three companies in the world can even play.

That reframes a lot.

The second signal was about energy.

Jensen said the real bottleneck for AI isn't chips. It's power.

He said he can solve any chip supply problem in 2-3 years. CoWoS packaging was a bottleneck. They doubled capacity multiple times. Done. Memory supply? Same playbook. EUV? Convince TSMC and ASML follows.

But energy takes decades. You can't scale AI factories, bring chip manufacturing home, and deploy robots.. without massive new power generation.

And then he said something about China that most people ignored.

China's real edge isn't better chips. It's abundant energy. They have data centers sitting fully powered and completely empty. Ghost data centers. Even stuck on 7nm, they just use 4x or 10x more chips to compensate.

Because energy is essentially free for them.

"If your amount of watts is completely abundant, what do you care about performance per watt for?"

Nvidia's entire engineering obsession.. Blackwell being 50x more efficient than Hopper.. exists because the US doesn't have abundant energy. Every watt matters.

But it also means the next big bet isn't chips. It's everything that generates, stores, and moves electrons.

Nuclear. Grid buildout. Power generation. Transmission infrastructure. The unglamorous stuff that takes 10-20 years to build.

That's where real money flows next.

And Jensen didn't stop there.

He quietly mentioned that CoreWeave, Nebius, Nscale wouldn't exist without Nvidia's backing. These neoclouds aren't random bets. They're structurally protected by the most important company in the industry.

He also said something I haven't seen anyone pull out yet.

He predicted the number of tool-users is going to skyrocket. Software companies like Synopsys and Cadence are going to explode in usage because every AI agent needs tools to work with. Excel, PowerPoint, every layout tool, every design rule checker.

"The number of instances of all these tools is going to skyrocket."

Think about what this really means. The doomer narrative is that AI kills software companies. Jensen is saying the opposite. Agents don't replace software. They become new users of it. Millions of agents, each needing licenses, each querying tools in parallel.

The software business doesn't shrink. It multiplies.

That's a direct prediction from the guy who sees demand signals before anyone else does.

And there's one more layer he mentioned..

AI cybersecurity. He described a future where every AI agent is surrounded by thousands of other AI agents keeping it safe.

"The idea that you're going to have an AI agent running around with nobody watching after it is kind of insane."

The AI safety ecosystem becomes its own industry. Startups built to monitor, audit, and protect agents. A whole layer nobody is pricing in yet.

Stack it all up.

Models at the top, locked in through equity, not just contracts. Energy at the bottom, becoming the actual constraint. Tool makers in the middle, quietly multiplying their seat counts through agents. AI security wrapping around the entire thing. Neoclouds compounding underneath.

That's the actual tech stack Jensen is building for. And if you know where to look, he just told you which layers are about to compound the hardest.

Most people watched this podcast looking for soundbites. And they got plenty.

But Jensen wasn't just giving interview answers. He was quietly telling you where the next decade of AI value creation is headed.

The labs. The energy. The tools. The safety layer. The neoclouds he's personally propping up.

He showed you his regrets, his investments, his bottlenecks, and his predictions. All in one sitting. If you line them up in order, you get a pretty clear picture of what matters next.

And here's what I keep coming back to.

Jensen is probably the best-positioned person alive to see what's coming. He sees every AI company's demand forecast. He knows what they're training. He knows what they need. He's in rooms with CEOs upstream and downstream of the entire stack.

When a guy like that tells you his biggest regret is not bankrolling the labs.. and that the real bottleneck is energy.. and that tool companies are about to explode.. you should probably sit with that.

Not as hype. Not as stock tips.

As signals from the guy who has the clearest view in the industry.

What do you think.. did you catch something in this podcast that everyone else missed?

If you made it this far, you're not a casual reader. You actually think about this stuff.

So here's my ask. If this article made you think, even a little, share it with one person. Just one. Someone who's in the AI space. Someone who reads. Someone who would actually sit with these ideas instead of scrolling past them.

That's how this newsletter grows. Not through ads or algorithms. Through you sending it to someone and saying "read this."