Yesterday morning, the most important man in artificial intelligence (AI) was out of a job.

That, my friends, was the definition of a black swan.

Nvidia's (NVDA) founder and CEO Jensen Huang landed himself on the naughty list and stepped down after a string of scandals.

Jensen was arguably the most important and central person to the whole AI trade. He’s been running Nvidia almost as long as I've been alive! So it’s no wonder AI stocks sold off hard.

This came on top of bad news from the midterms, where voters turned data centers into a ballot-box issue.

All in all, Nvidia alone has lost about $1.8 trillion of stock market value in five weeks. That's more than the entire value of Meta Platforms (META).

Everyone is asking: Is the AI trade over? More on that at the end of today’s essay.

• First, the Jensen saga.

It’s bad for Nvidia. I think it’s a huge mistake that he stepped down. But I don’t believe it changes the long-term trajectory of AI.

Over the three decades since he founded Nvidia, Jensen built relationships with the biggest players in tech. And he became a master at convincing people to buy into his vision.

That combination of technical know-how, connections, and salesmanship is nearly impossible to replace.

Roughly 60 people reported directly to him. Every big decision went through his desk. That's how Nvidia could turn on a dime. It's also why replacing him is such a tall order.

Jensen was more than Nvidia's CEO. He helped shape the entire AI supply chain as we know it. Without him, the world would almost certainly have far fewer AI chips today.

The good news is that Jensen built a machine that can keep running without him.

Nvidia has a huge head start over its competitors. It sells entire AI systems, from the chips themselves to the networking that connects thousands of them. And its software has become the industry standard, with millions of developers already using it.

Its chips are still the most energy-efficient on the market. That's a big advantage. Electricity is one of the biggest bottlenecks holding back the AI boom.

These advantages took years to build. Rivals won't erase them overnight.

• The backlash against data centers is real.

People would rather live next to a nuclear power plant than a data center. Nuclear is one of the cleanest, safest energy sources, so I'm glad people are coming around to it.

But fearing a building full of computers more than a reactor is still nuts!

Politics decides where data centers get built, not whether they get built. New York banned fracking in 2014. Next door, Pennsylvania kept drilling and became America's second-biggest gas producer. Nuclear had one regulator that could stop it. A data center has more than 3,000 counties to choose from. Plus the Gulf. And soon, space.

I'm a believer in markets and incentives. And money works!

Towns want data centers to pay their own way on power and water. Fine. OpenAI has already pledged to pay for its own electricity and grid upgrades at the campuses it owns.

For these companies, a bigger power bill is a rounding error. By my math, a month without power costs a 1-gigawatt AI campus $450 million to $800 million in lost output. Paying an extra $50 per megawatt-hour costs about $31 million a month. They'll pay the “bribe.”

I ultimately think whatever bottlenecks emerge from this will give us a longer, steadier boom instead of a 2000-style bust.

• Folks are getting overly emotional and mixing politics with investing.

The clearest sign that emotion is running the show was on election night, when Advanced Micro Devices (AMD) reported record quarterly sales (up 45% from a year ago). Its data center sales nearly doubled. It guided to another big jump next quarter.

Two days later, its stock fell 17%. Over the next four trading days, it shot back up 37% to well above where it started.

Nothing about AMD's business changed that week. Only “Mr. Market’s” mood did.

When markets get emotional, I do the same thing every time. I tune out the headlines and look at the facts. So that's what I did this week.

The facts all point the same way.

On Wednesday, November 18, I’m hosting a live Emergency Briefing to talk more about the Nvidia fallout, where we are in the AI boom, and what’s next for investors. It’ll be on Zoom at 11 AM ET. And it’s free to sign up. Go here to register. I hope to see you there.

Stephen McBride

Chief Analyst, RiskHedge

Reader Mailbag

If you missed my recent Jolt on the future of AI, I shared some incredible innovations I saw while meeting with more than 40 companies across the US. By far my most important takeaway is that Phase 2 of the AI trade is over, and Phase 3 has now begun.

From Kevin, who shared how he’s using AI at his work:

We’re using Claude Code to accelerate our creation of new products and features. But we’re also putting AI in front of our data products for our customers, which greatly amplifies their ability to explore vast, complex scientific & IP data to discover what they need. And that, in turn, accelerates their innovations in chemistry, drug discover, materials, etc.  

It’s maybe more basic/foundational than the cutting-edge stuff you typically look at, but I think it’s the kind of thing that’s going to drive a lot of economic value creation.

—Kevin

Kevin, I think you’re onto something big here. Most investors are still focused on the companies building AI. But as businesses start putting it to work in the ways your company has, I think we’ll see a whole new set of winners emerge.

I’ll touch on this more in my Emergency Briefing next Wednesday. Hope you can join!