There's a curious phenomenon in financial markets that I call "Jensen Huang Syndrome": you deliver the best quarter in your company's history, crush every estimate, and the market still slaps you across the face. That's exactly what happened to Nvidia last month.

Now, check out the irony.

Hock Tan, Broadcom's CEO — a guy who looks like he walked straight out of a Michael Mann movie, ice-cold, calculating, zero fluff — got on the earnings call and did what Jensen couldn't: he convinced the market.

The stock climbed 4% on Thursday. And it wasn't because of a pretty PowerPoint.

What the guy said that made Wall Street drool

Tan told analysts he expects AI chip revenue in 2027 to be "significantly above $100 billion." Read that again. Significantly. Above. One hundred billion dollars.

For those who don't grasp what that means: all of Broadcom brought in about $51 billion in the last fiscal year. The guy is basically saying that the AI division alone will be bigger than the entire company is today. And this isn't ten years from now — it's 2027.

JPMorgan pulled out the calculator and estimated the company could hit between $12 to $15 billion in revenue per gigawatt of capacity. They raised their estimates "conservatively" to $120 billion or more in AI revenue.

Conservatively. A hundred and twenty billion. Sure.

Why Tan convinced and Jensen didn't

Here's where the most interesting lesson of this whole story lives, and no suit-wearing analyst is going to tell you this.

Delivering numbers isn't enough. You need to kill the doubt sitting in the market's head.

The doubt with Nvidia was: "How long can this last?" Jensen didn't put that bullet down. Tan killed three at once:

First: he said he already has high-bandwidth memory and cutting-edge wafers locked in through 2028. At a time when everyone's fighting over HBM like dogs over a bone, the guy already secured the supply. That's management, damn it.

Second: he stated that AI business margins will be consistent with the rest of Broadcom's semiconductor business. In other words, none of that "we grew revenue but margins went to hell" nonsense. The yields are there, costs are under control.

Third — and this is the most elegant one: he used Nvidia itself as a shield. Tan argued that hyperscalers building their own custom chips can't afford to have "a chip that's just good enough." They need the best, because they're competing against each other and against Nvidia, "which isn't letting its guard down one bit."

It's like chess: he turned his biggest competitor into a reason for the customer to keep buying from him. Brilliant.

The ripple effect — and who got burned

A strong quarter from a giant always splashes around. Credo jumped 10% and Amphenol 4%, both copper connectivity manufacturers — which is what Broadcom uses to connect AI servers. The market's bet is that copper is winning the war against optical technology in this segment.

On the other side of the ring, Lumentum and Coherent — betting on optical tech — each dropped more than 4%. The market shows no mercy: it picked a side and punished the other on the same day.

What this means for the investor who isn't a sucker

Look, I'm not a fan of buying a stock after a 4% pop on earnings day. People who do that are usually the last minnow swimming into the school of fish.

But what's worth paying attention to here is Hock Tan's mental framework. The guy isn't a showman. He's not posting chip selfies on LinkedIn. He makes surgical acquisitions (remember VMware?), locks down supply chains years ahead, and talks to analysts like an adult talks to an adult.

In a world of tech CEOs who act like social media influencers, Tan is a welcome anomaly. He's the guy with real skin in the game.

Goldman Sachs said Broadcom delivers "the lowest inference cost for its hyperscaler clients" and is cutting costs at the same pace as leader Nvidia. If that holds up, $100 billion might actually be conservative.

So tell me: are you going to keep thinking AI is a bubble while these guys are locking up memory capacity through 2028, or are you at least going to consider that maybe — just maybe — this time the fundamentals actually support the narrative?