The big argument you hear from Nvidia bulls and bulls on the AI trade in general is that 'this isn't the dot-com bubble'. Their proof is compelling on the surface and it's that Nvidia is profitable. They're printing money in a way that's rarely been seen before.
It's true but this is a classic case of fighting the last war. Everyone is using the 2000s mind map where unprofitable, pets.com-style companies were the problem. But just because it's profitable doesn't mean it's not a bubble.
The bubble isn't in the current or forward P/E. It's in the market's, baked-in assumptions about the durability of these earnings. You aren't just paying for today; you're paying for two decades of perfect execution and effectively zero competition.
The numbers are pricing in a fantasy
Look at the projections. The consensus is pricing in 2028 revenue of around $360 billion. That’s 2.7 times higher than 2025's numbers. And in a sign of how parabolic this all is, revenue was just $11 billion in 2020. I actually believe in those numbers and expect the scale of AI spending to stay high for awhile, if not rise even more.
The first problem is that we are in the mother of all investment booms, and the price assumes this spending spree is permanent.
But the revenue is only half the story. The real fantasy is in the margins.
The current price assumes continuing gross profit margins of 75%. Not just for next quarter, but locked in for the next 20 years.
This implies that no one—not AMD, not Intel, not a wave of startups, not Amazon or Google's in-house chips—will ever catch up and compete on price.
In what world does a 75% margin in a high-tech hardware sector not attract rabid, margin-destroying competition? It's the biggest 'come and eat my lunch' sign in economic history. It's $5 trillion in market cap that's up for grabs.
Competition is coming, and it will be brutal
This valuation conveniently ignores two massive risks:
The Fabs: Nvidia doesn't even make its own chips. The market is assuming that TSMC and other fabs will happily let Nvidia keep all the profit and won't wreck their margins by raising their own prices. That's a bold assumption when you're the only game in town for high-end fab.
China (and likely others): You have the full, undivided weight of Beijing. Building its own high-end chips is a matter of national security for China. They are pouring everything into this. If—or when—they produce chips that are 'good enough,' do you think they'll be selling them at 75% margins?
That just doesn't happen in China. Try 5% margins. Try 'whatever it takes to gain market share and break the monopoly.' That single competitor changes the entire pricing model for the industry and pulls out the rug.
So, stop looking at the 20x forward P/E. It's a distraction.
A vote for NVDA at these prices is a vote that this investment boom never ends, that 75% margins are permanent, that competition never arrives, and that China's national ambitions will fail.
That's a precarious bet on perfection.
