"Is there an AI bubble?" is the question of the moment. The honest answer isn't a headline yes or no: it depends on which company you look at and, above all, on the numbers. Let's judge it the right way, with valuation instead of noise.
What a bubble is —and isn't
A bubble isn't something rising a lot. It's the price detaching from value: when a stock rises driven by enthusiasm and fear of missing out, more than by the profits it generates. NVIDIA being worth far more than two years ago isn't, on its own, a bubble: its profits have multiplied too. The right question isn't "has it gone up?", but "do the fundamentals back it?".
The two sides
There's a solid bullish case: AI is generating real revenue and profit at the companies selling the infrastructure, and big tech's spending on data centers keeps growing. And there's a cautious case: part of that spending is funded on the promise of future revenue that hasn't arrived yet, and some valuations assume growth no company can sustain for a decade. Both can be true at once.
How to judge it with numbers, not headlines
To tell whether a specific stock is in bubble territory, look at three things:
- The P/E versus its growth: a P/E of 50 can be reasonable if the company grows 40% a year, and absurd if it grows 8%. The number only makes sense next to growth.
- Are earnings catching up to the price?: in a healthy company, profits grow and "justify" the price over time. If the price runs and the profits don't show up, that gap is the bubble.
- Cash flow: real cash is harder to fake than the story. A company that generates cash withstands a pop better.
They're not all equally expensive
Here's the nuance almost nobody makes. Under the "AI" label live very different valuations: some tech names trade at extreme multiples, while certain equipment suppliers, server makers or even the utilities that power data centers trade at far more down-to-earth prices. Talking about "the AI bubble" as a block is a mistake: each company has its own price against its own growth.
The honest conclusion
Nobody knows whether "we're in a bubble" or when it would pop; anyone who tells you for sure is selling smoke. What you can do is not overpay: judge each company by its valuation and diversify. Type any AI stock's ticker into the analyzer and you'll see whether its P/E is sane for its growth. And to go deeper, read how to tell if a stock is cheap or expensive or the map of the companies behind the AI boom.