Lately, it seems like attaching the words "artificial intelligence" to almost any company is enough to send its stock price soaring. That pattern has understandably made investors nervous — the current climate carries some uncomfortable echoes of the dot-com bubble of the late 1990s, when speculative enthusiasm for anything internet-related eventually gave way to a brutal crash. The question worth asking isn't whether the comparison is tempting — it clearly is — but whether it actually holds up.
The Catalyst: A $100 Billion Bet on OpenAI
Much of the recent wave of "bubble talk" traces back to a single announcement: on September 22, Nvidia revealed plans to supply at least 10 gigawatts of its systems — representing millions of chips — to support OpenAI's infrastructure buildout, in a deal reportedly worth around $100 billion. Commitments of that scale, between two of the most closely watched names in tech, are exactly the kind of headline that can make any market participant pause and ask whether valuations have gotten ahead of themselves.
What's Actually Different This Time
Unlike many dot-com era companies, which often went public with little revenue and unproven business models, today's largest AI-related businesses are, in many cases, generating substantial and growing revenue, with real customers paying for real products. The infrastructure spending — from data centers to chips to power capacity — is also being funded largely by companies with existing, profitable core businesses, rather than by pre-revenue startups burning through IPO proceeds.
That doesn't mean today's market is free of speculative excess. Valuations for some AI-adjacent names have run well ahead of current earnings, and any slowdown in the pace of AI infrastructure spending could hit these stocks hard. But the underlying demand driving deals like Nvidia's OpenAI agreement appears to be tied to real, measurable compute needs — not simply the promise of future eyeballs, as was often the case in 1999.
The Takeaway for Student Investors
Bubbles are usually easiest to identify in hindsight, which is exactly what makes them dangerous in real time. Rather than trying to predict the exact moment sentiment turns, it's more useful to focus on the fundamentals: is the company generating real revenue, is the spending tied to measurable demand, and how much of the current price already assumes years of flawless execution. Those are the questions that mattered in 2000, and they're the questions that matter now.