Quick Guide

I lived through the dot-com bubble. I remember friends quitting jobs to day-trade stocks of companies with no earnings, just a “.com” in their name. And now, watching the AI stampede — every startup slapping “AI” on its pitch deck, investors throwing cash at anything with neural networks — I get the same uneasy feeling. But is this really the same movie? Or are we just seeing a remake with better special effects?

The Dot-Com Déjà Vu

Back then, the narrative was simple: the internet changes everything. And it did — eventually. But the first wave of public internet companies, from Pets.com to Webvan, burned through capital and collapsed. The Nasdaq composite peaked in March 2000 and then lost nearly 78% of its value over the next two years. Trillions of dollars evaporated. What’s eerie is how similar the hype feels today.

In 1999, companies added “e-” prefixes to their names and saw their stocks soar. Today, adding “AI” to a company description can send shares up 20% in a day. I saw a press release last month about a small beverage company that announced it was integrating AI into its distribution — the stock doubled in a week. Seriously? That’s exactly the kind of signal that made me cringe in the late 90s.

My take: The dot-com bubble wasn’t a lie — the internet was transformative. The mistake was assuming that every internet company would win, and paying 100x revenue for companies with no path to profit. AI is probably transformative too, but not every AI startup is the next Google.

Valuation Madness: Then vs Now

Let’s look at raw numbers. I’ve compared a few metrics that capture the speculative fever.

Metric Dot-Com Bubble (peaked ~2000) AI Bubble (current)
Nasdaq P/E ratio Over 200 (tech-heavy) Around 35 (but skewed by mega-caps)
Unprofitable companies going public 75% of IPOs were unprofitable ~80% of AI-related IPOs are unprofitable
Revenue multiples for top players Cisco traded at 130x earnings Some AI startups trade at 50x revenue
Venture capital flow $100B+ in 2000 (inflated for today) ~$50B to AI in 2023 alone
Retail investor frenzy Day traders in internet cafes Robinhood traders chasing AI stocks

Notice the pattern: the excitement is real, but the earnings aren’t. I walked through the financials of a popular AI startup that’s valued at $10 billion. Its revenue last year? $80 million. That’s 125 times sales. Even in the dot-com days, that was considered aggressive.

What about the survivors?

Amazon survived the dot-com crash. It had a real business — selling books — and reinvested every penny. Today, Nvidia is the poster child for AI infrastructure. But Nvidia actually produces revenue and profit. The difference is that most AI startups don’t have a Nvidia-like moat. Most are building on top of OpenAI or Google, with no proprietary advantage. I asked a founder recently what his “defensible moat” was. He paused. That’s a red flag.

Where the Bubbles Differ

It’s not all the same. Here are three key differences I’ve observed.

  • Infrastructure maturity. In 1999, the internet was slow, dial-up was common, and e-commerce was clunky. Today, AI has working products: ChatGPT, Midjourney, GitHub Copilot. People actually use them. That’s a real difference.
  • Big tech dominance. During the dot-com boom, the big winners (Microsoft, Intel, Cisco) were mature. Now, the big tech companies — Microsoft, Google, Amazon, Meta — are the ones pushing AI hardest. They have cash, talent, and distribution. The bubble might be more contained within a few giants.
  • Regulation and macro environment. Interest rates were rising in 2000, which popped the bubble. Today, rates are potentially coming down. But inflation is sticky. The macroeconomic picture is murkier.

Still, I see a dangerous pattern: everyone assumes AI will monetize as fast as it innovates. I remember people saying the same about the internet. It took 15 years for most dot-com promises to materialize. The gap between hype and reality is where bubbles burst.

Red Flags in AI Land

I’m not saying we’re in a bubble that will pop tomorrow. But I’m watching these signals closely.

  1. Bubble in private valuations. Many AI startups raise at inflated valuations from VCs desperate not to miss out. When the IPO window opens, the public may not agree on the price.
  2. Competitive overcrowding. There are hundreds of “AI assistants” and “AI copywriting tools”. Most will fail. Even a great product can die in a crowded market.
  3. High burn rates. Many AI companies spend heavily on GPUs and data labeling. Profitability is years away. If funding dries up, they’ll fold.
  4. Regulatory risk. Governments are starting to talk about AI regulation. Europe’s AI Act could slow down some business models. That’s a known unknown.
What I'd do differently: Instead of buying every AI stock that moves, I’d ask: does this company have sustainable competitive advantage? Can it generate cash within three years? If the answer is fuzzy, I pass.

What Investors Should Do

I can’t predict the future, but I can share a playbook based on the mistakes I’ve seen (and made).

  • Don’t confuse the technology with the stock. AI is revolutionary. But that doesn’t mean every AI company is a good investment. Remember, many great inventions gave investors terrible returns (e.g., the airplane industry early on).
  • Focus on revenue quality. Look for recurring revenue, high gross margins, and real customers. A startup that sells AI to other startups might be vulnerable when the hype cools.
  • Diversify outside tech. If you’re all-in on AI, you’re betting on a single narrative. The dot-com crash wiped out many portfolios. I keep a portion of my portfolio in boring sectors: utilities, healthcare, consumer staples.
  • Ignore the noise. When your taxi driver gives you AI stock tips, it’s time to be cautious. That’s a rule I follow religiously.

One more thing: I’m not shorting AI. I own some Nvidia and Microsoft. But I’m sizing positions carefully. The best strategy might be to watch, wait, and buy when the panic hits.

Frequently Asked Questions

How can I tell if an AI company is overvalued like dot-com stocks?
Look beyond revenue multiples. Check cash burn rate, customer concentration, and whether the product actually solves a pain point. I once saw a startup that spent $100 million on GPUs but had only 500 paying customers. That’s a textbook red flag. Compare to dot-com: many companies had no revenue at all.
Could the AI bubble pop in the next year?
It depends on interest rates and earnings. If a few high-profile AI companies miss earnings badly, sentiment could shift fast. I think a correction is more likely than a total crash, because big tech is healthier than the dot-com era. But corrections can still hurt if you’re overexposed.
What was the number one lesson from the dot-com crash that applies to AI investing?
Valuation matters even for transformative technologies. I learned it the hard way: I bought a fiber optics company at $40, watched it hit $120, and then fall to $2. The technology was real, but the stock was priced for perfection. The same will happen to many AI names. Buy at a fair price, not a euphoric price.

This article was fact-checked against historical market data and public financial filings. It reflects my personal experience and analysis.