Quick Navigation
I’ve seen this movie before. Actually, I lived through the dot-com crash. And watching the AI mania today gives me a strange sense of déjà vu. But it’s not a simple repeat — the AI bubble has its own flavor. Let me walk you through what I’ve observed, where the similarities are scary, and where this time might actually be different.
The short version: Both bubbles involve irrational exuberance around a transformative technology. But AI has real revenue, massive infrastructure spending, and a few dominant players. The dot-com bust wiped out companies with no earnings — today, many AI startups have similar fragility. History hasn’t repeated, but it has rhymed.
What Made the Dot-Com Bubble Different
Back in the late 90s, the internet was the shiny new toy. Every company that slapped a “.com” on its name saw its stock skyrocket — even if it had zero profit and questionable business models. I remember visiting a friend who worked at a startup that sold pet supplies online. The office had beanbags, free snacks, and a dog running around. They spent millions on TV ads during the Super Bowl. Revenue? Barely a trickle. The company, Pets.com, became the poster child of the bust.
Revenue vs. Hype
The dot-com bubble was fueled by pure speculation. Most companies had no clear path to profitability. They burned cash on marketing, built infrastructure they didn’t need, and believed “eyeballs” would eventually convert to dollars. When the music stopped, companies like Webvan and eToys collapsed. The Nasdaq fell nearly 80% from its peak. I lost a chunk of my savings — a painful lesson.
The Infrastructure Boom
Interestingly, the dot-com bust also gave birth to lasting giants. Amazon, Cisco, and Google survived because they had solid fundamentals. Cisco made the routers that powered the internet — real product, real customers. The lesson: the infrastructure layer often wins, while the application layer burns.
How the AI Bubble Compares (and Contrasts)
Fast forward to today. Artificial intelligence is the new internet. Every company claims to be “AI-first.” Valuations are sky-high. Nvidia, the chipmaker, briefly became the most valuable company in the world. But there are key differences that make me hesitate to call it a straight repeat.
| Dimension | Dot-Com Bubble | AI Bubble |
|---|---|---|
| Revenue generation | Most companies had zero revenue | Some AI firms have real revenue (e.g., OpenAI, Nvidia) |
| Capital intensity | Low: needed a website and a domain | Extremely high: GPUs, data centers, talent |
| Market concentration | Many small players | Dominance by Big Tech (Microsoft, Google, Amazon) |
| IPO frenzy | Hundreds of IPOs with no earnings | Fewer IPOs but massive private valuations |
| Hype cycle | Retail investors driven by FOMO | Institutional money + media hype |
What scares me is the valuation disconnect. Nvidia’s P/E ratio has been above 70 at times. That’s not crazy compared to some dot-com stocks, but it assumes years of perfect execution. Meanwhile, AI startups are raising billions at valuations that imply they’ll capture huge markets — but many have no moat. Anyone can build a chatbot using an API.
The Infrastructure Play Again
Just like the dot-com era, the infrastructure layer (chips, cloud, networking) is where the real money is flowing. Nvidia, AMD, and data center REITs are the modern equivalents of Cisco and Lucent. But even here, competition is heating up. I’ve seen small AI chip startups promise to beat Nvidia, but most will fail. The lesson: stick to proven leaders with customer diversification.
Key Lessons from the Dot-Com Bust That AI Investors Are Ignoring
I’ve been talking to friends in VC and founder circles. Many of them are too young to remember 2000. They think “this time is different” because AI has real use cases. I call BS. Here are three lessons they’re ignoring.
- Valuation matters, even for good businesses. Amazon’s stock fell 95% during the bust, even though it was a great company. If you bought at the top in 1999, you waited 10 years to break even. Many AI stocks could have similar drawdowns.
- Cash burn is a killer. During the dot-com era, companies like Webvan spent millions on warehouses before having demand. Today, AI startups spend millions on GPUs and salaries. If the funding environment changes (hint: it will), many will run out of cash.
- Diversification doesn’t mean buying every AI stock. In the late 90s, people bought a basket of internet stocks and thought they were diversified. They all crashed together. The same will happen in AI. Real diversification means owning assets outside the tech bubble — energy, healthcare, value stocks.
My personal rule: I limit AI exposure to no more than 20% of my portfolio. And I favor companies that have been profitable for multiple years, not just promise of future profits.
How to Navigate the AI Hype Without Getting Burned
If you’re like me — excited about AI but wary of a bubble — here’s a practical checklist I’ve developed after getting burned (and learning).
1. Focus on Free Cash Flow
Don’t get dazzled by revenue growth. Look at free cash flow (FCF) yield. Companies like Microsoft generate massive FCF and use AI to enhance their existing products. Pure-play AI startups often have negative FCF. I prefer the former.
2. Avoid the “Pick and Shovel” Fallacy
Everyone says “sell picks and shovels” (i.e., invest in infrastructure). That worked in the gold rush. But today, the picks and shovels are overpriced too. Nvidia’s stock already prices in years of 50% growth. If growth slows, the stock will get crushed. Instead, look for companies with pricing power and wide moats.
3. Use Dollar-Cost Averaging
Timing the top is impossible. I put a fixed amount into my chosen AI ETFs every month, regardless of price. This way, if the bubble bursts, I buy at lower prices. It’s boring but effective.
4. Watch for Insider Selling
When founders and executives start selling massive amounts of stock, it’s a red flag. I track insider transactions. If a CEO sells 50% of their holdings, I sell too.
Real-World Examples: From Pets.com to OpenAI
Let’s compare two iconic companies from each bubble.
Pets.com (Dot-Com)
- Founded: 1998
- Business: Online pet supplies
- Revenue at IPO: $0 (launched later)
- Peak market cap: ~$300 million
- Result: Went bankrupt in 2000, 9 months after IPO
OpenAI (AI)
- Founded: 2015
- Business: AI research and products (ChatGPT, API)
- Revenue: ~$3.4 billion in 2024 (estimated)
- Valuation: $150 billion+ (private)
- Profit: Negative (heavy spending on GPUs and talent)
The differences are striking: OpenAI has real revenue, but its valuation implies it will capture a huge part of the enterprise software market. It also faces intense competition from Google, Meta, and open-source models. I’m not saying OpenAI will fail — but at this valuation, the risk is high. I’d rather own Microsoft, which owns 49% of OpenAI’s profits and has a diversified business.
Frequently Asked Questions About AI vs Dot-Com Bubbles
Fact-checked: This article reflects personal observations and analysis. Always do your own research before investing.
