# Is AI a Bubble? What the Dot-Com Crash Teaches Investors

AI is real, but parts of the market price in perfection. Three bubble warning signs, what a crash would look like, and five ways to protect your money.

By Nikhil Varghese, CA · 2 December 2025 · https://finniki.com/blog/ai-boom-or-bubble

**Quick answer:** AI is both a real revolution and, in parts, a bubble. The technology already works, in hospitals, software and businesses, but some valuations assume a perfect, fast future: companies with no product or revenue valued in billions, share prices racing ahead of real adoption, and "AI-powered" stuck on everything. The dot-com era is the guide: the internet survived and thrived, but the Nasdaq fell **78%** and the pretenders vanished. Protect yourself by keeping AI to a limited share of your portfolio, favouring companies with real revenue, rebalancing, holding some cash, and building skills that AI enhances.

*Updated October 2026. Since the episode, AI shares have kept swinging: Nvidia became the first $5 trillion company in October 2025, then fell back sharply within weeks, a reminder of how quickly sentiment moves.*

[Watch the episode: AI bubble, boom or crash waiting to happen?](https://youtu.be/nHQ_h3ifXuM)

## Is AI the next big bubble?

Parts of it look like one. The International Monetary Fund and others have warned that AI-driven valuations could correct sharply, as the dot-com boom did. A handful of giant technology companies now make up a large share of the US market, around 40% of the S&P 500 for the ten largest, and a huge share of venture funding goes to AI: by the first half of 2025, AI start-ups were raising about half of all global venture capital. Every major bubble in history began with a real idea attracting too much money too fast.

## What happened in the dot-com bubble?

In the late 1990s, everyone wanted a piece of the internet. Companies added ".com" to their names and raised money without products. **Pets.com** spent about **$11.8 million** on advertising, including a Super Bowl commercial, while its revenue was about **$619,000**. The internet was the future, but being right about the technology is not the same as being right about the price. Between 2000 and 2002, about **$5 trillion** of value was wiped out and the **Nasdaq fell 78%**. Thousands of companies vanished; the internet itself kept growing.

![In the dot-com crash of 2000 to 2002, the Nasdaq fell 78% and about $5 trillion of value was wiped out.](https://finniki.com/blog/media/6585ecd340b74e24f5b7646b.jpg)

Today the label is "AI-powered" instead of ".com": by 2023, about 27% of listed companies worldwide mentioned AI in investor reports, compared with 18% of US companies mentioning the internet in 1999.

## Why is AI different from the dot-com bubble?

Because AI is already delivering. Many dot-com companies sold promises; AI is in use today. Hospitals use it to read scans and spot early cancer, developers fix code in seconds instead of hours, small businesses answer customers around the clock, retailers forecast demand to cut waste, financial advisers analyse portfolios in seconds, and every streaming and social app uses it to decide what you see. The technology is real and transformative. The question is price, not substance.

## What are the warning signs of an AI bubble?

Think of buying a house you love in a great location: a good investment, until you pay ten times its value because everyone else is rushing in. Three red flags:

1. **Valuations built on dreams.** Research labs with no product and no revenue are valued in billions on the promise of something that may be a decade away, or may never come.
2. **Prices ahead of adoption.** Real adoption is slow: companies must upgrade systems, train staff, handle regulation and test solutions. A factory or hospital cannot become "AI-powered" overnight, but share prices move as if they already have. Even large tech companies have delayed promised AI features.
3. **Fear of missing out.** Companies rebrand ordinary software as "AI" to attract money, and investors pile in so as not to miss out.

![Three bubble warning signs: valuations without revenue, prices running ahead of adoption, and everything labelled AI-powered.](https://finniki.com/blog/media/3bda2baf449d596a2fca5a5f.jpg)

## What would an AI crash look like?

A run of bad news, such as a major AI company failing to deliver or a big round of funding drying up, could send technology shares sharply lower. Start-ups burning cash without revenue would fail to raise money and close, hype-driven "AI specialist" jobs would disappear, and free AI tools would vanish or start charging. But, as with the internet, **AI itself would not disappear**: the technology would mature, the pretenders would exit and genuine innovation would continue.

## Which AI companies would survive a crash?

The ones building real things. In the dot-com crash, Pets.com, famous for its sock-puppet mascot, disappeared; Amazon, which was losing money but building real infrastructure and solving real problems, used the crash to strengthen its position. Today's equivalents: companies using AI to detect disease, run supply chains or genuinely raise productivity are likely to last; companies slapping "AI" on old products or promising what they cannot deliver are not.

## How do you protect your money from an AI bubble?

Five ways:

1. **Diversify.** Do not let your portfolio look like a tech conference guest list. Mix AI exposure with stable shares, index funds and some international exposure, like a balanced thali.
2. **Keep some cash ready.** When bubbles burst, prepared investors buy good businesses cheaply. As Warren Buffett puts it, be fearful when others are greedy.
3. **Apply the grandmother test.** If you cannot explain what a company does to your grandmother without buzzwords and future promises, stay away. Look for real customers, revenue and products.
4. **Use a barbell.** Keep most of your money in stable investments and limit high-risk AI bets to about **10 to 20%**, so you benefit if AI soars and survive if it crashes.
5. **Rebalance.** If AI shares grow from your planned 20% to 50% of your portfolio, sell some and bring it back to target.

![Protect yourself: cap AI at 10 to 20% of the portfolio, own companies with real revenue, and rebalance while keeping some cash.](https://finniki.com/blog/media/dc8341aa68c50e764c6ad7a0.jpg)

And the best protection is yourself: reduce debt, build more than one income stream, and learn skills that AI enhances rather than replaces. Be the person who uses AI as a tool.

## FAQs

### Is AI in a bubble?

Parts of the market show bubble signs, such as huge valuations for companies without revenue, but the technology itself is real and already widely used.

### What did the dot-com crash do to markets?

Between 2000 and 2002, the Nasdaq fell 78% and about $5 trillion of value was wiped out, but the internet kept growing.

### Should I sell my AI stocks?

Not necessarily. Keep AI to a share of your portfolio you could afford to see fall sharply, favour companies with real revenue, and rebalance when it grows too large.

### How much of my portfolio should be in AI stocks?

A limited slice, often suggested at 10 to 20% for high-risk themes, with the rest diversified across stable investments.

### Will AI survive if the bubble bursts?

Yes. As with the internet after 2000, the technology would continue while overvalued and weak companies fall away.

### How can I protect my career from AI?

Learn to use AI tools in your work and build skills AI enhances, rather than relying on hype-driven roles.

## Invest in the revolution, not the hype

Themes like AI belong in a portfolio in proportion, not as the whole of it. [Book a session](https://finniki.com/book) to check your exposure, or find more episodes on [FinNiki videos](https://finniki.com/videos).