As Lisbon buzzes with talk of agentic AI and multi-trillion dollar markets at Web Summit 2026, a ghost of tech past whispers a warning. Are we repeating the dot-com bubble? It's the question hanging over every billion-dollar valuation.
Eyeballs Then, Engagement Now
Remember the late
'90s? The key metric wasn't profit; it was 'eyeballs' or 'stickiness'. Companies like Pets.com burned through hundreds of millions on Super Bowl ads to attract users, assuming monetization would somehow follow. Today, the jargon has changed but the logic feels eerily similar. Instead of eyeballs, we hear about user engagement, model performance, or the sheer number of API calls. While these metrics aren't useless, the dot-com era teaches a brutal lesson: metrics that don't have a clear, plausible path to revenue are just vanity. The market eventually demands to see the receipts. The founders who survive will be those who can explain their business model without using the word 'magic'.
This Time It's Built on… Profit?
Here's the biggest difference between now and then, and it's a major one. The dot-com boom was largely fueled by speculative IPOs for companies with little to no revenue. In contrast, today’s AI infrastructure build-out is being driven by the world's most profitable companies—Microsoft, Google, Amazon—reinvesting immense cash flows. They see AI as essential for their survival and are spending hundreds of billions on data centers and chips. This isn't purely speculative; it's a strategic arms race funded by real profits. That provides a much stronger foundation than the flimsy capital of the dot-com days. While many AI startups are still burning cash, the ecosystem they live in is supported by giants with very deep pockets.
The Danger of 'AI-Washing'
In the late '90s, companies could boost their stock price simply by adding '.com' to their name. We're seeing a modern version of this with 'AI-washing', where any company that uses a chatbot or an API claims to be an 'AI company'. This is inflating valuations across the board. Recent data shows AI-native companies command revenue multiples of 25-30x, while traditional software-as-a-service companies languish at around 3-6x. The lesson from history is that the market eventually learns to tell the difference between companies using a new technology and companies whose entire business is built on a durable, proprietary advantage in that technology. The crash punished the imitators, but survivors like Amazon and eBay, which had real customer adoption, ultimately thrived.
So, Are We in a Bubble?
Yes and no. The consensus is that while some parts of the AI market are frothy, with valuations for earlynd stage startups doubling in a year, the underlying fundamentals are stronger than in 2000. Analysts argue the current moment feels more like 1995–1997, the 'displacement' phase where a new technology attracts initial investment, rather than the euphoric peak. Unlike the dot–com era's startups, today's top AI firms have real enterprise customers and are generating billions in recurring revenue. The AI boom is less a single bubble and more of a complex weather system, with pockets of extreme hype alongside genuine, transformative shifts. The primary risk might not be a single 'pop' but rather a high concentration of investment in just a few giant companies, creating its own set of dangers.
