There were quote screens, an IPO almost every week, and anything with .com or online in its name seemed to be worth the future. Promises were bought at the price of empires.
Part 1 replayed a day on the web before the platforms: searching, chatting, selling, listening, booking, often through European brands. Part 2 opened the atlas: ISPs, portals, mobile, mapping, classifieds, voice over IP. Europe did not own all of the Internet, but it did own several layers of use.
The final part is about the moment that advance jammed. Not because the ideas were bad. Because the bubble broke financing, loaded telecom balance sheets with debt, then installed lasting caution exactly when Europe needed to start again.
The Internet was real: the uses, the engineers and the first customers already existed. What breaks in 2000 is the price: an IPO is enough to turn a promise into public wealth, even without profits, sometimes without solid revenue.
For Europe, this moment matters because it hits at the worst intersection: Internet startups, portals, mobile and telecoms. It is a valuation bubble, but also an industrial bubble.
The mechanism in 5 points
A video explainer of the dot-com bubble, in English (CH 07) and French (CH 03). The opening image stays here; clicking opens YouTube in a new tab.
On March 17, 2000, the Dutch access provider and portal went public at 43 euros per share, with a valuation close to 12 billion euros. The story sold to the market was simple: free Internet, European expansion, a continental brand.
A pan-European promise: Internet access, portal, rapid expansion. The market was not buying profit. It was buying a place in the future.
Within days, the promise cracks. Nina Brink's pre-IPO share sale becomes a symbol: the public buys the future just as insiders have already stepped out.
Lastminute.com sold a very readable idea: using the Internet to fill unsold travel, hotel and leisure inventory. The model made sense. But in March 2000, the market was also paying for the symbol: a .com brand, future growth and a new way to consume.
The shares listed at 380 pence and jumped on day one. Then the music changed. The company survived, but its path shows the core problem: a good idea becomes fragile when it reaches the market priced as if success had already happened.
Lesson: real use, impatient price
Boo.com wanted to sell fashion online at global scale, with a rich experience, heavy visuals and an ambition well ahead of the connections of the time. The concept foreshadowed part of the future of online commerce.
But execution burned cash faster than usage grew. The site was heavy, the organisation expensive and revenues insufficient. In May 2000, Boo.com was liquidated after consuming around 135 million dollars.
Lesson: right intuition, wrong timingiFrance belonged to the first generation of the social web before social networks: personal pages, hosting, community, portal. It was not an empty shell. It was a real gateway into the French-speaking Internet.
The company was sold to Vivendi in stock. On paper, it was a brilliant exit. When Vivendi collapsed, the fortune turned into debt. This is the bubble from a founder's point of view: not only a market curve, but a personal reversal.
Lesson: the exit can become a trap3G licences are the right to use mobile frequencies to build the Internet on phones. Governments sold them through auctions. Operators paid before mass-market 3G phones, before usage, before revenue.
Why was it so heavy? Because a licence is not an app you close. It is a public entry ticket, paid at a high price, to which you still have to add antennas, network, marketing and years of waiting.
By the end of 2002, France Telecom carried around €68bn of net debt, Deutsche Telekom around €61bn. Before even thinking about the next move, they had to deleverage.
That is the European difference: the crash is not only financial. It becomes industrial.
Here's the curve everyone pictures: the NASDAQ, from its March 10, 2000 peak to its October 2002 trough. Below it, more quietly, Germany's NEMAX - which fell even further. Indices rebased to 100 at the peak.
Europe didn't just watch the American bubble: the NEMAX lost nearly 96%, even more than the NASDAQ (−77.9%).
Reading: indices rebased to 100 at the March 10, 2000 peak. NASDAQ Composite 5,048.62 → 1,114.11 (Oct 9, 2002, −77.9%). NEMAX All Share 8,583 → 349.02 (early Oct 2002, −95.9%). Schematic line between documented points. Hover the peak and troughs.
Journalist John Cassidy told, almost hour by hour, how the euphoria turned - and why so many investors wanted to believe, right to the end, in a story too good to be true.
Same damage on both sides of the Atlantic. The difference plays out afterwards: the United States went back to listing and scaling giants; Europe kept inventing, but funded shorter.
Note: VC amounts are orders of magnitude (sources: OECD, EVCA / NVCA; scopes and currencies not strictly comparable). Company milestones are dated; they illustrate a dynamic, not an exhaustive ranking.
Sold in May 2000. A few months before the crash.
Worldnet, one of France's first consumer ISPs, was profitable and still independent. Niel chose a sale rather than a stock-market listing. He exited in cash at the top of the cycle, then built Free. In the story of the dossier, this is the path of someone who crosses the bubble without losing his capacity to attack.
“They know, they look, but they see nothing.”
iFrance was sold to Vivendi in shares. When Vivendi collapsed, the brilliant exit became personal ruin and debt. Simoncini then started again with Meetic. This is the other face of the crash: ideas and founders remain, but financial violence changes their relationship to risk.
Talent does not disappear. The best founders keep going, but they learn that the market can destroy a trajectory in a few months.
After 2000, selling earlier, funding for less time or becoming profitable too quickly can look more reasonable than aiming for a global platform.
The question is not only whether Europe has engineers. It is whether Europe still accepts the long game, losses and scale.
* Yes, I really did want to find vintage pictures of these two entrepreneurs. The dossier is also about that era: its offices, exits, bets and grain.
The bubble didn't kill the ideas. It wore down Europe's tolerance for the long game.
European innovation didn't vanish. Founders kept going, took the hit, started over.
The ability to fund for the long haul, to list, to absorb losses, to keep its champions.
Turning products into global platforms - that's where some of the nerve was lost.
The common signal is clear: many AI players are not profitable, but are valued as if the future were already won. In 2000, World Online, Lastminute or Boo.com were already in that logic: less a profit than an option on the future.
The market pays for a promise of scale before the full economic proof.
Nvidia is already billing massive data-center revenue. Hyperscalers finance the wave with solvent balance sheets.
The EU share of global AI VC deal value in 2025, versus 75% for the United States.
So the question is not only: will there be an AI bubble? There will be excess. The question is: after the purge, who will keep the models, chips, cloud, data, customers and distribution channels?
In 2000 it kept its talent but lost the nerve: the ability to fund the long game, to list, to absorb losses and to keep its champions. Twenty-six years on, the scene has changed. AI already has real revenue, massive infrastructure and solvent customers. It is not a carbon copy of the Internet bubble.
But the European trap looks familiar: being present in research, in use cases, sometimes in products, without owning enough of the critical layers. Models run on mostly American compute, inside mostly American clouds, financed by balance sheets Europe cannot easily imitate. If the wave purges, what remains will matter more than peak valuations.
And dependency is usually noticed too late: one day a service shuts down, an API changes, a model becomes more expensive, a feature disappears for European users, and you discover there is no local equivalent to take over. The longer a dependency lasts, the harder it becomes to reverse.
The real question may no longer be whether Europe can still innovate. But whether it can still choose.
Save the ambition?The dossier “When Europe owned the Internet” is about to close. Do you want to save the changes made to ambition.dat?