AI bubble or no bubble? It is a legitimate question, especially if you remember 2000, when the Internet seemed to offer endless promise and then came the crash.
Today, though, the landscape is different. In the dot-com phase, there was often an enormous mismatch between expectations, real value and business models. With AI, by contrast, the impact is already tangible and cuts across sectors: productivity, customer care, research, coding and education.
The comparison with the dot-com bubble
The parallel keeps returning because we recognise the same mix of hype, capital and narratives. But there are two important differences.
In 2000, many companies were ‘Internet companies’ in name only: little product, little distribution and fragile revenues. Today, AI is already part of everyday processes and tools, with measurable usefulness.
That does not mean there cannot be excesses. It means that, if a correction comes, it could affect selected areas rather than the entire paradigm.
OpenAI today: high costs, extremely high usage, evolving monetisation
Take OpenAI. Infrastructure costs may indeed exceed revenues, but it is hard to imagine the model staying unchanged. On consumer platforms, monetisation often comes after distribution has been secured.
Estimates circulating in public debate put the user base close to a billion, with usage frequencies increasingly similar to those of a search engine, which it also partly replaces. If usage is that frequent, the natural question becomes: how do you monetise that ‘attention’?
Advertising inside AI: the next logical step
One plausible step is advertising. Recruitment for paid-ad roles suggests the subject is on the table, and the trajectory recalls what turned Google into a revenue machine: auctions, keywords and intent.
The potential difference is that here we are talking not only about keywords but also about more clearly defined intentions and profiles, because conversations contain context, goals and constraints. Google, too, can move in the same direction with Gemini.
A practical scenario: a free version with ads and a paid one without. It is an approach already seen in many digital products, and it keeps the barrier to entry extremely low.
The winner takes it all and the importance of distribution
In platform markets, there is often one rule: whoever becomes the standard wins. That is why investing heavily today makes sense, even at the cost of negative margins, to reap the rewards tomorrow. Distribution, even more than technology, can become the defensive line.
Cloud and vibe coding: a huge market opens up
In business and coding, OpenAI operates through partnerships and infrastructure. Historically, there is its relationship with Microsoft, but collaborations with other cloud players have also emerged. ‘Vibe coding’ opens up a gigantic market: if more people build software, the need for hosting, deployment and easy-to-use services grows too.
Here, simplicity is a critical success factor: whoever makes publishing, scaling and maintaining projects easy captures a huge share of the new demand.
Falling costs and competition: why the curve may change
Costs are another piece of the puzzle. While revenues can grow through new models, infrastructure could become relatively less expensive thanks to competition and hardware advances. We are still at the beginning of the technology cycle.
Where the real risk lies
There is no shortage of risks. For OpenAI, one is valuation: if it becomes excessive and Google catches up technologically and in distribution, pressure increases. Another is internal: when stock options are at stake, the temptation to cash in and change direction can accelerate talent turnover.
The bubble risk, however, is more tangible in the constellation of startups born out of the hype: high valuations simply because they ‘use AI’. The problem is familiar and keeps recurring: if the core platform adds a feature, the startup that depended on that feature gets ‘squeezed’.
Conclusion
So: bubble or no bubble? Perhaps both, but on different levels. The AI paradigm seems set to stay, while part of the ecosystem could face a harsh shakeout, especially where competitive advantage is thin.
And bubble or no bubble, we will get to the stock-market listing.
If you work in digital and marketing, the useful question is not only ‘who will win’, but ‘where value will move’: search, conversational advertising, cloud and tools that make technology accessible to many.