“If you are creating a company, what you will aspire to create a monopoly and avoid competition. The competition is of losers“. Those words were pronounced by Peter Thiel In a talk that gave Stanford students on October 7, 2014. Who gave way to talk, by the way, It was Sam Altman.
During those 50 minutes Thiel – Paypal and Palantir co -confounder, billionaire, successful investor, and obsessed with rejuvenation– It raised precisely that fundamental idea: that all companies aspire to become a monopoly. They do not say it publicly, of course – that entails legal and regulatory problems – but the goal is that.
It has happened in multiple cases in the world of technology. Windows is a de facto monopoly in desktop operating systems, Android and iOS are an accepted duopoly in mobile platforms, and Google is an indisputable monopoly in the world of searches. In all those cases, those who have tried to compete – and there have been attempts – have failed. The competition was indeed of losers.
Another monopoly in sight: that of AI
And here we are facing a situation that reminds us of all the above. In the world of AI we are living fierce competition. One in which dozens of companies try to develop their models and applications of AI to win the items to the others. To become monopolies.
OpenAi carries the lead. The question is whether you can maintain that leadership.
What model of AI is better than others? It is not entirely clear. The appearance of Grok 3 seems to have opted the balance in its favor, but its theoretical superiority in some tests sounds like the same as other previous releases sounded: if it is really the best, it will not last long to be.
In fact, competition between these models has made us in a situation in which, sincerely, they are all quite good. There will be, of course, use cases in which some will stand out on others – this best program, this writes better, this looks better – but everything seems to point to the differences will be less and less evident.
The improvements we are seeing in the market are getting smaller and, above all, more expensive. Grok 3 has been trained in the Gigantic Supercluster of XAI with at least 100,000 GPUS H100 of NVIDIA, but despite all those resources and that investment, which has achieved the startup of Elon Musk is to put themselves at the level of its compeditors, not to offer A product that suddenly is remarkably better. The same With the imminent GPT-4.5.
What are Ia companies trying now? Two things:
- Try to make your products simpler to use: Less variety of models, or at least hide that variety, as Openai proposes in the future and simplified chatgpt version.
- As traditional models do not advance so much, they raise New models reasoning (Deepseek R1, O3-mini) or agents (Operator) that encourage us to pay for increasingly faces.
That strategy is once again intended for the same: that their products stand out on the competition and achieve the desired objective: create the next great monopoly.
In that race there is for now an outstanding protagonist, at least if we take into account the number of usuals of each model. According to CNBC data and others collected By Ed Zitronthe estimated current situation is as follows:
- ChatgPPT: 400 million weekly active users
- Deepseek: 27 million active users monthly
- Gemini: 18 million active users monthly
- COPILOT: 11 million active users monthly
- Perplexity: 8 million active users monthly
- Claude: 2 million active users monthly
Especially the few users that Claude has theoretically, but what is evident is that Today for millions of users IA = chatgpt. Arriving the first here has made the difference for Openai, which has also constantly iterated to maintain that leadership.
Will they keep doing it? Will they become the de facto monopoly of AI? Of course they have ballots for it, but it is still very soon to be able to say it. The frantic advance of this technology makes it more difficult than ever who will win the race … if there is only one winner.
In fact, here the situation is very different because there are forces that Openai does not control. And they are too relevant forces.
- Windows: there is 1,400 million active PCs with Windows worldwide. That number includes both Windows 10 Machines with Windows 11.
- Android: there is 3.5 billion active devices based on Android.
- iOS/iPados: there is 2,350 million IOS active devices.
Do you think Microsoft, Google and Apple are going to let Openai take the cat to the water? Not much less. They will do everything they can so that the AI we use on their devices is yours. That’s why The idyll between Microsoft and OpenAi FLUQUEAand that’s why Google and Apple are little by little –too little by little In the case of Apple – integrating more and more functions of AI in its mobile phones.
These companies, de facto monopolies already in their markets, have as clear as Peter Thiel that competition is for losers. They probably have it much clearer, especially because they have been since before Thiel gave their famous talk.
We are therefore facing a situation that is analogous to that of the rest of the digital businesses with which we have lived. In the spotify audio streaming world it is almost a monopoly (31.7% share), while in the video the thing is much more distributed for the moment although Netflix stands out.
In the world of Electronic Commerce Amazon marks the passage, on social networks to Facebook there are almost no one to have more examples such as Uber, LinkedIn, Match Group (Tinder, Okcupid), or PayPal, which are also clear leaders in their respective markets.
Is there competition? Of course. Does competition change things? Normally, not too much: Firefox has not changed them In browsers, mobile operating systems that They tried to give options to Android and iOS either, and bing of course He has not managed to capture too much share In search engines.
The Open Source the AI as an alternative
There is another additional interesting factor in the field of artificial intelligence: Open Source models. Linux could not succeed on the desktop –It was never his year– But he did it in almost everything else, and the models of the Open Source may be surprising protagonists of this land.
In fact Depseek has made it clear how the models will end up becoming a Commoditysomething we already assume – as having a mobile in the pocket – and that will differentiate to each other is the application layer: what allow us to do their services and applications with those models. It is there where the models of the Open Source are winning battles, as perplexity or freeepik have shown, that without having their own models – they approve Open Source models – they propose a powerful idea: you do not worry about AI: the important thing is that you can do What you need.
Maybe there is the answer, after all. Perhaps who has the best of those answers manages to reach the Holy Grail of the companies in this segment. That is not so much to get an AGI. That too, of course, but that is a means to the ultimate goal.
Become in the AI monopoly.
Image | Adi Goldstein
In Xataka | Europe fine with 800 million euros. He just joined the select club of companies sanctioned by monopoly