Nvidia and OpenAI: Where Did the $100 Billion Go?

Remember those loud headlines where Nvidia and OpenAI were practically merging in a financial ecstasy, backed by a hundred billion dollars? It looked as if Jensen Huang and Sam Altman were building a new empire, one supplying endless computing power, the other endless intelligence. Reality turned out to be far more prosaic: the huge bag of money whispered about in the corridors of Silicon Valley and on the front pages of business media simply vanished from the radar. Investors who had already baked this growth into their strategies are now rubbing their temples in bewilderment, trying to work out whether it was a clever maneuver or just wishful thinking.

To understand why the deal has "dissolved" right now, look at the context of recent months. OpenAI is rapidly turning from a cozy research startup into an aggressive corporate machine. Sam Altman no longer wants merely to buy chips: he wants to build his own plants, design his own architectures and fully control the supply chain. Nvidia, for its part, stopped being just a "shovel seller" to gold diggers long ago. It has become the most valuable player on the planet, and such critical dependence on a single customer, even the most high-profile one, is starting to look like a strategic vulnerability rather than an advantage.

A $100 billion deal could have become either the crown of this alliance or a golden cage for both sides. Had Nvidia put in that money, it would effectively have bought itself a decisive voice in the future development of GPT and all later iterations of neural networks. But Altman, by all appearances, is not ready to hand over the keys to his kingdom. His ambitions to build a global network of data centers and semiconductor capacity conflict directly with Nvidia's business model. Why would Jensen Huang feed someone who may become his most dangerous competitor in specialized hardware tomorrow?

Regulators are also a factor that cannot be ignored. Antitrust authorities in the US and Europe are now examining the AI industry under a very powerful microscope. A tie-up between the world's largest chipmaker and the undisputed leader in language models would have triggered such a wave of investigations and lawsuits that the companies would have been bogged down in legal routine for decades. It is easier to part quietly and pretend that no official documents were ever signed than to explain to officials why you are not trying to monopolize the future of human civilization. It is a classic example of "too big" companies becoming hostages of their own scale.

For the rest of the market, this is a deeply worrying alarm bell. If the industry's two main locomotives could not agree on money, the era of "blind optimism" and endless checks is coming to an end. Every investment dollar will now be checked for payback, and every partnership deal for hidden knives in the back. Nvidia will keep selling its Blackwell chips to anyone who wants them, while OpenAI will have to find new allies, possibly among those who do not see it as a direct threat to their core business. We are entering a phase of harsh pragmatism, where old friends become competitors faster than the weight class of models gets updated.

Bottom line: The romance is over, and a harsh redistribution of spheres of influence has begun. Can OpenAI keep its lead without the exclusive backing of the planet's top chipmaker?