Nvidia to invest 100 billion in OpenAI for data centre construction

Nvidia will invest up to $100 billion in OpenAI under a new strategic agreement announced in recent hours. The agreement includes the deployment of at least 10 gigawatts of Nvidia systems dedicated to the artificial intelligence infrastructure of the Sam Altman-led company. The two companies have signed a 'letter of intent' that will be finalised in the coming weeks, with the first deployment phase scheduled for the second half of 2026.

Nvidia's financial commitment will not be immediate, but progressive: funds will be disbursed as the gigawatts come on stream. In this way, Jensen Huang's company avoids having to pay the full amount immediately and links its financial commitment to the achievement of specific, verifiable targets. The scale of the deal is unprecedented: 10 GW of capacity is equivalent to an infrastructure that will require enormous resources not only in terms of chips, but also in terms of power, cooling and connectivity.

It must also be said that the deal confirms the very close ties between Nvidia and OpenAI, which have been two of the main drivers of the AI boom for about three years now. It cannot be forgotten, in fact, that the demand for Nvidia GPUs began to grow precisely with the launch of ChatGPT in 2022 and, even today, OpenAI relies on those chips to develop its own models and make them available to users. A combination, in short, that has grown dramatically together. Nvidia has become the most valuable company in the world, OpenAI already has 700 million active users on a weekly basis

Going into technical detail, it is clear that the project demonstrates the huge amount of chips that OpenAI will need to develop the next generation of artificial intelligence, more advanced than current models.

"This is a monumental initiative," said Nvidia CEO Jensen Huang, in an interview in San Jose with OpenAI CEO Sam Altman and president Greg Brockman.