The AI ​​revolution begins with a gamble by Nvidia and OpenAI

On the same topic:

Records, as we know, are made to be broken. But at this rate, AI is rivaling Mondo Duplantis in pole vaulting. The agreement between Nvidia and OpenAI, announced as an investment of up to $100 billion to build at least ten gigawatts of data centers based on Jensen Huang's company's hardware, is destined to forever change the landscape of artificial intelligence : it's no longer just about developing algorithms or models, but about controlling the physical infrastructure that makes cutting-edge AI possible. The agreement reveals peculiar geopolitical dynamics, with Trump informed days in advance, demonstrating the strategic value of the operation in the competition with China.

All of OpenAI's infrastructure projects now fall under the Stargate umbrella, the $500 billion program announced by the president to consolidate American supremacy in AI. The move is a typical vertical supply chain integration, but on an unprecedented scale: not only does it grant OpenAI privileged access to millions of chips, but it also transforms Nvidia from a mere supplier to a strategic partner with an equity stake in the company. The first gigawatt of capacity is expected to arrive in the second half of 2026, based on Nvidia's new Vera Rubin platform. The first tranche of the payment—a $10 billion transfer—paid by Nvidia will be based on the American giant's current valuation of $500 billion, while subsequent payments will be based on the company's value at the time of disbursement. The timing is revealing: it comes just as the Federal Reserve has just cut interest rates, a sign of an economic slowdown. Yet the announcement has earned Nvidia, already at the top of the world's most valuable companies, an additional $170 billion in market capitalization. The deal consolidates Nvidia's dominance in an already unbalanced sector: controlling the hardware means dictating the pace and conditions of innovation, while competitors risk being relegated to marginal or defensive roles; adding the software component could prove to be the market's killer move.

Regulatory questions abound: a supplier simultaneously becoming an investor in one of the world's largest clients could raise antitrust objections, while OpenAI's corporate structure, transitioning to a public benefit corporation with a nonprofit board, will require a new balance. The geopolitical dimension is also evident: if most of these data centers are built in the United States, America will strengthen its leadership in AI, once again shifting the balance of power with China. Finally, some analysts wonder whether this move might bring us even closer, in a context where only those selling the hardware component are currently profiting in AI, to the bursting of a speculative bubble of unprecedented proportions . Because behind the mathematical abstraction of the models remain the very real balance of power between leaders and companies. It is at this level that who will control the next generation of AI is defined. The final question is what place Europe could have in this scenario.

More on these topics: