Nvidia has revealed its plans to invest up to USD 100 billion in OpenAI while also supplying it with advanced data centre chips.
The companies confirmed the plans in a joint announcement, describing the arrangement as two connected transactions: Nvidia will provide non-voting shares funding, which OpenAI can then use to acquire the hardware.
The first deliveries of Nvidia systems are scheduled for late 2026, with about one gigawatt of computing power expected to come online in the second half of that year on a platform named Vera Rubin. In total, the partners signed a letter of intent to deploy at least 10 gigawatts of capacity. That amount of energy usage is comparable to the annual consumption of more than eight million US households.
Implications for the AI sector
OpenAI, recently valued at around USD 500 billion, is already one of Nvidia’s major clients. Industry observers note that the agreement strengthens ties between two of the most influential players in artificial intelligence, raising concerns among rivals about reduced competition. Nvidia officials said that computing power is now the foundation of economic progress, adding that the collaboration is intended to accelerate new breakthroughs and expand access to AI tools.
The investment is expected to begin with an initial USD 10 billion once final terms are reached. According to Reuters, Nvidia shares climbed over 4% following the news, while Oracle stock rose nearly 6% as the company is part of a wider USD 500 billion project, codenamed Stargate, alongside OpenAI, Microsoft, and SoftBank to develop global AI data centres.
Wider industry dynamics may also be influenced. OpenAI continues to explore building its own chips in cooperation with Broadcom and Taiwan Semiconductor Manufacturing, and Microsoft remains a central backer. At the same time, regulators in Washington have been weighing closer scrutiny of the concentration of power in the AI sector.
