Earlier this week, Nvidia announced a staggering $100 billion investment in OpenAI, the company behind ChatGPT. This marks one of the largest investments ever in the AI sector and will significantly accelerate OpenAI’s efforts to develop more advanced AI models. New details about the collaboration have now begun to surface.
According to people familiar with the matter, OpenAI plans to lease Nvidia’s AI chips rather than purchasing them outright. Typically, AI companies acquire chips from semiconductor firms to deploy in their massive data centers. However, thanks to Nvidia’s recent investment, OpenAI will now be able to spread its costs over the useful life of the GPUs.
As part of the investment agreement, OpenAI is set to receive $10 billion from Nvidia in the near future, with the remaining funds to be injected into the company gradually over time.
According to Nvidia CEO Jensen Huang, building an AI data center with one gigawatt of capacity costs approximately $50 billion, of which roughly $35 billion is allocated toward acquiring Nvidia GPUs. By contrast, under the leasing option offered to OpenAI, the AI company can alleviate the financial strain of upfront purchases.
OpenAI, which was founded as a nonprofit in 2015, recently abandoned its shift toward a for-profit model and announced that it will continue to operate under nonprofit control.
In late July, OpenAI announced that its annualized revenue had reached $12 billion. For 2025, the company projects total revenue in the range of $15 to $20 billion. While OpenAI’s revenue has grown compared to previous years, the company is still operating at a loss. According to Bloomberg, OpenAI does not expect to become cash-flow positive until 2029, when its annual revenue is projected to reach $125 billion.
In addition to the steep costs of acquiring AI GPUs and training models, OpenAI has found itself in a talent war with Meta, forcing the company to offer substantial compensation packages to secure the market’s top talent.