Companies OpenAI and Nvidia have announced a strategic partnership to create a powerful AI infrastructure. They signed a letter of intent, under which Nvidia plans to invest up to $100 billion in building data centers for OpenAI. The agreement plans to deploy at least 10 gigawatts of capacity based on Nvidia systems, equivalent to millions of graphics cards, allowing for the processing of large volumes of data for new AI models.
The first gigawatt of capacity is planned to be launched in the second half of 2026 on Nvidia’s Vera Rubin platform. OpenAI has identified Nvidia as its primary strategic partner for computational and network infrastructure, which will help the company scale its services and introduce new features for users.
The partnership with Nvidia will allow OpenAI to reduce its dependence on Microsoft, which remains an important partner but is no longer the sole provider of cloud resources. OpenAI is also collaborating with companies like Oracle, SoftBank, and others to develop its own data centers and AI-based products.
According to Sam Altman, CEO of OpenAI, the new infrastructure will be the foundation for further breakthroughs in AI and will allow the creation of products for millions of users and companies. Recently, the number of active ChatGPT users exceeded 700 million per week. OpenAI plans to launch new features based on powerful computational resources, some of which will initially be available to Pro subscribers.
In parallel with this agreement, OpenAI is completing the development of its own AI chip, production of which is set to start in 2026. The company does not plan to sell this chip to external clients but will use it for its own services to ensure competitive advantages in the AI market.
