OpenAI pledges to buy Google AI chips in its first move away from depending solely on Nvidia

OpenAI intends to acquire Google's AI chips as their computing needs grow, diverging from their ongoing reliance on Nvidia processors. The Information reports that this purchase would involve Google's tensor processing units (TPUs), although financial specifics and delivery timelines have not been disclosed. This move comes as technology firms seek more processors to handle the escalating costs of model training and deployment.

Google's TPUs have proven attractive to companies like OpenAI, who are looking for alternatives to Nvidia's GPUs, which remain popular for training and operating AI models. Although OpenAI has already been a significant customer of Nvidia, purchasing GPUs for its AI systems, including the ChatGPT service, this reported Google purchase suggests OpenAI's intention to secure computing resources from a different supplier.

In 2025, OpenAI began renting Google's AI chips to bolster its products, implying that their relationship predated this latest purchasing strategy. The potential transaction also follows OpenAI's recent announcement that its advanced AI models will be available through Google Cloud, creating an additional commercial link between the two companies.

Despite their competitive nature in AI development and other tech sectors, Google and OpenAI have increasingly collaborated on computing resources and access to their respective products. OpenAI has also pursued its own AI chip development, unveiling an AI chip in partnership with semiconductor company Broadcom in June 2026. They intend to deploy this chip by the end of 2026, part of a multi-generational effort to create specialized computing hardware.

This custom chip program, coupled with purchases from Nvidia and the potential Google deal, indicates that OpenAI is exploring multiple sources of computing capacity. Google's expanding chip business may benefit from this demand, especially as Alphabet seeks financial returns from its significant investments in AI infrastructure.

Finance chief Anat Ashkenazi stated that Alphabet anticipates recognizing a small portion of revenue from TPU sales agreements by the end of 2026, with most revenue expected during 2027, suggesting a gradual financial contribution over time. The AI industry collectively is projected to invest as much as $600 billion in AI infrastructure by 2026, according to the report.

This spending has drawn investor attention, with inquiries emerging regarding the timing of investments in computing capacity to generate substantial financial returns.