OpenAI sticks with Nvidia, holds off on Google's AI chips

OpenAI has clarified that it has no active plans to use Google's in-house artificial intelligence chips, known as Tensor Processing Units (TPUs), to power its products, despite recent reports suggesting otherwise.

The statement comes after a report from The Information indicated that the ChatGPT-maker was exploring the use of Google’s TPUs to meet rising demand for AI computing. A spokesperson for OpenAI confirmed to Reuters on Sunday that while the company is in “early testing with some of Google’s TPUs,” it currently has “no plans to deploy them at scale.”

Although it is common practice for AI companies to test a variety of chips, deploying new hardware on a large scale typically involves significant changes to system architecture and software support. OpenAI continues to rely heavily on Nvidia’s graphics processing units (GPUs) and also uses chips from AMD to support its growing compute requirements.

The company has also partnered with Google Cloud to expand its computing capacity, as reported by Reuters earlier in June, a surprising collaboration between two direct competitors in the AI space. However, most of OpenAI’s computing power is still expected to come from GPU servers provided by CoreWeave, a specialised cloud provider.

As the AI race accelerates, OpenAI is working to reduce its reliance on third-party chips. It is currently developing its own custom AI processor and expects to reach a key “tape-out” milestone later this year. The tape-out milestone is when a chip’s design is finalised and sent for manufacturing. Taiwanese media outlet Commercial Times recently reported that OpenAI’s in-house chip, being developed with assistance from Broadcom and Taiwan Semiconductor Manufacturing Company (TSMC), could launch as early as the fourth quarter of 2025.

Meanwhile, Google has been expanding access to its once-internal TPU chips to external customers. This has already attracted major clients, including Apple, as well as OpenAI competitors Anthropic and Safe Superintelligence, both of which were founded by former OpenAI executives.

The growing demand for powerful AI chips, particularly those made by Nvidia, has reshaped the global tech industry and catapulted Nvidia to become the world’s most valuable listed company earlier this year. But as competition heats up and costs rise, OpenAI and others are exploring ways to diversify their hardware strategies.