"Nvidia Chips Are Too Expensive," OpenAI Also Turns to Google TPUs

After renting Google TPUs, OpenAI moves to buy them outright
Competition in the AI market, expanding cooperation on infrastructure

A foreign media report said OpenAI, the developer of ChatGPT, is looking to buy artificial intelligence (AI) chips made by its competitor Google. Amid a surge in demand for AI computing, the move is seen as an effort to diversify a semiconductor supply chain dominated by Nvidia.

AP Yonhap News

On October 27 local time, Bloomberg and U.S. technology news outlet The Information reported that OpenAI plans to buy Google's AI chips, known as tensor processing units (TPUs). OpenAI has been renting Google's AI chips since June 2025, and this latest move would take that relationship a step further by purchasing them. The size of the deal and its terms have not been disclosed.

Google developed its TPUs more than a decade ago to improve the processing speed and efficiency of its search engine, but their use has since expanded as they have also proved effective for AI computing. In particular, they have emerged as a relatively low-cost alternative in an AI chip market dominated by Nvidia. Google is also reportedly supplying TPUs to Anthropic, a competitor of OpenAI.

Big Tech companies face enormous costs for investing in AI infrastructure. Amazon, Microsoft and Alphabet, Google's parent company, plan to invest a combined total of nearly $600 billion in AI infrastructure this year. As investment has surged, concerns about profitability have also grown.

Revenue from Google's TPUs is expected to begin making a significant contribution starting next year. Alphabet Chief Financial Officer Anat Ashkenazi previously said that some revenue from TPU sales contracts would begin to be recognized toward the end of this year, with most of it recognized next year.

Google is also expanding its TPU supply network by partnering with other cloud companies. Oracle announced on October 28 that it plans to offer Google TPUs through its cloud infrastructure. This marks the first time a major cloud provider outside Google will offer TPUs, with the service set to begin next year. Oracle will add Google's TPUs to its existing Nvidia GPUs.

Amin Vahdat, Google's senior vice president of AI and infrastructure, said, "Next-generation AI requires a diverse range of computing resources," adding, "More companies will be able to use TPUs and develop and operate complex AI applications."

Meanwhile, Alphabet is scheduled to report its quarterly earnings after the market closes on October 28. Foreign media noted that investors are looking for signs that AI investment continues to drive growth in the cloud business.

This content was produced with the assistance of AI translation services.