OpenAI Targets Google’s AI Chips, Expands Beyond Nvidia

OpenAI Reportedly Weighs Buying Google’s AI Chips to Reduce Reliance on Nvidia

OpenAI is preparing to purchase AI chips from Alphabet’s Google, according to a report cited by Reuters, signaling a potential shift in how the ChatGPT maker sources the computing hardware needed to train and run its models.

Neither OpenAI nor Google confirmed the plan. OpenAI declined to comment and Google did not immediately respond to Reuters. The reported order size and pricing were not disclosed.

OpenAI is already among the largest buyers of Nvidia graphics processing units (GPUs), which remain the dominant standard for AI workloads. Reuters previously reported in June 2025 that OpenAI had begun renting Google’s AI chips, suggesting the company has been testing alternatives beyond Nvidia’s ecosystem.

Google’s chips, known as Tensor Processing Units (TPUs), have gained traction as an option for companies looking for more cost-effective AI compute. Google began selling TPUs directly to some customers this year, expanding access to hardware that was long used primarily within Google’s own infrastructure.

If OpenAI moves from renting to purchasing TPUs, it would also deepen a commercial relationship between two companies that compete in AI products and search, but have increasingly done business together. Last week, OpenAI said it would make its frontier AI models available through Google Cloud.

The chip strategy comes as OpenAI works to diversify its compute supply. In June, OpenAI unveiled its first custom AI chip designed with Broadcom and said it planned to deploy the processor by the end of this year as part of a multi-generation chip program.

For the broader AI and crypto-adjacent compute market, the developments underscore how demand for high-performance chips is pushing major AI developers to seek multiple suppliers. As AI infrastructure becomes a larger cost line for model providers, alternative hardware platforms such as Google’s TPUs are increasingly being positioned as viable complements to Nvidia’s GPUs.