OpenAI Plans to Buy Google’s AI Chips
OpenAI plans to buy artificial intelligence chips from Google to expand computing power for its growing AI operations. The reported deal could deepen ties between the companies, which compete in AI products and search. However, neither company has confirmed the proposed purchase.
OpenAI intends to acquire Google’s Tensor Processing Units (TPUs). The report cites a person familiar with the matter. Meanwhile, financial terms, purchase volumes and delivery timelines remain undisclosed.
The move would give OpenAI another source of AI computing hardware beyond Nvidia, whose graphics processing units (GPUs) currently play a major role in training and running its models. As demand for AI services rises, access to computing capacity has become a central concern for technology companies.
Google TPUs Emerge as an Alternative to Nvidia
Google developed TPUs specifically to handle AI workloads. Unlike general-purpose processors, these chips accelerate tasks such as model training and inference, which involves generating responses from trained AI systems. As a result, companies seeking alternatives to Nvidia’s widely used GPUs have turned to TPUs.
Google has traditionally provided access to its TPUs through Google Cloud. However, the company began selling the hardware directly to selected customers in 2026, expanding its reach beyond its own cloud infrastructure. This strategy has helped Google attract demand from businesses building large-scale AI systems.
OpenAI already has experience using Google’s chips. Reuters reported in June 2025 that the ChatGPT developer had started renting TPU capacity. The reported purchase plan could mark another step toward using Google’s hardware more extensively.
OpenAI Diversifies Its AI Infrastructure Strategy
OpenAI remains one of Nvidia’s largest customers. Nevertheless, the reported plan suggests the company wants more flexibility as it expands the infrastructure required to train and operate advanced AI models. Multiple hardware suppliers could also help it manage computing costs and capacity constraints.
OpenAI is also developing its own processors with Broadcom. In June 2026, the company outlined plans for a custom AI chip programme, with deployment expected to begin by the end of the year. Therefore, its reported interest in Google TPUs forms part of a wider effort to diversify its computing resources.
Meanwhile, commercial ties between OpenAI and Google continue to grow despite their competition. OpenAI recently announced plans to make its advanced AI models available through Google Cloud. Google is also expanding its TPU business through partnerships with other cloud providers, including Oracle.
The wider market reflects the scale of this infrastructure race. Amazon, Microsoft, and Google’s parent company, Alphabet, plan to spend up to $600 billion combined on AI infrastructure in 2026, according to Reuters. However, investors continue to question how quickly these investments will produce meaningful returns.
For now, OpenAI’s proposed TPU purchase remains unconfirmed. Still, the development highlights how leading AI companies are exploring different chip architectures and suppliers to meet rising demand. If the purchase proceeds, it could strengthen Google’s position in the AI semiconductor market while giving OpenAI another option for powering its models.
