OpenAI Plans to Buy Google AI Chips to Reduce Reliance on Nvidia

OpenAI Plans to Buy Google's Tensor Processing Units

OpenAI plans to buy Google AI chips, according to a report from The Information citing a person familiar with the matter1

1

. The planned purchase would give OpenAI access to Google's Tensor Processing Units (TPUs), processors specifically designed for AI workloads, marking a strategic shift for one of Nvidia's largest customers2

2

. Neither OpenAI nor Google has confirmed the report, and financial terms, volume, and delivery dates remain undisclosed3

3

. OpenAI declined to comment to Reuters, while Google did not immediately respond to requests for comment1

1

.

Diversify AI Hardware Supply to Reduce Reliance on Nvidia

The move would help OpenAI reduce reliance on Nvidia, whose graphics processing units (GPUs) currently power most of the ChatGPT maker's AI computing resources for training models and running inference workloads4

4

. OpenAI is one of Nvidia's biggest customers, and Nvidia invested $30 billion in OpenAI's recent funding round after signing a letter of intent for up to $100 billion tied to deploying at least 10 gigawatts of Nvidia systems3

3

. Google's TPUs have gained traction as an alternative to Nvidia's market-leading AI chips as companies seek cheaper options and chips suited to running AI models5

5

. A purchase would mean owning the hardware instead of paying Google Cloud by the hour, giving OpenAI more control over its supply chain3

3

.

Growing Commercial Ties Between Competing AI Giants

The planned purchase would deepen ties between OpenAI and Alphabet despite their competition in AI products and search1

1

. OpenAI recently announced that its frontier models would become available on Google Cloud, following a similar move with Amazon Web Services2

2

. Reuters reported in June 2025 that OpenAI had begun renting Google's AI chips, though OpenAI described it at the time as early testing with no plans to deploy them at scale3

3

. OpenAI is adapting its models for Google's TPUs and has been renting them via Google Cloud since 20254

4

. A job listing from OpenAI in San Francisco asks for an engineer to lead the optimization of its frontier models on TPUs, suggesting serious intent to deploy the hardware3

3

.

Google's TPU Business Gains Momentum

Google began selling TPUs directly to some customers this year after years of offering them through Google Cloud5

5

. In April, Google said it would sell TPU hardware directly to a small group of customers who can run the chips in their own data centers2

2

. Demand has grown on the back of deals with customers including Anthropic, and Google has sold so much TPU capacity that its own researchers have had to wait for access2

2

. Alphabet's CFO Anat Ashkenazi expects the company to book a small share of TPU sales revenue by the end of this year, with most revenue coming in 20272

2

. Anthropic signed an agreement with Google and Broadcom in April for multiple gigawatts of next-generation TPU capacity expected to come online starting in 20273

3

. Oracle and Google also recently signed a deal to bring TPUs to Oracle Cloud Infrastructure from 20274

4

.

OpenAI's Multi-Vendor AI Infrastructure Strategy

OpenAI is pursuing a multi-vendor approach to secure AI computing resources as tech companies spend heavily on AI infrastructure. Amazon, Microsoft, and Alphabet plan to spend as much as $600 billion combined on AI infrastructure this year, according to Reuters2

2

. Beyond Google, OpenAI gave AMD the right to buy up to 160 million AMD shares, roughly 10% of the company, tied to deploying 6 gigawatts of AMD chips3

3

. OpenAI is also building its own hardware with Broadcom, unveiling Jalapeño, its first custom AI chip designed for inference, with initial deployment planned for the end of 20263

3

. Richard Ho, who leads OpenAI's hardware program, spent nearly nine years at Google and was one of the leaders of its Cloud TPU project before joining OpenAI3

3

. This diversification strategy reflects the industry's recognition that relying on a single supplier for machine learning workloads creates supply chain vulnerabilities as demand for AI accelerators continues to surge.