ChatGPT maker OpenAI plans to buy Google’s tensor processing units, according to a media report, in what would be its first move away from relying only on Nvidia chips. The OpenAI Google TPU plan shows how the race for AI computing power is pushing companies to look beyond a single supplier.
Financial terms, the number of chips and delivery timelines have not been disclosed, and neither company has made an official announcement.
What the OpenAI Google TPU Plan Means
OpenAI is one of Nvidia’s biggest customers and uses its graphics processing units to build and run the models behind ChatGPT. It began renting Google’s AI chips through the cloud in 2025, and recently said its advanced models would also be available through Google Cloud. Buying TPUs would take that relationship a step further.
Google’s TPUs are custom chips designed for AI workloads. They were mainly used inside Google’s own data centres, but Alphabet has now expanded commercial sales so that selected outside customers can buy them, making them an alternative for companies seeking to reduce their dependence on Nvidia.
TPU vs GPU: What Is the Difference?
Graphics processing units, or GPUs, were first built for video games and graphics, but their ability to run many calculations at once made them the main hardware for training AI models. Nvidia leads this market, and demand for its chips has made it one of the world’s most valuable companies.
Tensor processing units, or TPUs, are chips that Google designed specifically for machine learning. Google first revealed them in 2016 and has used them to run services such as Search, Translate and its Gemini AI models. Because they are built for one type of work, TPUs can be more efficient for some AI tasks, while GPUs are more flexible.
For a company like OpenAI, using both types of chips can mean more capacity, better pricing power and less risk if one supplier cannot meet demand.
OpenAI’s Wider Chip Strategy
OpenAI is also working on its own hardware. In June 2026 it announced a custom AI chip designed with Broadcom, which it expects to deploy before the end of the year. Together with Nvidia GPUs and Google TPUs, this would give the company several sources of computing power.
The $600 Billion AI Spending Race
Alphabet, Amazon and Microsoft are together expected to spend as much as $600 billion on AI infrastructure in 2026. Alphabet’s finance chief Anat Ashkenazi has said the company expects to book a small part of its TPU sales revenue by the end of 2026, with most of it coming in 2027.
The huge spending has also raised questions among investors about how quickly it will produce returns, as AI companies continue to build larger data centres and sign long-term chip deals.
How the OpenAI Google TPU Shift Affects You
- ChatGPT users: more computing capacity can help keep services stable as demand grows, including in markets like Pakistan.
- Developers and startups: more competition between chip makers could, over time, lower the cost of running AI tools in the cloud.
- Tech industry: Nvidia still dominates AI chips, but big customers spreading their orders could slowly change the market.
- Investors: chip and cloud companies are being watched closely for signs that AI spending is paying off.
What Happens Next
Further details of the OpenAI Google TPU arrangement, including its size and timing, are expected once the companies confirm the deal. OpenAI’s custom Broadcom chip rollout later this year will also show how far the company can reduce its reliance on any single chip supplier.
For official updates, visit the Google Cloud TPU website.
Read More: Pakistan Climbs to 95th in Global Innovation Index 2026, Regains Overperformer Status
Follow the TaazaTaren channel for the latest updates.
