Editorial illustration for OpenAI Reportedly Plans to Purchase Google AI Chips

OpenAI Reportedly Plans to Purchase Google AI Chips

By AI Daily Post Edited by Brian Petersen, Editor-in-Chief

October 28, 2026 • 3 min read

OpenAI is in talks to buy AI chips from Google, according to a report from The Information, a move that would mark a notable shift away from its near-total dependence on Nvidia hardware. The chips in question are Google's Tensor Processing Units, custom-built processors that Google has spent years developing as an alternative to Nvidia's GPUs for training and running AI models. No pricing or volume details have surfaced yet.

The timing stands out. OpenAI has long counted as one of Nvidia's biggest customers, and Nvidia has poured billions into OpenAI as an investor. At the same time, OpenAI and Google are rivals across search and consumer AI products, which makes a hardware purchase from Google a strange wrinkle in that competition.

OpenAI already started renting TPUs through Google Cloud in 2025, after first renting Google chips the year before, and the company has been hiring engineers specifically to tune its models for TPU performance. That hiring push suggests the technical groundwork for a direct purchase has been underway for some time, not a sudden decision.

OpenAI plans to buy AI chips from Google, potentially reducing its reliance on NVIDIA as it expands computing capacity to train and run its AI models, according to a report by The Information.

OpenAI Plans to Purchase Google TPUs: Report - Analytics India Magazine

Why this matters

For developers and founders building on OpenAI's models, this is a reminder that the infrastructure beneath the products you rely on is shifting fast, and not always in ways that get announced with fanfare. OpenAI hedging away from pure Nvidia dependence while Nvidia simultaneously writes it billion-dollar checks is an odd arrangement, and it tells us compute scarcity still dictates terms more than loyalty or rivalry does. Google selling TPUs to a company it competes with in search and chatbots is the more telling detail here.

That's not friendship, it's Google's cloud and chip division deciding revenue from a hot customer matters more than protecting market position elsewhere. For researchers, the practical question is whether TPU-trained or TPU-served OpenAI models behave any differently, in latency, cost, or availability, once this migration happens. For anyone planning infrastructure spend around a single vendor's roadmap, the lesson is to watch what the biggest labs actually buy, not what they say about competitors.

OpenAI hasn't confirmed terms, and that opacity itself is worth tracking as this deal, if real, gets finalized.