Analysis
OpenAI plans to buy Google's AI chips, The Information reported Tuesday, in a shift that would reduce OpenAI's dependence on Nvidia, the company that supplies most of the hardware behind the AI boom, according to a Reuters wire report carried by Investing.com. The report did not disclose financial terms, chip volumes, or a delivery timeline, and neither company confirmed it: Google did not immediately respond to Reuters' request for comment, and OpenAI declined to comment.
OpenAI's compute shopping spree
The Google TPU report lands weeks after OpenAI closed its latest financing round, giving the company an $852 billion post-money valuation, with SoftBank co-leading alongside strategic backers Amazon, Nvidia, and Microsoft. Per Yahoo Finance:
- Total committed capital — $122 billion.
- Amazon — committed $50 billion, the round's largest single commitment.
- Nvidia and SoftBank — each committed $30 billion.
“Per Yahoo Finance: - Total committed capital — $122 billion.”
Separately, per Decrypt:
- Nvidia — signed a letter of intent to invest up to $100 billion in OpenAI tied to chip deployment commitments.
- AMD — gave OpenAI the right to buy up to 160 million AMD shares, tied to OpenAI deploying 6 gigawatts of AMD chips.
In June, OpenAI also unveiled its first custom chip, code-named Jalapeño, built with Broadcom and planned for initial deployment by the end of 2026.
Why Google chips, and why now
OpenAI remains one of Nvidia's biggest customers, and a move toward Google's in-house Tensor Processing Units would mark one of the clearest signs yet that leading AI labs don't want a single hardware supplier controlling their compute roadmap. It also puts OpenAI in a position its chief rival already occupies: Anthropic signed its own multi-gigawatt compute agreement with Google and Broadcom in April, part of roughly $138 billion in debt-financed compute deals Anthropic has lined up since September alone. If OpenAI follows a similar path, it would confirm that Google's TPU business, not just Nvidia's GPUs, is becoming a second must-have compute source for frontier labs.
The numbers in context
Big tech's own capital commitments dwarf any single chip deal: Amazon, Microsoft and Google parent Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure, per the same Reuters report. Against that backdrop, even a large OpenAI-Google chip agreement would be one more entrant in a market already defined by hundred-billion-dollar compute contracts — Anthropic's own compute obligations alone run into the hundreds of billions once new and expanded deals are counted.
What the headline misses
None of this is confirmed. The Information's report is sourced to a single person familiar with the matter, financial terms are undisclosed, and both companies involved either declined to comment or didn't respond — the gap between "plans to buy" and an actual signed, priced agreement is wide, and past reports of AI chip deals, including Nvidia's own OpenAI letter of intent, have taken months to convert into disclosed terms.
What to watch next
Whether OpenAI or Google confirms the chip discussions, what volume of TPUs would actually change OpenAI's Nvidia dependence meaningfully, and whether Anthropic or other labs follow with their own non-Nvidia chip commitments.