DigestAI news desk

Cut through the AI noise.

Hardware & Compute4 min read

Google expands TPU sales to external customers

Google has begun selling its proprietary Tensor Processing Unit (TPU) chips outside of its own cloud, partnering with data‑center operators such as Fluidstack and Oracle. The move follows a decade of internal development that started in 2015 and was opened to external users in 2018.

1 source

Key points

  • Google opens TPU sales to external customers via Fluidstack and Oracle
  • OpenAI, Meta, Anthropic are buying TPUs to diversify GPU supply
  • TPU revenue projected to reach $138 billion by 2028, 34 % of Google Cloud

Major AI firms are already adopting the chips.

Google’s CFO Anat Ashkenazi said TPU revenue was recorded in the latest quarter and is expected to explode from 2027 onward. Analysts project standalone TPU sales of $9.4 billion in the second half of 2026, rising to $138 billion by 2028, which would represent about 34 % of Google Cloud’s revenue. Nvidia, which holds roughly 80 % of the AI accelerator market, is countering with a $3 billion investment in Meta’s $97.5 billion stock offering to lock in customers.

The story so far

5 episodes →
  1. Google expands TPU sales to external customersthis story
Full story from note.com · by whereverjp · via Search: NVIDIAOpen source ↗

[2026 Latest] Google's Proprietary 'TPU' Chip is Transforming the AI Semiconductor Market | Scenarios for the Collapse of Nvidia's Monopoly and the Intentions of Tech Giants_927

note.com · 4 November 2026

Loading the full article…

This text was published by note.com and written by whereverjp. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page
GoogleNvidiaOpenAIMetaAnthropicOracleAnat AshkenaziJohn Blackledge

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories