Google shows off its semiconductor prowess with a new Anthropic deal
By Emily Bary and Christine Ji
Alphabet's cloud business secured a key AI partnership with Anthropic, signaling an increasingly competitive landscape for high-performance chips
The deal grants Anthropic access to up to 1 million of Google's tensor processing units to train and run its Claude large language model.
Alphabet Inc.'s cloud and chip businesses just got a big endorsement in the form of a refreshed agreement with artificial-intelligence startup Anthropic.
Under the terms of the newly announced deal, Anthropic will be able to access up to 1 million of Google's (GOOGL) (GOOG) tensor processing units, or TPUs, to train and run its Claude large-language model. The company, which is a rival to OpenAI, will also use various Google Cloud services.
In a release, Anthropic said it went with Google TPUs "due to their price performance and efficiency," as well as the company's past experience with the chips.
The two businesses previosuly struck an arrangement in 2023, and the expansion "will help us continue to grow the compute we need to define the frontier of AI," Anthropic Chief Financial Officer Krishna Rao said in a statement.
AI players have been hungry for computing power, as evidenced by OpenAI's recent deals with Advanced Micro Devices Inc. (AMD), Nvidia Corp. (NVDA) and Broadcom Inc. (AVGO) And as infrastructure costs ramp up, companies have been paying more attention to factors like a chip's performance relative to its price.
The Anthropic-Alphabet deal is a validation of Google's TPU business, which analysts see as a unique asset. TPUs use vertical power delivery to provide greater power efficiency than Nvidia's chips, which use lateral power delivery. Google partners with Broadcom for the design and manufacturing of the chips, although it's reported that Google could be bringing on Taiwan-based MediaTek Inc. (TW:2454) as a cheaper option.
Not only does Google use TPUs itself, but it sells them to others in the technology world. One D.A. Davidson analyst recently estimated that a combined spinoff of Google's TPU unit and DeepMind AI research lab could be worth $900 billion, though the analyst was doubtful the company would pursue such a move.
Read: Google may be sitting on a $900 billion gem that could disrupt Nvidia's dominance
Anthropic's partnership with Google comes after Anthropic and Nvidia publicly clashed over AI policy regarding U.S. export restrictions on chips in May of this year. Anthropic called for stricter controls to promote national-security interests, while Nvidia criticized Anthropic's stance and called for fewer regulations.
Alphabet's stock was up 1% in premarket trading on Friday, while Amazon.com Inc.'s stock (AMZN) was off fractionally. Both technology giants are Anthropic investors, but some on Wall Street have grown concerned about the competitiveness of Amazon's cloud and other services in the AI era.
Jefferies analyst Blayne Curtis wrote in a note last week that Anthropic's use of TPUs doesn't seem to bode well for Amazon's custom Trainium chip, while UBS analyst Karl Keirstead wrote in a note earlier this week that Anthropic could be using Google Cloud as its primary compute provider over Amazon Web Services.
"We remain committed to our partnership with Amazon, our primary training partner and cloud provider, and continue to work with the company on Project Rainier, a massive compute cluster with hundreds of thousands of AI chips across multiple U.S. data centers," Anthropic said in an announcement on Thursday.
"AWS continues to be Anthropic's primary training partner and cloud provider," an AWS spokesperson told MarketWatch.
Melissa Otto, head of research at S&P Global Visible Alpha, said she doesn't think the Anthropic-Google partnership is indicative of Amazon falling behind. "They're all talking to each other - that's the sort of market that we're in right now," Otto told MarketWatch prior to the announcement.
In an ecosystem where access to power and computing resources are major limiting factors, Otto expects to see many alliances. "Companies are trying to get as much compute as possible," she said.
Anthropic emphasized that its compute strategy uses a combination of Google's TPUs, Amazon's Trainium and Nvidia's GPUs.
Representatives for Google and Nvidia declined to comment.
Read: Here's what really matters for Amazon's stock - beyond the AWS outage
-Emily Bary -Christine Ji
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
10-24-25 0732ET
Copyright (c) 2025 Dow Jones & Company, Inc.The articles, information, and content displayed on this webpage may include materials prepared and provided by third parties. Such third-party content is offered for informational purposes only and is not endorsed, reviewed, or verified by Morningstar.
Morningstar makes no representations or warranties regarding the accuracy, completeness, timeliness, or reliability of any third-party content displayed on this site. The views and opinions expressed in third-party content are those of the respective authors and do not necessarily reflect the views of Morningstar, its affiliates, or employees.
Morningstar is not responsible for any errors, omissions, or delays in this content, nor for any actions taken in reliance thereon. Users are advised to exercise their own judgment and seek independent financial advice before making any decisions based on such content. The third-party providers of this content are not affiliated with Morningstar, and their inclusion on this site does not imply any form of partnership, agency, or endorsement.
Popular
4 Stocks to Buy for the Long-Term While They’re Still Undervalued
What’s Ahead for Stocks This Quarter—and Where to Invest
Are TIPS ETFs the Best Way to Fight Inflation?
10 Best Blue-Chip Stocks to Buy for the Long Term
