Anthropic has stated that Google is giving them $10 billion in cash (and valuing them at $350 billion, the same as a funding round in February). They could get an additional $30 billion if they reach certain goals, and will get a lot of help increasing their computing resources.
Deal terms and why they matter
Anthropic says that Google Cloud will supply 5 gigawatts of computing power over the next five years, and possibly even more. That’s a huge amount of support for developing and operating large AI programs. To give you an idea of scale, one gigawatt can power roughly 750,000 homes in the US at any one time.
Anthropic has been getting more funding because lots of people are now using Claude Code, an AI program that speeds up software creation. Earlier this week, Anthropic told us they got $5 billion from Amazon (again at a $350 billion value), and Amazon may give them up to $20 billion more in the future.
Anthropic’s momentum and fresh capital
In February, Anthropic raised $30 billion, and since then people who invest have wanted to invest in Anthropic with a value of $800 billion or more. The company also says Cowork, an AI program designed to do a wider range of jobs without needing to write code, is growing rapidly.
Google’s TPUs (tensor processing units) are an important alternative to Nvidia’s chips and are hard to get in the AI world. Anthropic is a major buyer of Google’s chips and cloud services, and these are areas Google is focusing on as their main source of money from advertising starts to change.
Here are the headline deal points, as stated by the companies and the startup:
– $10 billion from Google now at $350 billion valuation
– Up to $30 billion more tied to targets
– 5 gigawatts of compute from Google Cloud
– Option for further gigawatts if needed
Chips, capacity, and a deepening cloud tie-up
This new deal builds on an agreement made earlier this month between Anthropic, Google and Broadcom. Last year, Google said they would give Anthropic up to 1 million TPU chips (worth tens of billions of dollars) after investing about $3 billion in the company.
Despite all the money changing hands, the competition between the two is very clear. They are both trying to create AI that can do as much as people can, and to get businesses to use it. People who work or used to work at Google say Google’s leaders are becoming more worried about their place in the AI coding market, where Anthropic is doing extremely well.
Partners and competitors in AI coding
Claude Code has quickly become a favorite tool for software developers in Silicon Valley, and even at Google itself. Anthropic was started in 2021 by Dario Amodei and people who previously worked at OpenAI, and they still have a strong connection to Google, where Amodei used to be an AI researcher.
Anthropic is thinking about offering shares to the public as soon as October, as they look for the infrastructure they need to deal with demand. They are also fighting against the Department of Defense calling them a risk in the supply chain, after disagreeing publicly about whether their technology could be used by the military.
Risks, scrutiny, and what to watch next
Some financial experts are worried about “circular” deals in AI, where the companies that provide cloud computing and chips invest in AI startups, and then sell those startups the computing power they need. This criticism has been made about deals involving Anthropic and other AI companies.
What happens from here depends on Anthropic achieving the goals that will get them more money and computing power from Google. If Anthropic continues to improve their products and get more of the necessary infrastructure, this expanded partnership could completely change the way AI chips, cloud services, and tools for developers compete.
Key areas of scrutiny highlighted in the source include:
– Pentagon supply-chain risk designation
– Concerns about circular financing and capacity sales
– Intensifying competition in AI coding
