November 3, 2026, (Inside AI) — Anthropic has committed $895.4 billion through 2045 to computing power and chip deals it largely cannot back out of, according to its public offering prospectus. The disclosure comes as the San Francisco AI lab prepares what could be the largest initial public offering in history, expected to raise $100 billion this month.
The company's scramble for compute began late last year when executives realized they were running out of capacity to train and serve their models. That shortage forced Anthropic to limit free use of its products and raise prices for some customers after its Claude Code coding tool became a hit. The company now says its ability to stay at the AI frontier depends substantially on securing compute at scale.
Anthropic's predicament illustrates a widening divide in the AI industry. Giants like Amazon, Google, Microsoft, and Meta have deep pockets and credit histories to build their own data centers. Younger labs like Anthropic and OpenAI must rely on others for the most vital resource in the AI boom. Anthropic's response has been a deal spree spanning dozens of contracts with former crypto miners, data center startups, chipmakers, and tech giants.
The company has committed $121.3 billion to Google, $110 billion to Amazon, $44.9 billion to Microsoft, and $120.5 billion to SpaceX through 2036. It also struck a $50 billion deal with neocloud provider Fluidstack and a $41.3 billion agreement with Lambda. This year, Anthropic expects to have more than 3.5 gigawatts of computing power, enough electricity to power all the homes in Chicago. Next year, it aims to double that amount.
"Anthropic is gobbling up a lot of the capacity," said Erik Bernhardsson, CEO of Modal, a New York company that sells computing power.
The financial burden is immense. Anthropic is raising debt financing, including $63 billion last week from Wall Street banks and private credit firms. Even so, the company has warned it may need additional financing sooner than anticipated. The IPO proceeds are meant to fund the compute buildout, but the commitments extend decades beyond the offering.
Tom Brown, Anthropic's chief compute officer, has spearheaded efforts to diversify beyond Nvidia's AI chips, using chips from Amazon, Google, and Advanced Micro Devices. Finding enough computing power is "very clearly the bottleneck for all of our progress," Brown said at a Group of 20 event in September. "We're in a huge shortage of the power and labor needed to support the demand for AI," he said.
Scavenging Power Lines and Poaching Electricians
Anthropic's strategy goes beyond signing checks. The company is helping design data centers that other firms will build and operate. It has hired construction and supply chain experts, bought chips directly from Nvidia, AMD, and Broadcom, and purchased cooling systems so operators can install components faster. It met with developers in Texas, Indiana, and other states about building data centers, asking about permits and grid connections.
The efforts have hit hurdles. Fluidstack pushed back against some Anthropic data center designs over concerns about power fluctuations that could cut off electricity and damage equipment, before reaching a solution. In upstate New York, TeraWulf competed for electricians with the owners of the Buffalo Bills' football stadium. Anthropic and TeraWulf ultimately asked the local electricians union for names of traveling electricians from other areas, even though it would make the project more expensive.
"We operate as one team," said Asher Genoot, CEO of Hut 8, a former crypto miner working with Anthropic.
Anthropic has started considering sites with at least 200 megawatts, enough electricity for about 140,000 U.S. homes. That is less power than the company originally wanted but a faster way to get what it needs. It has since become open to data centers as small as 50 megawatts and would supplement the sites with batteries and natural gas.
John Tough, managing partner of Energize Capital, which invests in data center technology, said Anthropic did not control the pipelines of computing power, so it might have to settle for building "gigawatts, megawatts at a time." "The next best thing is to scavenge the world to figure out where there's power lines," he said.
Anthropic's compute crunch reflects a broader shift in AI economics. The company's $895.4 billion in commitments dwarfs its current revenue and underscores how capital intensity has become the defining feature of frontier AI. Unlike software businesses of the past, AI labs must secure physical infrastructure, long-term energy contracts, and specialized labor before they can serve a single customer. That reality is forcing Anthropic to go public not just to fund growth but to service debt taken on to secure compute.
Read: Nvidia in Talks to Back Nscale's Anthropic Data Center with Billions in Credit Guarantees
The company declined to comment on the record before its IPO. Its prospectus devotes more than 10 pages to the stakes of securing compute, a sign of how central the issue is to its investment thesis. Anthropic was founded in 2021 and initially turned to Amazon and Google, which invested a combined $6 billion in 2023, for computing power. It then spent much of that money on their cloud services.
As the IPO approaches, Anthropic's ability to deliver on its compute promises will be tested by construction timelines, power availability, and competition for the same resources from OpenAI and other labs. The company's bet is that securing megawatts today will translate into model superiority tomorrow. Whether public markets will tolerate such heavy upfront spending remains an open question.
If a research paper dropped this week, Mahesh has already read it. He covers AI tools and vibe coding for builders who need to know what is worth their time and what is not.
