Anthropic Reveals Computing and Infrastructure Commitments of Up to $567 Billion
Key Facts
Anthropic’s offering prospectus disclosed future computing and infrastructure commitments ranging from $542 billion to $567 billion, putting the securing of computing capacity at the heart of the company’s expansion ahead of its public offering. The figures include existing commitments, agreements and additional contracted capacity, and do not simply represent an estimated spending budget available for use. Fortune reported that computing agreements signed as of December 31, 2025, totaled $417 billion with various providers. The terms of those agreements begin in 2026 and 2027, and run from one to 20 years. They also include additional capacity commitments beginning between 2028 and 2031, valued at between $125 billion and $150 billion. The figures underscore the financing and operating infrastructure required to expand the company’s models.
The long-term commitments coincide with rapid revenue growth, but the prospectus also records large losses in the first half of 2026. Revenue was $16.286 billion for the six months ended June 30, 2026, compared with $1.163 billion in the same period of 2025. Net loss was $80.748 billion, versus $12.969 billion a year earlier. At the same time, the company moved to adjusted operating income of $1.6 billion, after an adjusted operating loss of $1.8 billion in the first half of 2025. On a quarterly basis, second-quarter 2026 revenue was about $11.6 billion, while computing cost the company $5.6 billion and was its largest operating expense. The figures reflect the contrast between improved adjusted operating performance and a widening net loss, as computing costs continued to absorb a large share of quarterly revenue.
The “Anthropic Direct” initiative explains part of the company’s strategy for securing computing capacity while it remains reliant on external partners. Those partners will secure sites, power and facilities and manage data centers, while Anthropic leases computing capacity from them. The company says the initiative will make up an increasing share of its computing investments over time. Around September 30, 2026, Anthropic began entering into purchase commitments and lease agreements related to the initiative. Lease payments over five years total $125.2 billion, excluding interest. The company describes this approach as potentially reducing its reliance on competitors and delivering long-term economic advantages, but it does not make Anthropic solely responsible for building and operating the infrastructure.
The company’s relationships with computing providers continue to combine commercial reliance and competition. Anthropic’s sources include Amazon, Google and Microsoft, as well as agreements with SpaceX that give it access to up to $120.5 billion in computing capacity through 2029. According to Reuters, 47% of the company’s sales in 2025 passed through cloud providers Amazon and Google. The two companies are also investors, computing suppliers and AI competitors. The offering prospectus warned that cloud providers, as competitors, may “prioritize their own computing needs.” The company said its proprietary orchestration layer distributes workloads across AWS Trainium chips, Google TPU units, Nvidia processors and AMD processors, helping cut costs and reduce reliance on a single supplier.
Commitment data show how quickly the scale of pledges accelerated over a short period. Non-cancelable hosting and computing commitments totaled $54.6 billion at the end of 2025, before total long-term commitments exceeded $417 billion by early 2026, according to Reuters. Those commitments cover 3.5 gigawatts of dedicated computing capacity. Fortune reported potential infrastructure funding in 2026, including $42 billion from Broadcom, $20 billion from Google and $15 billion from Amazon. These figures link computing needs to sources of financing in which suppliers also act as investors. Execution of the pledges remains tied to the availability of sites, power and equipment, and the company’s ability to meet its financial commitments.
Anthropic warned in its offering prospectus that direct infrastructure expansion could result in significant cash outflows affecting liquidity, financial condition and operating results. It said building computing capacity would change its cash-flow profile and represent a significant use of net cash in operating and investing activities “for the foreseeable future.” It also identified execution risks including power delivery, obtaining permits, legal challenges, supply-chain disruptions, and shortages or delays in chips and data-center equipment. These risks add to lease and purchase commitments that extend over years. At the same time, a public offering could provide resources to accelerate expansion, while the offering’s value, size and share price remain undetermined, according to an Associated Press report. The agency said the stock would be listed on the Nasdaq Global Select Market under the ticker ANT.
The offering prospectus comes as investors watch whether AI companies can finance infrastructure amid rising borrowing costs and market volatility. A strategist quoted by Reuters said Anthropic’s offering “will set a benchmark for valuing leading AI labs” and could affect future offerings and the confidence of companies financing infrastructure. After the prospectus was announced, U.S. stock index futures briefly rose, and the S&P 500 gained 0.5% in morning trading on October 26, 2026. The Associated Press reported that the disclosure came weeks after Chief Executive Dario Amodei called for slowing the development of the most advanced AI models, while the offering could give the company resources to accelerate its expansion. This follows SpaceX’s June 2026 offering, which raised $86 billion and set a record, a potential benchmark for any large Anthropic offering.