The Growing Race for AI Computing Power
The global artificial intelligence industry is facing a critical challenge: securing enough computing power to develop, train and operate increasingly advanced AI models. Anthropic, the company behind the Claude chatbot and Claude Code, is moving aggressively to secure this essential resource as demand for its AI products continues to increase.
The company has entered into dozens of major agreements with technology giants, data centre operators, chipmakers and infrastructure providers. Its objective is to ensure that it has enough computing capacity to support future AI development while meeting the growing demand from businesses and individual users.
Computing power has become one of the most valuable resources in the AI industry because advanced models require enormous amounts of specialised hardware, electricity and data centre infrastructure.
Anthropic’s Massive Infrastructure Commitments
According to information disclosed in its public offering prospectus, Anthropic has committed approximately $895.4 billion through 2045 to computing power and chip-related agreements. These long-term commitments highlight the scale of investment required to remain competitive in the rapidly expanding AI market.
The company is expected to have access to more than 3.5 gigawatts of computing power in 2026, with plans to approximately double that capacity in the following year, according to people familiar with its plans.
Anthropic’s infrastructure strategy includes agreements with some of the world’s largest technology companies:
- Google: Approximately $121.3 billion in computing commitments.
- Amazon: Approximately $110 billion in commitments.
- Microsoft: Approximately $44.9 billion in commitments.
- SpaceX: Approximately $120.5 billion in commitments through 2036.
These agreements are designed to provide the computing resources required to train advanced AI systems and operate products such as Claude and Claude Code.
Partnerships Beyond Traditional Cloud Providers
Anthropic is also looking beyond major cloud computing companies to secure additional capacity.
The company has partnered with specialised infrastructure providers, commonly known as neocloud companies, including Fluidstack and Lambda. These businesses focus on providing high-performance computing resources for AI workloads.
It is also working with former cryptocurrency mining companies such as TeraWulf and Hut 8 to develop data centre facilities.
These partnerships offer an alternative route to expanding computing capacity. Some former crypto-mining operators already have access to land, electricity connections and infrastructure that can potentially be adapted for AI data centres.
Anthropic is also working directly with chip manufacturers, including Nvidia, AMD and Broadcom, while investing in supporting equipment such as cooling systems to accelerate data centre deployment.
Why Computing Capacity Has Become a Major Challenge
The rapid adoption of generative AI, coding assistants and autonomous AI agents has increased demand for specialised processors and high-performance data centres.
Developing advanced AI models requires powerful chips that can process enormous volumes of information. Operating those models at scale requires even more resources because millions of users and businesses may access AI services simultaneously.
Anthropic has already experienced capacity constraints as the popularity of Claude Code and other products has increased. Limited computing resources have affected product availability and contributed to changes in usage limits and pricing.
The challenge is not limited to purchasing chips. Companies must also secure suitable locations, electricity supplies, cooling equipment, construction teams and connections to power grids. These requirements can make data centre development expensive and time-consuming.
The Financial Pressure Behind Anthropic’s Expansion
Securing computing power on this scale requires enormous financial resources. Anthropic’s long-term infrastructure commitments increase its need for capital as it attempts to expand its AI business.
The company is pursuing major financing arrangements and preparing for a potential public offering that could raise substantial funds. Its disclosed commitments demonstrate that the AI race is increasingly becoming a competition not only between models and software products but also between companies capable of financing large-scale infrastructure.
The financial challenge is particularly significant for AI companies that depend on external cloud providers rather than owning all their computing infrastructure.
As demand grows, these companies must balance infrastructure spending, product development, operating costs and the need to generate sustainable revenue.
What Anthropic’s Strategy Means for the AI Industry
Anthropic’s infrastructure push reflects a broader shift in the artificial intelligence industry. Access to advanced chips, computing facilities and reliable electricity is becoming just as important as developing better algorithms.
Companies that secure sufficient computing capacity may gain an advantage in launching new models, improving performance and serving larger customer bases. Those unable to obtain enough infrastructure could face delays, higher operating costs and limitations on product availability.
The competition is also creating opportunities for semiconductor manufacturers, data centre developers, cloud computing providers and companies specialising in power and cooling technologies.
As AI adoption continues to expand, computing infrastructure is likely to remain one of the industry’s most important strategic assets.
