Anthropic proposes to use U.S. government land to build large-scale AI data centers

Anthropic, creator of the IA Claude model, released a document proposing the use of federal lands of the Departments of Defense (DoD) and Energy (DOE) to build data centers dedicated to the training of large-scale artificial intelligence. The initiative seeks to accelerate the infrastructure of AI in the United States, suggesting programmatic environmental reviews and public-private partnerships to enable the rapid implementation of data centers.

Anthropic's proposal for AI data centers in federal lands

The document, entitled “Building AI Infrastructure in the United States”, recommends that the federal government provide land under the jurisdiction of the DoD and the DEO for the construction of AI data centers. The proposal includes programmatic environmental reviews — a process that analyses impacts in an aggregate and early manner rather than time-consuming individual assessments — reducing bureaucratic barriers. In addition, Anthropic advocates a public-private investment model, in which the government provides the land and part of the funding, while technology companies bear the costs of construction and operation.

Importance of federal land in the development of AI data centers

The scale needed to train border AI models requires huge amounts of energy, water and physical space. The acquisition of private land for data centers faces obstacles such as zoning, local opposition and long environmental licensing processes. By using land already owned by the federal government — many of them in remote areas with access to clean energy sources — the proposal eliminates stages of land negotiation and accelerates deployment. AI demand follows 'almost unlimited' even with companies demanding more return on investment, which makes agility a critical competitive factor for the US in relation to China and other nations.

Impact of the AI data centers proposal on the artificial intelligence sector

If implemented, the measure can significantly reduce the time to build data centers — currently a bottleneck for companies like Anthropic, OpenAI and Google. The use of programmatic environmental reviews can cut off years of schedule. The proposal also signals an alignment between government and large AI players, potentially directing public resources to an infrastructure that will benefit the private sector. However, critics may question the granting of federal land to corporations and the environmental risks associated with massive energy consumption.

Connection to the global AI data center infrastructure

Other countries, such as the United Kingdom and Japan, have announced plans to build AI data centers with government support. Anthropic's proposal puts the US into this global infrastructure race. The company points out that the ability to train ever-increasing models is vital to American leadership in technology and national security — an argument that can gain weight in an election year.

The full document is available on the Anthropic website. The initiative reflects the increasing pressure for AI infrastructure, also echoing the AI race that exchanges giant models for smart and cheap systems, while demanding colossal physical investments.

Frequently asked questions about AI data centers in government lands

Has this proposal been approved?

No, this is an Anthropic suggestion document to the American government; there is still no sign of formal adoption.

What are the environmental benefits?

Program reviews can identify cumulative impacts and plan mitigation more efficiently, but the main gain is licensing speed.

Would Anthropic have exclusive access to the land?

The document does not mention exclusivity; the proposal is for land to be made available to AI companies in general, under competitive criteria.