Anthropic signs a $10 billion computing power deal to continuously expand AI infrastructure.

According to informed sources, the artificial intelligence startup Anthropic has signed a $10 billion power collaboration agreement with AI infrastructure startup Volta Infra Holdings Ltd. to meet the growing computational power demands of its Claude model. This marks the latest move by Anthropic in recent months to expand its AI infrastructure layout. Sources indicated that Anthropic has contracted to use data centers operated by Volta. Earlier on Tuesday, Volta announced that it has reached a $10 billion collaboration agreement with an unnamed AI laboratory and will work with Bitcoin mining company Bitdeer Technologies (BTDR.US) to provide computational power services from its data center in Norway, with a contract duration of six years. However, Volta CEO Ricard Boada declined to disclose the identity of the client, and both Anthropic and Bitdeer did not comment on the matter. Boosted by the news, Bitdeer's stock price rose by over 8.5%. As more companies and consumers use Claude for programming and other AI tasks, Anthropic has been steadily increasing its investment in computational power in recent months to ensure it can meet the rapidly growing market demand. Previously, the company has signed computational power collaboration agreements with Elon Musk's SpaceX (SPCX.US), AMD (AMD.US), and Akamai (AKAM.US). Bloomberg also reported earlier that Anthropic is in talks with Meta (META.US) to rent data center computational resources. Earlier this year, Anthropic completed a $65 billion financing round to support the massive capital requirements for AI model development. At the same time, the company is also considering launching an initial public offering as early as this year to raise further funds through the capital markets. According to Voltas disclosure, the Norwegian data center involved in this collaboration will provide 133 megawatts of computational power and deploy NVIDIA Corporation (NVDA.US)'s latest generation Vera Rubin AI chips. Volta was established this January by former Brookfield Asset Management executives, primarily focusing on providing computational power leasing to AI companies and assisting clients in financing the procurement of expensive AI chips. The company also announced on Tuesday that it has secured $300 million in venture capital financing, bringing its latest valuation to $2.4 billion. As a partner in this collaboration, Bitdeer is also one of the representatives of Bitcoin mining companies transitioning to AI infrastructure in recent years. As Bitcoin prices remain low, more mining companies are beginning to convert some of their data centers from cryptocurrency mining to AI computational power services. Bitdeer previously indicated plans to transform some of its facilities in Texas, Tennessee, and Washington into AI data centers. In fact, both Anthropic and its competitor OpenAI are increasingly relying on large tech companies and emerging cloud computing enterprises to provide data center resources necessary to meet the substantial computational power needs for advanced AI model development and large-scale commercialization. OpenAI CEO Sam Altman has previously stated that the company plans to invest trillions of dollars in AI infrastructure in the future. Currently, OpenAI is planning to build a new data center that could cost over $30 billion, and according to prior reports by Bloomberg, NVIDIA Corporation is also in discussions with OpenAI to support its rental of an AI data center campus in Ohio, which is constructed by SoftBank Group, with a total investment of about $500 billion and a capacity of 10 gigawatts. However, over the past year, some financing and computational power collaboration models between Anthropic and OpenAI have attracted market attention. Some analysts believe that these financing and computational power partnerships exhibit certain characteristics of circular financing, creating an increasingly complex relationship of financial and business dependencies among AI developers, chip manufacturers, and infrastructure providers. If future AI demand does not meet the current high market expectations, this highly interconnected collaboration model could further amplify operational risks across the entire industry chain.