AI基建竞赛再升温，Anthropic拟斥资500亿美元建数据中心

Group 1 - Anthropic, supported by Amazon, plans to invest $50 billion in building AI infrastructure in the U.S., starting with customized data centers in Texas and New York, with the first facilities expected to be operational by 2026 [1] - Anthropic has partnered with Fluidstack to develop AI infrastructure that supports rapid business growth and long-term research plans, leveraging Fluidstack's ability to provide gigawatt-level power supply quickly [1] - The investment positions Anthropic as a core player in U.S. AI infrastructure, with CEO Dario Amodei emphasizing the need for robust infrastructure to accelerate scientific discovery and create jobs [1] Group 2 - Microsoft announced the construction of its first AI super factory in Atlanta, Georgia, which will connect with another data center in Wisconsin to form a large supercomputer powered by hundreds of thousands of NVIDIA chips [3] - The Fairwater AI data centers will be interconnected, enabling high-speed data flow and collaborative operation across different states, significantly reducing the time required for training new AI models [3] Group 3 - Amazon has established a dedicated data center park for Anthropic in Indiana, costing $11 billion, which is now operational [4] - Competitors like OpenAI are also aggressively pursuing large-scale infrastructure plans, with OpenAI securing over $1.4 trillion in long-term investment commitments from major tech companies [4] - Concerns have been raised about the feasibility of these massive expenditures and whether the U.S. has the industrial capacity to fulfill such commitments, leading to questions about a potential bubble in the AI sector [4] Group 4 - The role of the U.S. government in funding AI infrastructure has become a contentious issue, highlighted by OpenAI's request to expand the scope of a key tax credit under the CHIPS Act to include AI data centers [5] - OpenAI's previous proposal for government backing of its computing partnerships was withdrawn after facing opposition, underscoring the political and financial uncertainties surrounding AI infrastructure funding [5]