Ann Davis Vaughan

The scale of AI infrastructure investment, estimated at $15 trillion, is fundamentally changing how data centers are financed and built. Traditional financing models are being disrupted as developers and investors grapple with unprecedented capital requirements and construction velocity. Industry leaders are rethinking risk assessment and funding approaches to keep pace with the buildout.

Amazon, Microsoft, Oracle and other AI data center developers are increasing financial offers to municipalities and regulators to secure approval for new facilities, while also opposing utility proposals to pass infrastructure costs to consumers. The companies are moving beyond messaging campaigns to offer tangible community benefits. Local opposition from activists remains strong despite financial incentives, setting up years of regulatory and community battles ahead.

Four U.S. states have rolled back or paused data center tax incentives, while nine others are considering repeal measures, according to the National Conference of State Legislatures. The moves could raise equipment costs by 7% or more, adding several billion dollars to the price of every gigawatt of AI computing. States that previously courted data center investment are now reversing course in little-noticed legislative actions that accelerated this summer.

Crusoe Energy, the data center developer building OpenAI's Stargate supercomputer facility in Abilene, Texas, is facing higher-than-expected costs and technical challenges integrating natural gas turbines with the AI infrastructure. Engineers have been working overtime to resolve compatibility issues between the power generation system and one of the most expensive AI supercomputers ever built. The project, part of OpenAI's broader Stargate computing initiative, is encountering obstacles that were not anticipated during initial planning.

Fervo Energy, a leader in enhanced geothermal systems, raised nearly $2 billion in its initial public offering. The funding positions the company to scale production and compete directly with natural gas as a power source for AI data centers. Enhanced geothermal technology, which uses drilling techniques borrowed from oil and gas to access heat from the earth, has become a focal point for solving the energy demands of large-scale AI infrastructure.

GridCare, a Stanford-backed startup, claims to have identified substantial hidden power capacity in existing U.S. grids by using AI to analyze siloed utility data and run simulations. The company argues that American power networks operate at only 30% utilization on average and could safely reach 60% through software optimization alone, potentially unlocking 300 gigawatts of already-paid-for capacity. GridCare has already helped enterprise AI services provider Gruve locate 150 megawatts in congested markets like Santa Clara and Seattle. The startup raised $13.5 million last year and is announcing a new funding round, but faces the historical challenge of convincing risk-averse utilities to adopt efficiency measures when they profit from building new capacity.