The United States is experiencing an unprecedented boom in data center construction as technology companies race to build the infrastructure required to power artificial intelligence workloads. Global data center dealmaking surged to $61 billion in 2025, with the Stargate Project alone committing $500 billion to build next-generation AI facilities. The construction frenzy has transformed the competitive landscape, created supply shortages in key markets, and raised critical questions about energy infrastructure capacity.
How Much Capital Is Flowing Into Data Center Construction?
According to S&P Global, more than $61 billion flowed into the global data center market in 2025, slightly higher than the $60.8 billion invested last year. US spending reached a record $40 billion in June alone, representing a 30 percent increase from the previous year according to Bank of America Institute research.
The Stargate Project, a joint venture between OpenAI, Oracle, and SoftBank, has emerged as the most ambitious initiative. The partners announced in January plans to invest $500 billion building a new generation of AI-ready data centers across the country.
By September, OpenAI, Oracle, and SoftBank expanded Stargate with five new sites bringing planned capacity to nearly 7 gigawatts and investment exceeding $400 billion over the next three years. The expansion puts the project on track to secure its full commitment by the end of 2025, ahead of schedule.
Debt issuance nearly doubled to $182 billion in 2025, up from $92 billion in 2024. Meta raised $62 billion in debt since 2022, with nearly half issued this year alone. Google and Amazon raised $29 billion and $15 billion respectively.
Which Major Projects Are Under Development?
Vantage Data Centers announced plans to develop a mega-scale 1.4 gigawatt campus in Shackelford County, Texas. The “Frontier” project requires $25 billion in investment and will employ over 5,000 individuals during construction. Once complete, it will rank among the largest single-site data center capacities globally.
Amazon announced a $10 billion investment for a high-tech cloud computing and AI campus in Richmond County, North Carolina. The facility will contain 20 buildings at full build-out, each spanning more than 200,000 square feet.
Microsoft announced what it calls the first AI superfactory in Atlanta, linking multiple facilities within its Fairwater project with hundreds of thousands of advanced GPUs running AI workloads. The company has pledged over $60 billion to neocloud data center companies.
Anthropic announced plans to spend $50 billion on US AI infrastructure, starting with custom data centers in Texas and New York developed in partnership with Fluidstack. The first locations are expected to go live in 2026.
What Challenges Do Developers Face?
Power availability has become the dominant factor in data center site selection, surpassing traditional considerations such as proximity to metropolitan areas. Grid capacity constraints are forcing developers to seek non-traditional locations with abundant electricity.
The North American data center sector faces a deepening supply crunch. According to JLL’s midyear report, colocation vacancy has fallen to just 2.3 percent even as installed capacity rose to 15.5 gigawatts. Northern Virginia continues dominating with 5.6 gigawatts of capacity.
Water consumption represents an emerging concern as AI workloads generate intense heat requiring cooling systems. Data centers in water-scarce regions face increasing scrutiny from regulators and investors.
Public acceptance remains mixed. Industry surveys found that while 93 percent of Americans recognize the importance of AI data centers, only 35 percent support construction in their communities. Concerns center on environmental impact, energy consumption, and land use.
How Is the Energy Challenge Being Addressed?
Interest in nuclear-powered data centers and small modular reactors has grown substantially. Texas Nuclear Ventures signed agreements to bring four large reactors to the state, while Blue Energy announced plans for a power plant southwest of Houston supplying up to 1.5 gigawatts to Crusoe Energy Systems’ data center.
The US Department of Energy is opening its Oak Ridge Reservation in Tennessee for private development of AI data centers with on-site power generation, marking the first step toward using federal land for AI infrastructure.
Pennsylvania unveiled a $70 billion initiative to attract major data center investments, with key sites adjacent to nuclear power plants providing direct power connections. Amazon announced a $20 billion Pennsylvania investment to develop multiple AI and cloud infrastructure campuses.
Google’s head of infrastructure told employees the company must double AI serving capacity every six months to meet intense demand, highlighting the extraordinary growth trajectory the industry is attempting to maintain.
What Does This Mean for the US Economy?
The data center construction boom is creating significant employment opportunities. Amazon’s North Carolina project alone is expected to add substantial jobs during construction and operations. Vantage’s Texas development will employ over 5,000 workers.
JLL forecasts up to $1 trillion in North American data center investment from 2025 through 2030. The pace of growth in the United States is leaving Europe “in the dust” according to ING research, which predicts US data center investment could be fivefold higher.
Hyperscalers are increasingly working with AI labs to finance construction in arrangements that underscore the massive capital requirements. The competitive dynamic among frontier AI model providers is changing quickly, driving sustained investment even as some investors question valuations.
Federal policies including the ‘One Big Beautiful Bill’ Act and AI Action Plan provide supporting frameworks but are unlikely to significantly accelerate growth that is primarily driven by private investment. Long-term success depends on balancing innovation, regulation, and infrastructure development while ensuring the grid can support increasingly power-hungry AI workloads.