A growing backlash against data center construction across the United States is threatening to slow the artificial intelligence infrastructure boom, as local communities push back against the power demands and environmental impact of facilities that can each consume as much electricity as a nuclear reactor.
"Communities are waking up to what a 1-gigawatt data center means for their water, their grid and their quality of life," said Sergio Toro, founder of Aterio, a data provider that tracks data center construction through satellite imagery and utility filings. "The era of building these facilities with minimal local scrutiny is over."
The opposition comes as the four largest cloud providers — Amazon.com Inc., Microsoft Corp., Google parent Alphabet Inc. and Meta Platforms Inc. — race to secure power for an estimated $300 billion in planned data center capacity. Amazon's self-built US data centers already consume up to 9 GW of power, comparable to the generation capacity of North Dakota, according to Aterio. Google and Microsoft each operate about 5 GW of self-built capacity, while Meta runs roughly 4 GW.
At the center of the conflict is Meta's "Prometheus" data center in New Albany, Ohio, a 1 GW facility that will house as much as $30 billion worth of semiconductors when it comes online later this year. The project transformed a landscape of farmland and clapboard houses into a complex of military-grade weatherproof tents and gas turbines — a visual symbol of the industry's rapid expansion that has galvanized local opposition.
The power bottleneck is forcing hyperscalers to make difficult trade-offs among speed, cost, reliability and environmental impact. Amazon's strategy emphasizes cost efficiency and scale, leveraging two decades of construction experience and long-standing relationships with utilities and equipment suppliers. It was the first hyperscaler to sign a power purchase agreement with an existing nuclear plant, a move since replicated by Microsoft, Google and Meta.
Google has taken a different path, prioritizing clean energy even at the expense of speed. The company acquired renewable developer Intersect Power earlier this year and is building at least three Texas data centers co-located with solar and wind projects, allowing them to skip the long queue for grid connection under Texas power market rules. About a quarter of Google's expected 2030 capacity will come from leases rather than self-built facilities, according to Aterio.
Other hyperscalers are turning to off-grid natural gas to bypass grid constraints entirely. Microsoft struck a 20-year agreement with Chevron Corp. to power an AI data center in Texas with a dedicated natural-gas-fired plant. Meta and Amazon have similar projects in development, according to data from Cleanview, an industry research firm. Amazon has not publicly confirmed its involvement in a Fayette County, Ohio project, but Cleanview's data shows it is the only planned large power demand near a permitted off-grid gas facility there.
The regulatory environment remains favorable to natural gas for now, but that could shift. All four hyperscalers have also placed bets on next-generation energy technologies including small modular nuclear reactors, advanced geothermal systems and even space-based solar power. SpaceX and Google are exploring data centers in orbit.
For investors, the divergence in strategy creates a clear competitive wedge. Amazon's scale advantage — 9 GW of self-built capacity versus 5 GW for Google and Microsoft — gives it a structural cost edge that compounds over time. But Google's clean-energy focus may prove more resilient if the political winds shift against fossil fuels. Microsoft and Meta face the most exposure to construction delays and community opposition, given their heavier reliance on projects that require new grid connections or face local permitting battles.
The ultimate winner of the AI infrastructure race will depend on which company can navigate the growing tension between the industry's insatiable demand for power and the communities that must live next to it. With opposition spreading from Ohio to Virginia to Texas, the path to $300 billion in new capacity is no longer just an engineering challenge — it is a political one.
This article is for informational purposes only and does not constitute investment advice.