AI Data Centers Spark Land Rush and Political Battles as Big Tech Seeks 85 Gigawatts by 2030

Demand for AI Data Centers Sends Prospectors Hunting for Land and Power

Tripp Mickle has been reporting on the artificial intelligence boom since it began in late 2022. In January, employees with the Seattle start-up Cloverleaf Infrastructure, which secures power and land for data centers, huddled around a map of Wisconsin farmland. Brian Janous, the firm's co-founder, was in the process of completing an agreement with a local utility for enough power to light a small city. He and his team were hunting for land where that electricity could fill an artificial intelligence data center. Local dairy farmers were open to a sale, with a caveat, said Aaron Bilyeu, the firm's chief development officer. To comply with environmental laws, they still needed to spread manure nearby. "In Nebraska, we call that the smell of money," said Mr. Bilyeu, who grew up in Omaha. As the A.I. boom enters its fourth year, Mr. Janous and his team have become modern-day land men. They work at the intersection of utility companies and tech giants, securing the power and sites necessary for the hundreds of billions of dollars of data centers being built across the country. Their product -- powered land, they call it -- has become one of the nation's most valuable commodities. A.I. companies are seeking 85 gigawatts of power for new data centers by 2030, about a fifth more than the power grid can currently supply, according to S&P Global, a market research firm. The demand has tech companies scrambling to secure power and land as quickly as possible. The power crunch has created business opportunities. Bloom Energy, which started in 2001, has made billions of dollars by selling fuel cells to companies to power data centers. Williams Companies, a natural gas pipeline operator, has a deal to build a pipeline and power station in Ohio for Meta. And Crusoe won a contract to develop data centers for OpenAI after tapping excess gas from oil fields to power mobile data centers. Mr. Janous, 48, spent a decade overseeing Microsoft's energy strategy. He is trying to solve the power problem by finding unused capacity on the electricity grid. He and his team study electricity transmission flows and regional power sources. Then they strike deals with utility companies, secure permits and buy hundreds of acres nearby. They package the land and power and sell it to data center developers for OpenAI, Meta and Google, with much of the money flowing to Cloverleaf as a profit. "He's a wildcatter," said Dan Krueger, an executive vice president at WEC Energy Group, an electric services company based in Milwaukee. "The lifeblood of A.I. is electricity." Early oil prospectors, who are said to have gotten their nickname from a 19th-century Pennsylvania speculator who shot a wildcat and put it atop a backwoods derrick, were known to take big risks to strike oil and get rich. Modern land men bring more rigor to their high-stakes gambles by studying seismic data to discover oil fields. Mr. Janous and his team have more in common with modern-day oilmen than early speculators, but their business has its own reputational and financial risks. Being a data center middleman has thrust Cloverleaf into heated community debates. The team has faced hostile town halls and abandoned at least one proposed project in the face of local concerns that data centers increase electricity prices, strain water resources and create noise pollution. The company also faces the risk that the A.I. boom fizzles out. If companies cut back on data center plans, it could be left with millions of dollars in sunk costs and commitments for power equipment or land. "It's hard, and it's risky," said Ben Alingh, co-founder of Monarch Energy, a San Diego-based firm that is among a handful of companies that, like Cloverleaf, line up power and land for data centers. "It costs millions to do studies with utilities, and that money is not refundable. You have to have a stomach for that risk." In their haste to build data centers and circumvent red tape, some tech companies like Meta and xAI are building on-site power sources, much as factories did in the 1800s with their own power plants. The practice was abandoned long ago after Samuel Insull, a Chicago industrialist, showed it was cheaper to connect factories to power from a single plant. Like Mr. Insull, Mr. Janous and his team believe that connecting data centers to established utilities is preferable -- assuming they are operating in an area with capacity to spare. In places where there is excess capacity, data in a recent federally funded study shows, bringing new power customers on the grid like data centers can lower prices for all. When Mr. Janous left Microsoft in late 2023, he didn't have Cloverleaf in mind. But as the A.I. boom accelerated, he heard stories from friends in the industry about landowners calling tech companies with offers to sell acres without sufficient access to power. It was the equivalent of selling a house miles from the nearest sewer line. "It's a business that really shouldn't exist, but it does," said Mr. Janous, who raised $300 million from the private equity firms NGP and Sandbrook Capital. He has used the money to hire about two dozen people, buy land and purchase equipment. Relationships are one of Mr. Janous's biggest advantages. In 2021, he spoke on a panel at an energy conference in Chicago about Microsoft's trouble finding sites with adequate power for data centers. Mr. Krueger of WEC, who was in the audience, later introduced himself to Mr. Janous and said, "We've got the site for you." Mr. Krueger explained that WEC had provided power to Mount Pleasant, Wis., for a Foxconn complex that turned out to be much smaller than had been promised. He and Mr. Janous later put together a deal for Microsoft to buy the land and build data centers. When Mr. Janous started Cloverleaf in 2024, Mr. Krueger called with another opportunity. A project had fallen through for a manufacturing plant north of Milwaukee in a town called Port Washington. The site had ample land and a utility plan but no transformers or power lines. Cloverleaf signed contracts to pay for equipment and other costs at the site. It then worked with WEC to finish the power plan for a data center. Last year, Cloverleaf sold the 1,900-acre site with 1.3 gigawatts of power to Vantage Data Centers in a deal estimated to be worth more than $200 million, according to analyst estimates of land with that much power. Vantage plans to develop a $15 billion complex for Oracle and OpenAI with 1 gigawatt of A.I. capacity and four data centers. Cloverleaf has put together similar deals in Georgia and Alabama. In some instances, it doesn't even purchase land. Instead, it signs multiyear options for parcels at a fraction of the purchase price, which can later be transferred to a buyer. As communities increasingly question data center developments, Cloverleaf has tried to win towns over with commitments to address concerns about power, noise and water usage. It doesn't sign nondisclosure agreements with government officials that would make its projects secret. It supports data centers that need a one-time provision of water, which is then continually recycled. It suggests towns create fees if noise or light pollution exceeds local wishes. And it provides Mr. Bilyeu's phone number and email, should anyone have questions or comments. But its commitments have been met with skepticism. The project it sold to Vantage Data Centers in Port Washington faced blowback after the sale. It abandoned another project this year in Greenleaf, Wis., after some community members worried it would sap local resources. In Frenchtown Township, Mich., more than a hundred residents turned up at a community meeting in November with Cloverleaf to voice their opposition to developing a data center site on a former golf course, said Kim Giarmo, who joined a local committee challenging the project. She said people were worried that the project could create environmental issues at a nearby state park on Lake Erie, and that its underground batteries could harm wetlands. Residents doubt the project will deliver the 100-plus promised jobs and are concerned that the community won't know who will operate or use the future data center until after the permitting process is complete, Ms. Giarmo said. "They make it sound like it will be a bed of roses," she said. "Citizens aren't buying it." Mr. Janous said his firm doesn't force data centers into communities. It emphasizes that the centers may be less disruptive to their towns than a factory, and bring some high-paying jobs and tax revenue. "I understand it's scary to think about these giant facilities coming into your community," Mr. Janous said. "We try to find places where it's truly a win-win for a community, and that's not everywhere."

AI data centers are the next political fight. Big Tech is ready

Once people in his Republican-leaning Virginia legislative district stopped shutting their doors in his face last year, Democrat John McAuliff realized he had unexpectedly landed on a hot-button issue: Voters were getting frustrated with the armada of AI data centers popping up all around them. McAuliff, a former White House climate adviser under President Joe Biden, ended up narrowly defeating the Republican incumbent by less than two percentage points in November, and is now a member of the state's House of Delegates. He attributes his victory partly to a focus on rising utility costs from data centers across the region, notably in nearby Loudoun County. The enormous concentration of 200 power-hungry data facilities and counting outside Washington, D.C. is cementing Loudoun County's reputation as the data center capital of the world -- although Texas is not far behind in claiming that title. Democrats won control of both chambers of the Virginia legislature and the governor's office last year, a political trifecta. McAuliff now has a chance to deliver on a core campaign promise to make sure that tech companies, not his new constituents, are the ones paying up for utility cost increases. He supports legislation making its way through the legislature that would shift certain grid-connection costs from Dominion $D Energy customers to "high-load" users like data centers. And now it's the powerful tech industry that has been regularly knocking on McAuliff's door. "Someone in the industry, either the representatives of the actual companies or their chosen lobbyists, have been in my office probably every other day since I got here," McAuliff said in an interview. "They're a very aggressive industry, and not wholly unhelpful. Sometimes they'll offer to help contextualize something, or sometimes they'll offer to water it down. ... But they are aggressive and they're talented." Jesse Hunt, a spokesperson for the pro-AI group, said they want to elect "as many pro-innovation candidates as possible." Leading the Future isn't advocating for a specific set of policies beyond saying that AI standards should be set at the federal level, not left to states. AI firms are also spending tens of millions of dollars on lobbying efforts such as those in Virginia. It all comes as Big Tech companies kick off an unprecedented capital expenditure spree to build data centers and related infrastructure across the country. As the compute capacity needed to power the AI boom rapidly grows, Meta, Amazon $AMZN, and Google $GOOGL are leading the charge with more than $650 billion in planned spending this year, a sum rivaling Sweden's gross domestic product. Yet that tidal wave of AI-related spending is crashing against pervasive and growing skepticism within the American public about how AI affects jobs, the cost of living, and more. Surveys have shown opposition to data centers is rising fast. "People are skeptical of AI. People want the government to put guardrails on AI," said Alex Jacquez, chief of policy at the left-leaning think tank Groundwork Collaborative. "They would prefer that the pace of AI acceleration slow down so that we can get things right versus speed up so that we can get more innovation and productivity." Even some powerful Silicon Valley investors are warning about nascent opposition to data centers. Chamath Palihapitiya, a host of the "All-In" podcast, pointed to local backlash to data centers with 25 projects cancelled last year alone. "The core problem as I see it: Local residents are being asked to subsidize AI infrastructure through higher electricity bills with no upside. That's not a sustainable ask," he said last month. President Donald Trump nodded to rising voter frustration with higher power bills during his State of the Union address last week. He announced a "ratepayer protection pledge" aimed at ensuring tech companies shoulder all the electricity costs tied to bringing data centers online. "We're telling the major tech companies that they have the obligation to provide for their own power needs, so that no one's prices go up," Trump said. The policy was short on details. A White House official later said Amazon, Google, Meta, Microsoft $MSFT, Oracle $ORCL, and OpenAI are among the firms signing onto the pledge at a Wednesday event in Washington. But some analysts warn that it would be a mistake to explain away the public's frustration with data centers as only being about higher utility bills. At a recent municipal government meeting in Sunbury, Ohio, residents packed two rooms to voice opposition to a proposal that would rezone 330 acres for an Amazon data center. Their concerns ranged from diminished home values, the potential for damaging effects to public health, and other unexpected consequences of building a data center so close to their community. "I think that it's often reduced down to like energy and water," said Nat Purser, a senior policy advocate at the think tank Public Knowledge. "But it's the noise of the servers, it's the level of electricity consumption. People just don't like the aesthetics of big warehouse buildings. So I think lawmakers are trying to be responsive to all that." Most lawmakers are interested in some form of regulation that upholds AI's role in the U.S. economy and as a technology that's here to stay. As a result, current lawmakers and political candidates are beginning to roll out AI policy platforms. Rep. Ro Khanna, a Democrat who represents a district encompassing Silicon Valley, called last month for a "new tech social contract" that has AI powering middle-class wealth, not extracts from it it. "If we continue with the status quo or adopt poll-tested incrementalism, we will leave ordinary Americans out in the cold and modern prosperity will be only for the privileged," Khanna said. Part of that, he said, hinges on compelling tech companies pay for electricity bill increases instead of passing off those costs to families. It's a increasingly common position in Congress. Missouri Republican Sen. Josh Hawley Connecticut Democratic Sen. Richard Blumenthal rolled out bipartisan legislation last month compelling data centers develop their own power sources so they don't add to mounting utility bills. Potential 2028 Democratic presidential candidates are also dialing down their enthusiasm for data centers, Axios reports. In Virginia, the state's legislative session still has a few weeks left to go. Gov. Abigail Spanberger is eyeing new ways to bring down utility bills, after she campaigned last year in favor of data centers "paying their own way" for power. Virginia state lawmakers are still weighing a raft of bills dealing with data centers and their effect on electricity grids, water systems, and residents. The data center boom won't be paused anytime soon, but tech companies can probably expect more voter backlash -- and perhaps an initial swing at government oversight. "I've been really pleased by the number of legislators who have taken a hard position against continued unregulated industry growth -- including our utilities," McAuliff said.