- Key insight: Heavy community opposition is presenting a new risk to banks that finance data centers.
- Expert quote: "It's a rising risk, especially if the assumptions people are making — that in 10 years' time we're going to use AI in 95% of everyday life — don't happen. That's why they are trying to offload or at least find solutions like syndicated loans and securitization." —Arnaud Journois, senior vice president of Morningstar
- What's at stake: Communities face noise and air pollution and water and power shortages due to the enormous new data centers.
Power-plant engineers like to tell an apocryphal story about going door to door, asking people if it would be OK to put a wind or solar farm near their house. If the homeowner says no, the next question is, would you promise to stop using computers, TVs, blow dryers, refrigerators, air conditioning and heat? If the answer to that is no, an engineer takes a giant pair of scissors and cuts that home off the grid.
The point is, no one likes having a power plant in their neighborhood, but everyone needs them. The escalating use of generative AI has driven a boom in construction of data centers and the power plants upon which they rely: the International Energy Agency expects data-center electricity consumption to more than double to around 945 trillion watt-hours by 2030 — more energy than the entire country of Japan uses today.
Communities have been pushing back. A Gallup survey released this year found 71% of U.S. adults oppose building data centers in their area, including 48% who said they strongly oppose them. According to Data Center Watch, 75 data center projects worth $130 billion were blocked or delayed in the first quarter of this year.
President Trump, on the other hand, has strongly supported the data-center boom. "The only reason that communities throughout the U.S.A. should not want Data Centers is if they want to end up being backwards and poor," he wrote in a Truth Social post on Monday. "If they want to be successful and rich, with far lower taxes and jobs all over the place, let Data Reign."
Data-center opposition has started affecting the banks that provide loans to developers, hyperscalers and infrastructure operators and provide capital-markets financing through structured-finance transactions.
Opposition to bank-backed data centers
The biggest data center financers, according to Morningstar, are
Morgan Stanley and KKR Capital Markets were among the lead arrangers of a $9.7 billion warehouse credit facility for data center developer CyrusOne, which is building a $500 million data center in Sangamon County, Illinois. Local residents have strongly objected to the project and have sued the county over it, citing concerns about diesel generator noise, exhaust emissions, high water consumption, power grid strain, loss of farmland and zoning violations. The project is proceeding, but the financers have built in clauses saying construction funding can be used only once the necessary permits and leases are in place.
In Virginia, QTS, which is owned by Blackstone, ultimately terminated its Prince William Digital Gateway project after strong local opposition. QTS did not approach lenders for bank financing for that project.
In addition to local opposition, data center financing carries other risks.
"Concentration risk is one of the most significant credit considerations for banks active in the sector," Morningstar analysts wrote in their report. "Projects are typically large and often dependent on a limited number of hyperscalers, AI providers or specialist operators."
There's also the risk that the AI boom is a bubble that will eventually burst, leading to decreased use of and potentially abandonment of the new data centers.
"It's a rising risk, especially if the assumptions people are making — that in 10 years' time we're going to use AI in 95% of everyday life — don't happen," Arnaud Journois, senior vice president of Morningstar, told American Banker. "That's why they are trying to offload or at least find solutions like syndicated loans and securitization."
Pricing for added risk
The big banks are not backing down from financing controversial data centers, but they are trying to mitigate the risks.
One way is through pricing. "They are increasingly trying to price, structure and condition financing around the risks created by community opposition," Massimo Buonomo, an advisor to the European Commission, told American Banker. "Community acceptance is becoming part of credit underwriting. Banks are increasingly asking not simply 'Will this project generate enough cash flow?' but 'Can this project actually be built, permitted and operated on schedule?'"
Lenders are favoring jurisdictions with stronger permitting prospects and lower risks of project delays, he said.
Syndication and securitization
Banks are also mitigating data-center risks by keeping the financing off their own balance sheets, according to Journois.
One way is through syndication, where banks distribute loans across a broader group of lenders and investors.
"Banks are also making greater use of significant risk transfer, or SRT, transactions, enabling them to transfer portions of credit risk to external investors whilst preserving client relationships and maintaining lending capacity," the Morningstar report stated. "In addition, banks increasingly use loan sales and secondary-market transactions to reduce retained exposures, free up capital and enhance balance sheet flexibility."
However, the report warned, these mechanisms create new vulnerabilities. "Strong investor demand could weaken underwriting discipline, whilst the migration of exposures from regulated banks to nonbank financial institutions may reduce transparency and complicate the assessment of systemwide risk," the report stated.
This spreading of risk reminds Journois of the 2008 financial crisis, when several Wall Street firms collapsed or got into trouble.
"People argue that data centers are now part of everyday infrastructure financing, a stable asset class with only one or 2% maximum of the total exposure," Journois said. "But yet, if in 10 years' time promises are not kept and AI is disappointing, or we don't use as much AI as we thought we would, then the problems might arise."
Read more:
- Banks are derisking balance sheets, cutting crypto, exclusive survey finds
- How banks are navigating the AI data center backlash
- Here's what a CLARITY-free crypto market structure looks like
- What's driving activist investors, and how banks can be ready
He's concerned that systemwide financial risk could end up at insurance companies, private credit funds and nonbank institutions.
"This is where the risk lies," Journois said. "Because if this becomes a bubble at some point, these firms are not as regulated as the banks are. So in terms of financial systems, we might have a problem."
The Bank of England's July 2026 Financial Stability Report found that AI infrastructure financing is spreading across public debt, private credit, leveraged finance, structured finance, securitized data-center assets and special-purpose vehicles.
"A bank may reduce its direct exposure by syndicating a loan," Buonomo said. "An infrastructure fund may then acquire part of the exposure. Another portion may ultimately be financed through a structured or securitized product. The individual institutions may therefore appear less concentrated, while the financial system as a whole becomes more interconnected."
The Bank of England also noted that private-credit financing of AI investment increased from 9% in 2024 to 34% in 2025.
Buonomo does not compare data center financing to the mortgage market before the 2008 financial crisis. "The underlying assets are real infrastructure, many projects have strong counterparties, and hyperscalers such as Microsoft, Google, Amazon and Meta have substantial financial resources," he said.
But he does see a potential correlation risk.
"If banks, private-credit funds, infrastructure investors and securitized products are all effectively underwriting the same assumptions — perpetual AI demand growth, high utilization rates, continued hyperscaler spending, abundant electricity and stable data-center valuations — then diversification at the investor level does not necessarily mean diversification at the system level," he said.
Buonomo sees the data center financing sector moving from a phase of enthusiasm to a phase of discipline.
"The next stage will be determined not only by the availability of capital, but by power availability, permitting, community acceptance, financing structure and the resilience of the underlying infrastructure," he said. "In other words, I think the key question for banks is gradually changing from, 'How do we finance the AI build-out?' to 'How do we finance it without creating concentrated exposure to the same infrastructure, energy and demand assumptions?'"
Renewable energy
Another answer to data center opposition is to build facilities that use renewable energy, which comes with low carbon emissions, reduced water consumption and less community opposition.
Today, more than 40% of data centers in the U.S. are powered by natural gas, according to the International Energy Agency. Almost a quarter (24%) use renewable energy sources including solar and wind. A fifth (20%) use nuclear power and 15% use coal-fired power plants. The IEA expects that natural gas will be the largest source of additional supply over the next five years, followed by renewables.
According to Chris Castro, chief sustainability officer at Climate First Bank, renewable energy is the cheapest source of new power generation in the U.S. and globally.
"We made that crossover around 2022 when the price of renewables was more economically viable than new fossil fuel power plants," Castro told American Banker.
A July 2026 Lazard report on the cost of electricity found that "wind and solar, including when coupled with storage, remained the cheapest sources when compared to conventional gas peakers, nuclear and coal," Castro said. Combined cycle gas plants remained competitive but are more vulnerable to fuel price sensitivity, he said.
Because renewables are cheaper, 90-95% of all new energy over the last five years, and in the first six months of 2026, has been carbon-free clean energy, according to the U.S. Energy Administration Agency. Climate First Bank has made solar loans to more than 10,000 households as well as commercial businesses and communities. It has financed $500 million worth of clean-energy projects, Castro said.
Google, for one, is using solar power to fuel its new data centers in Texas and Ohio.
"Renewable energy is increasingly becoming a strategic necessity rather than simply an ESG consideration, as energy costs and grid capacity may become decisive factors in the profitability of AI data centers," Buonomo said.
