The investment risks of AI data centers

From AI Boom to Investment Risk: Rethinking pension fund data center exposure

As community opposition, regulatory scrutiny, and infrastructure constraints mount, some public pension funds are beginning to reconsider their exposure to data center investments. Their decisions raise broader questions about the sustainability of the AI infrastructure boom and who ultimately bears its financial and environmental costs.

The rapid expansion of artificial intelligence has fueled an extraordinary investment boom in data centers, with private equity firms, infrastructure investors, and technology companies committing substantial capital to build the physical infrastructure needed to power AI.

But as the industry races to expand, a different set of signals is emerging. Communities are increasingly challenging proposed developments, while policymakers are scrutinizing the economic and environmental costs of data center growth. For institutional investors, these pressures are translating into potentially material financial risks, including permitting delays, rising infrastructure costs, regulatory uncertainty, and questions about long-term demand and valuations. Together, these factors raise important questions about risk-adjusted returns and investors’ fiduciary responsibilities.

An additional factor surfacing among institutional investors is exposure concentration as AI investments permeate across strategies and asset classes. Recent decisions by a few major public pension funds suggest that these concerns are beginning to influence institutional investment strategies.

Pensions & Investments reported that the $330 billion New York City pensions’ Chief Investment Officer Monte Tarbox said that the AI boom echoes previous financial crises and that the city’s systems “passed on a technology-focused private equity fund because of AI exposure concerns, both in that strategy and across the NYC portfolios.”

P&I also reported that the $3.1 billion Jacksonville Police & Fire Pension Fund declined to make a new commitment to Blue Owl Capital’s Digital Infrastructure Fund IV, citing concerns about political opposition, utility access, regulatory uncertainty, and community resistance. Meanwhile, the $66.7 billion Arizona State Retirement System is exploring ways to reduce its increasingly concentrated data center exposure. Data centers already account for approximately 13% of its $9.3 billion real estate portfolio, with that share projected to exceed 20% by the end of 2026 as development assets are revalued.

Community opposition is becoming an investment risk
As data center development accelerates, communities across the country are increasinglypushing back against proposed projects, raising concerns about noise, air pollution, environmental impacts, and quality of life. This growing opposition is already shapinglocal zoning and permitting decisions and promptinggreater scrutiny from state lawmakers. For developers and investors, these developments introduce mounting regulatory uncertainty, potential project delays, and financial risks that could challenge the pace and profitability of data center expansion.

In early September, Massachusetts Governor Maura Healey issued an executive orderprohibiting state agencies from issuing permits for data center development without prior approval from the municipality where the project would be located. Meanwhile, in Vermont, local governments aremoving to ban data center development following Governor Phil Scott’s veto of legislation that would have imposed stricter regulations on the industry. Indeed, lawmakers in 16 states engaged in policy discussions about banning datacenter development. Together, these developments illustrate how growing state and local scrutiny is creating new regulatory hurdles for data center expansion, potentially increasing permitting uncertainty, delaying projects, and raising financial risks for developers and investors. 

Community opposition, legal challenges, and regulatory uncertainty are translating into financial risks for data center developers. In Virginia, QTS Data Centers and Compass Datacenters abandoned plans for the Prince William Digital Gateway, a proposed development involving tens of billions of dollars in investment, after a prolonged legal challenge resulted in the project’s approval being overturned. In Pacific, Missouri, Beltline Energy withdrew its proposed $16 billion data center development in August 2026, shortly after local officials enacted a one-year moratorium amid community concerns about environmental impacts. Similarly, in Texas, Diode Ventures abandoned a proposed data center near Cedar Creek Reservoir following sustained community opposition over water consumption and concerns about local infrastructure. A July 2026 analysis by Carbon Direct identified at least 46 U.S. AI data center projects delayed or canceled following community opposition between January 2024 and May 2026, representing approximately $170 billion in announced investment. 

A changing legislative landscape could reshape the economics of data centers
In addition to outright moratoriums, state policymakers are increasingly responding to concerns about the financial and environmental costs of data center expansion through other legislation addressing permitting, electricity affordability, water consumption, and public transparency. In September 2026, Californiaenacted seven data center bills, establishing new electricity and water reporting requirements, strengthening protections against shifting infrastructure costs onto ratepayers, and tightening environmental review requirements. In New Jersey, lawmakersestablished a separate electricity rate class for data centers and new energy requirements, alongside legislation requiring greater transparency about energy and water consumption. Oklahoma also enacted the Data Center Customer Ratepayer Protection Act of 2026, designed to protect existing utility customers from infrastructure costs associated with large electricity users.

Other states are introducing more direct constraints on development. New York lawmakers passed theResponsible Data Center Development Act, proposing a one-year moratorium on new permits for hyperscale data centers, environmental impact assessments, public hearings, and separate electricity and water rate classes. In Massachusetts, Governor Maura Healey issued a September executive order requiring local approval before state permitting can proceed, alongside new expectations for energy costs and environmental protections. Virginia has also approved legislation strengthening water-use reporting and emissions standards for data center backup generators. Together, these developments signal an evolving regulatory landscape in which developers and their financial backers may face greater compliance costs, longer approval timelines, and increased responsibility for infrastructure investments previously assumed to be borne by utilities or the public.

For private equity and infrastructure investors, these policy developments challenge a central assumption underpinning the data center investment boom: that projects can move rapidly from capital commitment to construction and revenue generation. As regulatory requirements evolve, investors may face longer development timelines, increased operating and infrastructure costs, and capital tied up in projects awaiting approval. These risks are especially consequential for speculative developments without secured utility access, permits, or long-term customers, raising questions about asset valuations, anticipated returns, and the potential for stranded investments.

The wider AI infrastructure ecosystem carries interconnected risks
Data center expansion is driving investment across electricity generation, transmission, natural gas infrastructure, water systems, construction, and land development. Private equity firms and institutional investors may hold overlapping interests across these sectors, creating cumulative financial exposures that are not always apparent at the individual project level.

These investments also raise questions about who benefits and who bears the costs. While developers, utilities, and technology companies may profit, communities and ratepayers can face higher infrastructure costs, environmental burdens, and forgo public revenue through subsidies and tax incentives. If anticipated demand fails to materialize, investors could face stranded assets while utilities and communities remain responsible for infrastructure built to support unrealized growth.

The central question is whether the pace, scale, and financing assumptions underlying AI infrastructure expansion remain sustainable as regulatory scrutiny, community opposition, and investment risks grow. For pension funds and private equity managers, greater transparency into ownership structures, cumulative exposure, project valuations, and environmental and community impacts is essential to protecting beneficiaries and ensuring accountability for the broader costs of the AI infrastructure boom.

What responsible investors should be asking
As AI infrastructure investment expands, private equity firms and institutional investors must look beyond projected returns to assess financial, regulatory, and community risks. Investors and asset managers are seeking to develop guidelines to support holistic risk considerations. Key questions include:

• Regulatory risk: Have projects secured permits, utility access, and local approvals? Could moratoriums or new legislation delay development?
• Infrastructure costs: Who bears the costs of electricity, water, and supporting infrastructure?
• Demand and valuations: Are returns backed by long-term tenants or speculative demand projections?
• Portfolio exposure: Are investors accounting for overlapping investments across data centers, energy, and related infrastructure?
• Community impacts: Are environmental justice concerns and community opposition reflected in investment diligence? How are impacts on electricity costs for utilities and consumers managed?
• Environmental impacts: How are developers mitigating impacts of water usage? How are developers determining power sources and emissions mitigation? 
• Downside risks: How would delays, rising costs, or shifting hyperscaler demand affect valuations and exit opportunities?

These questions are particularly consequential for public pension funds responsible for protecting workers’ and retirees’ savings.

A potential turning point for AI infrastructure investment
The AI infrastructure boom is far from over, but growing community opposition, regulatory scrutiny, infrastructure constraints, and speculative development risks are challenging assumptions about its long-term profitability.

As stewards of long-term capital investors are exploring how to assess and manage these emerging risks before losses and costs are passed on to pension beneficiaries, ratepayers, and communities. Greater transparency, rigorous due diligence, and accountability across the AI infrastructure ecosystem will be critical as the next wave of investment takes shape.