There is a version of the AI infrastructure story that the industry tells publicly, and there is a version visible in SEC filings, class-action complaints, and the growing gap between announced timelines and operational reality. For most of the past three years, those two versions coexisted without much friction. Capital markets were willing to underwrite the optimistic version. Communities were still in the early stages of understanding what data center development meant for their electricity bills. Regulators had their attention elsewhere.
That window is closing. The Fermi America class-action lawsuit is the most visible illustration of where the accountability trajectory is heading. Investors filed after the first anchor tenant pulled out while management was still telling them commitments were in advanced stages. This is not, however, an isolated case. Rather, it is the leading edge of a broader reckoning. The gap between what AI infrastructure companies say and what is operationally true is widening.
The Gap Between Announcement and Reality Is Widening
The AI infrastructure industry has developed a communication culture that systematically overstates readiness and understates risk. Developers announce projects at sizes that reflect aspirational capacity rather than confirmed commitments, for instance. Developers present timelines based on engineering assumptions that exclude interconnection queue realities, supply chain lead times, and regulatory approvals. These factors consistently push actual commissioning dates well beyond initial projections. Sustainability commitments are, moreover, made at a level of specificity that implies operational confidence the underlying data does not support.
None of this is unique to AI infrastructure. Ambitious capital-intensive industries have always balanced the optimism needed to raise capital with the honesty needed to execute. What is changing, however, is the scale of the gap and the number of parties now measuring it precisely. Investors who bought Fermi America at IPO have a class-action attorney measuring exactly what management said versus what was true. Communities in Pennsylvania told data center investment would be rate-neutral have their electricity bills measuring the same gap. The measuring instruments are multiplying, and they are increasingly connected to legal remedies.
The Sustainability Commitment Problem
The transparency problem extends well beyond project timelines and tenant disclosures. Hyperscalers have made some of the largest public corporate pledges of the past five years in the sustainability category. The metrics against which they were made are, notably, difficult to verify independently. Carbon-free by 2030, water-positive by 2030, net-zero by 2040. Companies made these commitments when AI consumed far less energy. The political environment, moreover, rewarded ambitious targets without requiring rigorous accounting.
The energy reality of 2026 is testing those commitments in ways that 2020 board rooms did not anticipate. Microsoft has acknowledged that its carbon emissions have increased significantly as AI data center expansion has accelerated. Google has disclosed that its water consumption has risen sharply. Both companies are managing that gap through accounting flexibility, timeline extensions, and renewable energy certificate purchases. Critics argue those certificates do not represent the firm, hourly carbon-free power their commitments implied. That approach is not fraudulent. Regulators examining greenwashing claims are, however, trained to scrutinise it closely.
SEC climate disclosure rules require public companies to treat material climate risks with the same rigour as financial risks. By that standard, a hyperscaler’s 2030 carbon-free commitment is a material risk disclosure question if the energy trajectory makes it unachievable. It is not, in other words, primarily a sustainability question. And the legal infrastructure for securities fraud and greenwashing enforcement is increasingly equipped to address it.
The Investor Disclosure Dimension
The Fermi America situation illustrates a specific dimension of the transparency problem: the disclosure of material information about project status to investors. On the March 30 earnings call, Fermi’s CEO told analysts the company was signing letters of intent with potential tenants. The complaint alleges, however, that the company already knew the anchor tenant problem was worse than it had publicly stated. Whether that specific claim ultimately constitutes securities fraud is a question for courts to determine. What is not in question: the gap between public statements and operational reality triggered litigation within weeks of the CEO’s departure.
That sequence, public optimism followed by operational failure followed by litigation, is not, consequently, going to remain confined to Fermi America. A wave of SPACs and IPOs created a cohort of AI infrastructure companies with public shareholders and disclosure obligations. Many of those companies made statements about project timelines, tenant commitments, and power access that operational reality is now testing. The ones that fall short will, in turn, face the same scrutiny that Fermi America is facing. The legal framework for holding them accountable is, notably, already in place.
What the Industry Needs to Do Differently
The transparency problem is not primarily a legal problem. It is a credibility problem that legal consequences are making visible. An industry that relied on optimistic projections to raise capital is discovering those projections create liability when they fail to materialise. That liability, in other words, is not easily absorbed by the capital raised. The solution is not better lawyers. It is more honest communication with the full range of stakeholders now in a position to hold the industry accountable.
For investor communications, that means presenting timelines that reflect realistic interconnection, permitting, and supply chain scenarios rather than best-case assumptions. It means disclosing anchor tenant status with specificity rather than vague references to advanced discussions. None of this is legally required beyond what disclosure rules already mandate. It is, however, the standard that the market is beginning to impose through litigation and regulatory scrutiny.
For community communications, it means raising electricity rate impacts with communities before announcing projects, rather than after opposition has formed. It means making commitments about rate neutrality backed by specific utility agreements rather than general assertions. It means, ultimately, treating community trust as an operational asset rather than a PR variable. As we have covered in our analysis of how the AI data center backlash has become a swing-state political issue, the political consequences of the trust deficit are already materialising in ways that can block projects regardless of their technical merits.
Getting the Sustainability Disclosure Right
For sustainability communications, it means distinguishing between purchased renewable energy certificates and actual carbon-free power, between projected emissions reductions and measured ones, between aspirational targets and operationally achievable commitments. That regulatory infrastructure is developing faster than most AI infrastructure executives realise. Companies that adopt more rigorous disclosure standards voluntarily are building a compliance posture significantly less expensive than one imposed by regulators.
For three years, the AI infrastructure industry has built at a pace that prioritised speed over accountability. A legal and regulatory environment is now imposing the accountability the industry did not self-impose. Companies that treat that shift as a compliance burden to manage will find it expensive. Those treating it as an opportunity will find genuine credibility with investors and regulators. That credibility is harder to replicate than any engineering specification.
