FuelCell Energy, Inc. is stacking new contracts and expanding its generation fleet at a moment when U.S. data centers, driven by artificial intelligence workloads, face an electricity crunch that federal officials say could double or triple demand by 2028. The Danbury, Connecticut-based company reported a 62.8 megawatt operating portfolio as of October 31, 2025, and has signed deals on two continents that signal an aggressive growth push. Whether that expansion can keep pace with the grid strain AI is creating remains an open question, one shaped by execution risk, supply chain constraints, and the sheer scale of power the tech sector now requires.
AI Workloads Are Straining the Grid
The electricity math behind AI is stark. The 2024 assessment from Lawrence Berkeley National Laboratory provides historical estimates of U.S. data center electricity consumption stretching back to 2014 and projects scenario ranges through 2028, framing accelerating compute and AI trends as a primary force behind rising consumption. The U.S. Department of Energy, drawing on that research, characterized the outlook bluntly: data center electricity use is expected to double or triple by 2028, with AI applications identified as a key driver and hyperscale campuses emerging as a focal point for planners worried about regional grid reliability.
That projection matters for any company selling on-site power generation. Grid operators across the country are already struggling to approve new interconnections fast enough, and traditional utilities face multi-year permitting timelines for large-scale plants. Fuel cells, which generate electricity through electrochemical reactions rather than combustion, can be sited close to the load they serve, cutting transmission losses and bypassing some of the bottlenecks that slow conventional power buildouts. FuelCell Energy’s carbonate technology produces electricity and heat simultaneously, which in theory makes it attractive for facilities that need reliable, round-the-clock power. The gap between that theoretical fit and a signed data center contract, however, is one the company has not yet publicly bridged with named AI clients or megawatt allocations specific to that sector, leaving investors to infer potential from broader market trends rather than disclosed deals.
Hartford Project Anchors Near-Term Growth
FuelCell Energy’s most concrete domestic expansion move is a power purchase agreement signed in January 2025 with Eversource and United Illuminating for a 7.4 MW carbonate fuel cell system in Hartford, Connecticut. According to the company’s quarterly filing for the period ended April 30, 2025, construction is expected to wrap up in calendar year 2026, with commercial operations targeted for December 2026, a schedule that would add meaningful capacity to a generation portfolio that stood at 62.8 MW as of the end of fiscal year 2025. The Hartford system is designed as a grid-supporting asset rather than a behind-the-fence installation for a single customer, positioning FuelCell as a contributor to regional reliability at a time when New England faces capacity constraints and aging thermal plants.
Separately, a company-referenced press release described a $160 million commitment to support the Hartford-area grid, which has occasionally been conflated with the company’s overseas service business. That figure and the $159.6 million long-term service agreement disclosed in SEC filings for a South Korean project have created some confusion in coverage, as both large-dollar contracts were discussed around overlapping periods. The Hartford PPA and the South Korean service deal are, however, distinct arrangements serving different geographies and customers, a distinction that matters when assessing where FuelCell’s revenue pipeline is actually concentrated and how diversified its contracted cash flows have become.
South Korea Deal Tests Manufacturing Scale
The company’s largest disclosed service arrangement sits on the other side of the Pacific. According to FuelCell Energy’s quarterly filing for the period ended April 30, 2024, the company entered into a long-term service agreement on May 28, 2024, with Gyeonggi Green Energy for the 58.8 MW Hwaseong-si platform, valued at $159.6 million. The contract calls for replacement of 42 modules, each rated at 1.4 MW, a substantial manufacturing commitment that will test FuelCell’s production capacity and supply chain resilience over the life of the agreement, especially as the company balances domestic project execution with international obligations.
This South Korean deal is significant for reasons beyond its dollar value. It locks in recurring service revenue and forces the company to maintain a module production cadence that could, in principle, be redirected or expanded to serve new customers if demand materializes from data center operators or other high-consumption users. The risk runs both directions: if FuelCell struggles to deliver 42 replacement modules on schedule, it signals manufacturing constraints that would undermine any pitch to power-hungry AI facilities; if it executes cleanly, the Hwaseong-si contract becomes a proof point for the kind of large installed base the company argues it can support, a narrative that underpins its growth story in investor communications.
Portfolio Expansion and Policy Tailwinds
The company’s most recent annual filing underscores how rapidly its operating footprint has grown relative to earlier years. In its Form 10-K for the period ended October 31, 2025, FuelCell reported a 62.8 MW portfolio of operating projects, reflecting a mix of utility-scale and distributed assets under long-term contracts. That filing also details a development pipeline that includes additional generation and service opportunities, suggesting that the Hartford plant and the Hwaseong-si service agreement are part of a broader effort to secure predictable revenue streams as the company moves beyond a purely project-by-project model. For AI-intensive customers, a larger fleet can be read as evidence of operational maturity, though the absence of named hyperscale clients remains a notable gap.
Public policy may help close that gap over time. The U.S. Department of Energy’s advanced energy initiatives have emphasized high-efficiency, low-emission power technologies that can integrate with a changing grid, and carbonate fuel cells fit squarely within that efficiency-first framing when fueled with natural gas, biogas, or hydrogen blends. In parallel, DOE’s new project clearinghouse for large loads, accessible through the Genesis portal, is intended to streamline coordination among developers, utilities, and federal agencies as data centers, manufacturing plants, and other energy-intensive facilities seek reliable power. For companies like FuelCell, that kind of centralized visibility into future demand could create opportunities to position on-site generation as part of integrated solutions rather than as stand-alone, speculative builds.
Can Fuel Cells Ride the AI Power Wave?
Even with policy support and a growing contract base, FuelCell Energy faces a scale challenge. The company’s current 62.8 MW operating portfolio, while material for a niche technology provider, is small relative to the multi-hundred-megawatt campuses that leading cloud and AI firms are planning across the United States. A single hyperscale facility can require 100 MW or more of firm capacity, implying that a full-site solution would dwarf FuelCell’s existing fleet. That mismatch does not preclude participation, fuel cells could be deployed as a resiliency layer alongside grid power and batteries, but it does mean the company must pick its spots, targeting configurations where its high-capacity-factor output and combined heat-and-power capabilities offer clear advantages over alternatives.
Execution risk also looms large. The Hartford project must be delivered on time and on budget to validate the company’s ability to build new utility-facing assets while simultaneously fulfilling its obligations to Gyeonggi Green Energy and other customers. Any slippage in module manufacturing for the Hwaseong-si platform, or delays in commissioning Hartford, would raise questions about how quickly FuelCell can scale to meet AI-driven demand that, according to federal projections, could double or triple data center electricity use by 2028. Conversely, if the company hits its milestones, maintains uptime across its fleet, and leverages policy tools aimed at advanced energy deployment, it will be better positioned to argue that fuel cells belong in the power mix for an AI era defined by relentless, around-the-clock computing loads.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
