Executive Summary
AI infrastructure competition is expanding from compute to power. Over the past two years, the market has focused on GPUs, HBM, advanced packaging, and cloud capex. However, as AI data centers become more power-intensive, electricity availability is becoming a prerequisite for computing capacity to come online. For large data centers, the key question is no longer only the cost of electricity, but whether power can be secured quickly and reliably.
The U.S. power system is at the center of this constraint. AI data centers are creating large, concentrated, and continuous loads, while the U.S. grid remains aging, regionally fragmented, and slow to expand. This creates an “impossible trinity” between low cost, fast deployment, and high reliability. As a result, the market is reassessing multiple power pathways, including gas turbines, storage, coal-backed reliability, microgrids, and SOFC.
SOFC is not positioned to replace mainstream power sources. Gas turbines will likely remain the main source of incremental dispatchable power, while storage and coal serve different reliability and balancing roles. SOFC’s value lies in high-value, localized power-constrained scenarios where customers need fast-deployable, modular, and relatively lower-emission onsite power.
Bloom Energy’s investment logic is built on this backdrop. The company is one of the few U.S.-listed platforms with SOFC as its core technology and direct exposure to data centers and other critical-load customers. Its Energy Server is not simply a fuel-cell product; it is an onsite power system that offers “time-to-power” when grid access is delayed or traditional power equipment is constrained.
Bloom’s core alpha comes from scarcity value, data center demand upgrading its use case, existing customer validation, and capacity expansion that could support larger orders. The key debate is not whether SOFC becomes a dominant power source, but whether Bloom can continue winning and delivering high-value orders during this AI power bottleneck window.
The main risks include valuation volatility, project execution delays, high system costs, technology durability, competing power pathways, fuel-price exposure, and customer concentration.
1. Industry Background: AI Is Pushing the U.S. Power System into an “Impossible Trinity”
Over the past two years, the core bottleneck in AI infrastructure has largely been understood through the lens of computing hardware: GPUs, HBM, advanced packaging, server clusters, and cloud capex. However, as AI training and inference workloads continue to expand, the constraint is gradually moving from “can we secure enough chips?” to a more fundamental question: can we secure enough power, quickly and reliably?
For large-scale AI data centers, electricity is no longer just an operating cost. It is becoming a prerequisite for computing capacity to come online. AI data centers require higher power density, more stable electricity supply, and faster deployment timelines than traditional internet data centers. As a result, electricity availability is becoming a key determinant of project delivery, asset utilization, and return on invested capital.
The U.S. power system is particularly exposed to this challenge. On the demand side, AI data centers are creating a new class of large, concentrated, and highly reliable power load. On the supply side, the U.S. grid is aging, fragmented by region, and slow to expand. According to the materials reviewed, natural gas accounted for roughly 45% of U.S. power generation in 2024, renewables including wind, solar, and hydro accounted for around 40%, and coal accounted for about 15%. At the same time, the average age of the U.S. grid is over 40 years, while new grid construction has fallen short of expectations. This creates a structural mismatch between rapidly rising power demand and limited grid expansion.
The problem is not only a matter of national power supply, but also of regional constraints. The U.S. grid lacks the kind of large-scale cross-regional dispatch capability seen in more centralized systems. Even if power exists at the national level, it may not be available in the right region, at the right site, and within the required construction window. For AI data center developers, this means that grid connection delays and local capacity bottlenecks can directly delay project monetization.
This brings the U.S. AI power system into an “impossible trinity”: electricity must be low-cost, fast to deploy, and highly reliable. Cost determines long-term data center competitiveness. Speed determines whether projects can come online as scheduled. Reliability is non-negotiable for both data center operators and the power system. Under normal demand conditions, these objectives could be balanced through utility expansion and long-term planning. Under the rapid growth of AI workloads, they are increasingly in conflict.
Traditional solutions are struggling to keep pace. Grid upgrades require long approval and construction cycles. Large power plants face equipment delivery, permitting, interconnection, and ramp-up constraints. The materials suggest that meaningful improvement in the U.S. grid may not arrive until around 2035, implying that the next five years could remain structurally constrained.
As a result, data center customers are no longer relying solely on the traditional grid. The market is beginning to reprice power solutions that can be deployed closer to load, including natural gas turbines, storage, coal-backed reliability, microgrids, onsite generation, and solid oxide fuel cells. These technologies do not serve the same role. Gas turbines remain the most important mainstream source of new dispatchable power. Storage is primarily a balancing and backup solution. Coal is regaining relevance as a reliability backstop. SOFC, by contrast, is emerging as a niche but increasingly relevant onsite power solution, supported by modularity, relatively low emissions, low noise, and faster deployment.
Therefore, the current power-chain revaluation is not simply a clean energy trade or a defensive utility trade. It is a repricing of power availability in the AI era. In the past, the market focused mainly on who could provide the cheapest electricity. In a power-constrained environment, the more important question becomes: who can provide usable electricity quickly, reliably, and at an acceptable cost?
This is the context in which Bloom Energy and the SOFC value chain should be understood. SOFC is unlikely to become the dominant source of U.S. power in the medium term, nor is it positioned to replace gas turbines, coal, or utility-scale generation. Its value lies in specific high-value use cases where grid access is slow, turbine delivery is constrained, and customers need power quickly. In this setting, SOFC is not a broad replacement for the grid, but a potential solution for localized power bottlenecks.
Bloom Energy’s investment logic should therefore be framed within this U.S. AI power “impossible trinity.” The company does not need to become a mainstream power supplier. If it can provide fast, reliable, and environmentally acceptable onsite power for data centers, semiconductor facilities, and industrial campuses, its valuation framework may shift from that of a traditional fuel-cell company to that of a scarce AI power infrastructure asset.
2. Power Supply Pathways: SOFC as a High-Value Complement to Mainstream Power Sources
As the U.S. AI data center power gap widens, the market is reassessing multiple power supply pathways, including gas turbines, renewables plus storage, coal, nuclear power, and SOFC. These technologies should not be viewed through a simple “better or worse” framework. Each serves a different role: some are more suitable as large-scale baseload or dispatchable power sources, some function mainly as backup or balancing resources, while others are better suited to solving localized power bottlenecks for specific customers and sites.
Natural gas turbines remain the most important source of incremental dispatchable power in the U.S. Gas already accounts for a large share of U.S. power generation, and combined-cycle gas turbines (CCGT) generally offer strong economics and dispatchability. For data centers that require stable power supply, gas generation remains one of the most realistic and scalable solutions. The challenge is that gas turbines cannot be deployed instantly at large scale. The global advanced turbine market is highly concentrated among a few suppliers, and current order backlogs are significant. Even after equipment is delivered, projects still require construction, grid connection, and capacity ramp-up, meaning actual power delivery can lag market expectations.
Open-cycle gas turbines (OCGT) and small gas turbines offer more flexibility, but their role is closer to peaking and transitional supply rather than long-duration baseload power. OCGT units are less efficient and are mainly used for start-stop and peak-load functions. Small turbines can help relieve near-term pressure, but their gas consumption is meaningfully higher than that of CCGT. If natural gas prices rise, their economics become even less attractive. As a result, small gas turbines may serve as a temporary bridge, but they are unlikely to become the core long-term solution for AI data center power demand.
Renewables plus storage face a different set of constraints. Wind and solar projects can be built relatively quickly, but the key issue in the U.S. is not only whether generation assets can be built, but whether they can be connected to the grid, provide stable power, and match 24/7 data center load requirements. The U.S. interconnection queue remains very large, historical completion rates are limited, and grid-connection timelines are long. Policy shifts have also increased uncertainty around future renewable deployment. Storage is an essential supplement, but it is primarily a balancing and backup resource. Data centers require continuous and reliable power, while storage depends on charge-discharge cycles. On its own, storage is not sufficient to support long-duration, high-reliability data center operations, especially under extreme weather conditions.
Coal and nuclear represent two different ends of the reliability and long-term low-carbon spectrum. Coal had been in structural decline in the U.S., but its reliability value is being reassessed as AI power demand rises and the deployment of gas, renewables, and storage faces constraints. For data centers, gas and coal remain high-confidence power sources. However, coal faces environmental, policy, and asset-aging constraints, making it more suitable as a system-level reliability backstop rather than a preferred source of incremental power for newly built data centers. Large-scale nuclear and SMR offer long-term potential, but permitting, construction, and commercialization timelines are too long to address the urgent 2026–2030 power gap.
Against this backdrop, SOFC has a clearer role. It is not a mainstream solution for the entire U.S. power system, nor is it a full replacement for gas turbines. Instead, it is an onsite power solution for high-value loads. SOFC converts chemical energy directly into electricity through electrochemical reactions, avoiding the traditional multi-step process of converting chemical energy into heat, then mechanical energy, and finally electricity. This gives SOFC certain advantages in efficiency, emissions, noise, and part-load operation. SOFC systems can also use multiple fuel types, including hydrogen, natural gas, and biogas.
More importantly, SOFC’s commercial value lies in its deployment speed and modularity. It can be installed in a modular format and deployed faster than gas turbines or SMR, making it relevant as a rapid power expansion tool in power-constrained data center markets. When grid access is slow, turbine delivery is delayed, and customers urgently need electricity, SOFC can function as a practical “time-to-power” solution.
Therefore, SOFC should not be evaluated using the same framework as mainstream baseload power sources. Its total market size may remain smaller than gas turbines, coal power, and utility-scale generation. However, in AI data center use cases—where power density is high, reliability requirements are strict, and customers are willing to pay a premium for faster power availability—its marginal value can be significant. For Bloom Energy, the key question is not whether SOFC can become a dominant source of U.S. power, but whether it can become one of the few fast-deployable, scalable, and publicly investable onsite power assets during this window of localized power shortages and accelerated data center construction.
3. SOFC Technology: A Fast-Deployable Onsite Power Solution for Data Centers
Solid oxide fuel cells, or SOFCs, generate electricity by converting the chemical energy of fuel directly into electricity and heat through electrochemical reactions. Unlike gas turbines, which follow a multi-step process from chemical energy to heat, mechanical energy, and then electricity, SOFCs do not rely on combustion or large rotating machinery. This gives them advantages in efficiency, emissions, noise, and modular deployment.
Structurally, an SOFC system consists of fuel cells, stacks, and balance-of-system components. The cell includes an anode, electrolyte, cathode, and interconnect. The electrolyte conducts oxygen ions while separating fuel from air, and the interconnect links individual cells while managing gas flow and electrical conduction. These materials and manufacturing processes are critical to both system performance and cost.
For data centers, SOFC’s first advantage is efficiency. Because it converts chemical energy into electricity more directly, SOFC power generation efficiency can approach 60%. If combined with combined heat and power applications, overall system efficiency can be even higher. For high-load, long-duration data center operations, this can help reduce fuel consumption and improve energy productivity.
The second advantage is fuel flexibility. SOFC systems can use hydrogen, natural gas, biogas, and other fuels, and they do not depend entirely on low-cost, high-purity hydrogen supply. This is important because the green hydrogen value chain is still immature. In the near term, SOFC can rely on existing natural gas infrastructure, making commercialization more realistic.
The third advantage is modularity and speed. SOFC systems can be deployed in modular formats and installed faster than large gas turbines or SMR projects. This directly addresses the “time-to-power” problem facing AI data centers. When grid access is delayed and turbine delivery is constrained, faster onsite power becomes commercially valuable.
SOFC also has better onsite characteristics than many traditional power sources. It produces lower noise and lower local emissions than combustion-based systems, and its carbon intensity is generally below coal and conventional gas turbine generation. This matters for large technology companies that need electricity quickly but still face emissions, community impact, and long-term sustainability constraints.
The technology still has limitations. SOFC systems remain expensive, with costs elevated by high-temperature materials and complex manufacturing processes. High-temperature operation also creates durability and degradation challenges. In addition, SOFC systems are less suitable for rapid start-stop or fast load-following applications than some conventional power assets.
Therefore, SOFC should not be viewed as a universal power solution or a near-term replacement for gas turbines. Its value lies in specific use cases that require fast deployment, continuous operation, modular expansion, lower noise, and relatively lower emissions. For AI data centers, its appeal comes not from one single superior metric, but from the balance it offers across efficiency, fuel flexibility, deployment speed, reliability, and environmental acceptability. As electricity becomes a bottleneck for data center commissioning, that balance is becoming more valuable.
4 Market Opportunity and Competitive Landscape: A Small but Fast-Growing SOFC Segment
SOFC remains a niche technology within the broader power system. It should not be directly compared with gas turbines, coal-fired generation, or utility-scale power assets in terms of absolute installed capacity. However, its low penetration also creates higher growth elasticity. For investors, the industry beta of SOFC lies not in its total share of the power mix, but in the potential acceleration of adoption from a small base, particularly as data center demand shifts from pilot projects to larger-scale procurement.
Data centers have become one of the most important use cases for SOFC. Current applications include distributed power generation, data centers, commercial and industrial combined heat and power, transportation and marine auxiliary power, hydrogen production and storage, as well as off-grid power supply for remote sites. Among these, data centers account for a meaningful share of SOFC demand and represent one of the most attractive growth channels. This suggests that SOFC growth is no longer driven only by clean energy demonstration projects or policy incentives, but is increasingly linked to the investment cycle of AI data centers.
The market remains at an early stage of commercialization. North American SOFC annual installations could reach roughly 0.5–1.25GW during 2026–2030, while the cumulative global SOFC market could exceed $60 billion over the same period. Although SOFC will likely remain small relative to the overall U.S. power system, this level of incremental demand is already meaningful for the current industry base and for companies with established production capacity.
That said, the addressable market should be framed with discipline. SOFC is unlikely to become a mainstream power source in the U.S. over the medium term. Compared with CCGT, coal, utility-scale generation, and storage, SOFC remains smaller in scale and still faces constraints in system cost, durability, manufacturing capacity, and supply-chain maturity. Current system costs remain high, partly due to high-temperature materials and complex manufacturing processes. Durability degradation, start-up and load-following limitations, and insufficient industrial scale are also important bottlenecks. Therefore, SOFC is better defined as a fast-deployable supplementary power solution for high-value use cases, rather than a broad replacement for mainstream power generation.
The competitive landscape is still evolving. The global SOFC market is led mainly by companies in Japan, the U.K., and the U.S., with representative participants including Mitsubishi Power, Aisin, Ceres Power, and Bloom Energy. Bloom Energy holds an important position in the global market, but the industry has not yet reached the same degree of concentration as the large gas turbine market. Technology pathways, customer segments, and business models remain open to further differentiation.
Along the value chain, SOFC value capture is not limited to downstream system integration. Upstream materials and midstream stack manufacturing are also important. Electrolytes, interconnects, cathode materials, stacks, thermal management, power electronics, and control systems can all benefit from industry growth. Interconnects are one of the higher-value components in an SOFC system, and materials such as metal interconnects, chromium-based alloys, and rare-earth-related inputs are important to both cost reduction and performance improvement. For U.S. equity investors, however, the most direct investment exposure is likely to come from downstream system platforms, as they are best positioned to secure data center orders and convert technology capability into revenue.
Overall, SOFC should be understood as a small-base, high-growth segment with a competitive structure that is not yet fully settled. Its total market size should not be overstated, but its marginal impact on relevant companies can be significant. As AI data centers increase demand for fast-deployable onsite power, companies with customer access, system integration capabilities, production capacity, and scarcity value in public markets are more likely to become the primary vehicles for investors seeking exposure to SOFC growth. For Bloom Energy, the key is not to prove that SOFC will become a dominant power source, but to demonstrate that it can continue to win orders and expand delivery capacity in a high-value, fast-growing niche market.
5 Company Overview: Bloom Energy’s Business Model and Strategic Positioning
Bloom Energy is one of the few publicly listed U.S. companies with SOFC as its core technology platform. Unlike many fuel-cell companies that focus primarily on transportation, hydrogen mobility, or demonstration projects, Bloom is more focused on stationary power generation. Its key end markets include data centers, industrial campuses, commercial facilities, and critical infrastructure. This positioning gives the company more direct exposure to the power shortage theme driven by AI data center expansion.
Bloom’s core product is the Energy Server, a modular SOFC power generation system. Customers are not simply buying an individual fuel-cell stack; they are purchasing an onsite power solution that can be deployed close to load. The system can generate electricity using natural gas, biogas, or hydrogen, and can be configured in modular blocks based on customer load requirements. It can also be integrated with microgrids, backup power systems, and thermal management infrastructure. For data center customers, the main value is not merely lowering electricity cost, but shortening time-to-power, improving reliability, and reducing dependence on traditional grid expansion.
Bloom’s revenue structure reflects the characteristics of a project-based infrastructure business. Revenue mainly comes from product sales, installation, and long-term services. Product revenue reflects equipment delivery capability, installation revenue reflects project execution and site progress, while service revenue comes from long-term maintenance and performance support. As the installed base grows, service revenue could improve business stability. However, near-term growth is still primarily driven by new project deliveries, especially from data center and industrial customers.
The company’s business model can be summarized as equipment sales plus system delivery plus long-term service. This is not an asset-light software model; it requires manufacturing capacity, supply-chain management, and project execution capabilities. At the same time, compared with traditional large-scale power projects, Bloom’s modular SOFC systems offer shorter construction cycles and better replication potential. In practical terms, the company is selling “time-to-power”: the ability to provide deployable onsite electricity when customers cannot quickly secure grid capacity or traditional power equipment.
Customer mix is central to Bloom’s revaluation. Earlier applications included commercial buildings, retail facilities, telecom infrastructure, and selected technology data centers, where Bloom was often viewed mainly as a distributed clean energy provider. As AI data center power demand accelerates, the use case is shifting from energy efficiency and low-carbon generation toward speed of power availability and reliability. Data center projects are larger in scale, supported by customers with stronger purchasing power, and more sensitive to delivery timelines. This could meaningfully increase Bloom’s order size and revenue elasticity.
However, this business model also brings execution risk. Revenue recognition depends not only on signed orders, but also on production capacity, equipment delivery, customer-site construction, fuel access, permitting, and project acceptance. As data center orders become larger, market attention is likely to shift from whether demand exists to whether Bloom can deliver on time, sustain margins, and manage working capital and cash flow pressure.
In this sense, Bloom Energy should not be viewed simply as a traditional fuel-cell concept stock or a utility company. It is better understood as a distributed onsite power platform serving high-reliability load customers. As AI-driven power bottlenecks intensify, the core value of its business model is shifting from low-carbon distributed energy toward fast-available AI power infrastructure. This provides the foundation for analyzing the company’s investment logic.
6 Investment Logic: Bloom Energy’s Core Alpha
Bloom Energy’s investment case is not based on SOFC becoming a mainstream power source for the U.S. grid. Rather, the company sits in a high-growth, low-penetration niche with scarce public-market exposure, while its product directly addresses the “time-to-power” bottleneck facing AI data centers. Its core alpha can be summarized in several areas.
First, scarcity value in the U.S. equity market. Within the AI power value chain, there are many listed companies exposed to gas turbines, utilities, grid equipment, and energy storage. However, there are very few U.S.-listed companies with SOFC as their core business and direct exposure to onsite power demand from data centers. Bloom therefore serves as one of the most direct public-market vehicles for investors seeking exposure to the theme of AI data center power shortages, fast-deployable onsite generation, and SOFC adoption.
Second, data centers are upgrading the use case. Bloom was previously viewed mainly as a distributed clean energy provider, serving commercial buildings, retail facilities, telecom infrastructure, and industrial customers. As AI data center power demand rises, the value proposition is shifting from low-carbon distributed energy to fast access to usable power. Data center customers have larger loads, stronger purchasing power, and greater sensitivity to delivery timelines, which could increase Bloom’s order size and revenue elasticity.
Third, modular SOFC directly addresses a clear power bottleneck. Compared with large gas turbines and grid expansion, SOFC’s advantage is not the lowest absolute cost. Its strength lies in shorter construction cycles, modular deployment, proximity to load, and fuel flexibility across natural gas, biogas, and hydrogen. When grid access is delayed, turbine delivery is constrained, and data centers need to come online quickly, Bloom’s Energy Server is well positioned for this specific use case.
Fourth, customer validation and project experience create first-mover advantages. Bloom has accumulated project experience across data centers, commercial facilities, industrial sites, and critical infrastructure. For data center customers, onsite power is not just an equipment purchase; it requires reliability, operations and maintenance, permitting, fuel access, and system integration. Existing deployments and customer references can reduce adoption barriers and improve Bloom’s credibility in large project bids.
Fifth, capacity expansion raises the revenue ceiling. SOFC remains a low-base industry, and production capacity is closely tied to revenue conversion. Bloom’s existing capacity and expansion plans provide the foundation for larger order intake and delivery. If data center orders continue to materialize, capacity release could become an important driver of revenue growth and operating leverage.
Sixth, the valuation framework may be reset. The market previously viewed Bloom mainly as a fuel-cell or clean energy equipment company. As AI power bottlenecks intensify, the company may be reclassified as an AI power infrastructure asset. This is not merely a narrative shift; it depends on improving revenue visibility, higher-quality customers, larger order sizes, and stronger delivery execution.
Overall, Bloom Energy’s investment logic can be summarized as follows: as U.S. AI data centers face power shortages and delayed grid expansion, SOFC is being repriced as a fast-deployable onsite power solution for high-value use cases. Bloom, as a scarce U.S.-listed SOFC platform, offers multiple sources of alpha, including thematic exposure, customer validation, capacity expansion, and potential valuation-framework reset.
7 Risk Analysis
First, valuation volatility risk. Bloom Energy’s recent share price performance has already reflected high expectations for AI data center power shortages, SOFC adoption, and future order growth. If order conversion, revenue recognition, or margin improvement falls short of market expectations, the stock may face meaningful valuation compression.
Second, project execution risk. Revenue recognition depends on multiple steps, including equipment production, customer-site construction, fuel access, permitting, installation, commissioning, and final acceptance. Data center projects are typically large in scale, and any delivery delay could create volatility in revenue timing and affect market confidence in the company’s execution capability.
Third, cost and profitability risk. SOFC systems still face relatively high manufacturing costs. High-temperature materials, stack production, interconnects, thermal management, and system integration all require strong cost control. If cost reduction progresses more slowly than expected, or if yield and supply-chain management fall short during capacity expansion, gross margin and cash flow may come under pressure.
Fourth, technology reliability and durability risk. SOFC systems operate under high-temperature conditions and may face stack degradation, material aging, thermal-cycle stress, and higher maintenance costs. If actual system life, availability, or operating costs fail to meet customer expectations, adoption among large data center and industrial customers may slow.
Fifth, substitution risk from competing power pathways. Gas turbines, backup power systems, storage, microgrids, utility expansion, and longer-term nuclear solutions may all compete with SOFC in different scenarios. If turbine bottlenecks ease, grid expansion accelerates, or alternative onsite power solutions become more competitive in cost and delivery, SOFC’s addressable market could be constrained.
Sixth, policy and fuel-price risk. Bloom’s systems can use natural gas, biogas, or hydrogen, and project economics are affected by natural gas prices, electricity prices, carbon policy, subsidies, and customer ESG requirements. A sharp increase in fuel prices or a reduction in policy incentives could reduce project returns.
Seventh, customer concentration and order visibility risk. AI data center orders can be large, but project timing may be affected by customer capex plans, campus construction progress, financing arrangements, and power planning. If major customer orders are delayed, cancelled, or become overly concentrated, the stability of Bloom’s revenue growth could be affected.
Disclaimer:
The Information presented above is for information purposes only, which shall not be intended as and does not constitute an offer to sell or solicitation for an offer to buy any securities or financial instrument or any advice or recommendation with respect to such securities or other financial instruments or investments. When making a decision about your investments, you should seek the advice of a professional financial adviser and carefully consider whether such investments are suitable for you in light of your own experience, financial position and investment objectives. The firm and its analysts do not have any material interest or conflict of interest with any stocks mentioned in this report.
IN NO EVENT SHALL SAHM CAPITAL FINANCIAL COMPANY BE LIABLE FOR ANY DAMAGES, LOSSES OR LIABILITIES INCLUDING WITHOUT LIMITATION, DIRECT OR INDIRECT, SPECIAL, INCIDENTAL, CONSEQUENTIAL DAMAGES, LOSSES OR LIABILITIES, IN CONNECTION WITH YOUR RELIANCE ON OR USE OR INABILITY TO USE THE INFORMATION PRESENTED ABOVE, EVEN IF YOU ADVISE US OF THE POSSIBILITY OF SUCH DAMAGES, LOSSES OR EXPENSES.