Eos Energy Enterprises, Inc. (EOSE) | The Buildout — AI Infrastructure

The Verdict

Eos Energy Enterprises manufactures zinc-based battery energy storage systems for grid-scale and commercial applications, and sells into the grid rather than into the data center itself. The flagship Z3 battery is positioned for mid-duration applications, which the FY2025 10-K links particularly to AI and data center markets, and the company emphasizes safety and earth-abundant materials. The connection to the buildout is indirect: data-center load growth pulls grid capacity forward, and market and state rules reward resources that hold output across a full system need rather than a short peak. Management describes two placement paths — storage co-located behind the meter at a data center, which is still in qualification, and grid-level or generation-adjacent storage in regions where data centers are being built. The exposure is second-derivative, and substitutes are plentiful: the criticality read is that data-center development would likely shift to alternative long-duration or lithium-ion storage providers if Eos could not deliver.

Earnings

Eos reported second-quarter 2026 revenue of $68.8M, up 351% year over year and 21% sequentially, the highest quarterly revenue to date. Gross loss was $48.8M, with adjusted gross margin of negative 62%. Adjusted EBITDA loss was $71.4M, a negative 104% margin, and the net loss was $276M, driven primarily by noncash fair-value adjustments related to the capital structure. Backlog reached a record $807M, six customers placed orders (four new, two repeat), and first-half revenue of just under $126M already exceeded all of the prior year's revenue.

Management tone: The second-quarter call was framed as a deliberate trade-off: record shipments and a record backlog alongside a guidance cut, with management calling the cut a business decision rather than an operating surprise. Management gave the reduction a mechanical cause — Line 1 being down during its move and upgrade — volunteered the roughly 80% revenue concentration in prepared remarks, and volunteered the ex-tariff material-cost figure of 1% sequential rather than burying the 10% number. Several prior specifics were not restated: the end-2026 adjusted EBITDA timeline, Indensity shipping and the TURBINE-X joint development agreement drew no update. The Frontier ownership percentage moved from about 49% to about 36% between calls and was not reconciled, and management declined to rank the co-location and grid-level data-center pathways, saying both were moving.

Management Guidance

Management tightened full-year 2026 revenue guidance to $300M to $350M from $300M to $400M, saying the upper end was reduced by $50M because Line 1 would be down during its move and upgrade into Thorn Hill. The stated shape is the second half above the first half and the fourth quarter above the third, with the bottom of the range requiring roughly $50M of second-half growth over the first half. On margin, management points to "a clear path to over 72 points of adjusted gross margin improvement over the next 12 months, assuming we execute our plan and achieve expected production volumes," split into roughly 25 points from materials, 20 from conversion, 20 from project and field services and 8 from yield; a Q&A reference puts gross margin at 10% by the second quarter of 2027. Adjusted EBITDA is guided as improving with operating leverage, with execution and volume growth determining the pace. The Thorn Hill consolidation targets an additional 10% to 15% reduction in conversion costs with an estimated payback of about nine months, and the second-year Department of Energy tranche was expected by quarter end subject to loan-agreement conditions.

Trajectory

Revenue is growing fast off a small base and the growth rate is decelerating as the base builds: revenue went from $10.5M in the March 2025 quarter to $57.0M in the March 2026 quarter and $68.8M in the June 2026 quarter, with year-over-year growth easing from 445% to 351% and the sequential dollar increment at about $11.8M. Volume is the driver — cube deliveries rose 207% year over year and 20% sequentially — and unit costs are falling: direct labor per cube down 20% sequentially, manufacturing overhead per cube down 4% sequentially (about 16% at Turtle Creek standalone), and scrap dollars down 63% on the same volume, though material cost per cube improved only 1% sequentially once the tariff benefit is excluded. On the reported figures, gross margin moved from negative 234.7% in the March 2025 quarter to negative 71.0% in the June 2026 quarter, while EBITDA (operating income plus D&A) widened from a $73.1M loss to a $77.7M loss; management attributes the dollar pressure to Thorn Hill under-absorption and field costs and expects the field-cost pressure to diminish significantly by the fourth quarter.

The Model

The model projects FY+1 revenue of $322.5M with an EBITDA loss of $295M, a negative 91.45% margin, and FY+2 revenue of $575.0M with an EBITDA loss of $80M, a negative 13.85% margin. The near term is anchored by converting backlog into shipped product, with Thorn Hill's Line 2 ramp and the Line 1 relocation setting the volume. The FY+2 step-up depends on that conversion continuing and on the itemized margin bridge — materials, conversion, project and field services, and yield — showing up in reported results. The FY+2 revenue spread across the model's runs is 19%, from $540M to $650M.

Projections are the median of 4 independent model runs.

Management tightened full-year 2026 revenue guidance to $300M to $350M from $300M to $400M, saying the upper end was reduced by $50M because Line 1 would be down during its move and upgrade into Thorn Hill. The stated shape is the second half above the first half and the fourth quarter above the third, with the bottom of the range requiring roughly $50M of second-half growth over the first half. On margin, management points to "a clear path to over 72 points of adjusted gross margin improvement over the next 12 months, assuming we execute our plan and achieve expected production volumes," split into roughly 25 points from materials, 20 from conversion, 20 from project and field services and 8 from yield; a Q&A reference puts gross margin at 10% by the second quarter of 2027. Adjusted EBITDA is guided as improving with operating leverage, with execution and volume growth determining the pace. The Thorn Hill consolidation targets an additional 10% to 15% reduction in conversion costs with an estimated payback of about nine months, and the second-year Department of Energy tranche was expected by quarter end subject to loan-agreement conditions.

What Could Go Right — and Wrong

Looking Ahead

The next twelve months are about conversion. Thorn Hill is the center of it: Line 2 is adding shifts, Line 1 is being relocated and upgraded, and the company targets 10% to 15% additional conversion-cost savings with an estimated payback of about nine months. The second-year Department of Energy loan tranche was expected to close by quarter end, subject to loan-agreement conditions, and more than 200 MWh of additional fleet is expected online over the next six months on current customer project schedules. Frontier Power USA's deployment engine is funded with about $263M of equity and an estimated $1B of supported project deployment, but its first projects are not expected online until the third quarter of 2027, and the TURBINE-X joint development agreement targeting 2 GWh points to initial deployments in 2027.

Financials

Annual Summary

Key Ratios (Trailing)

The Company

Eos Energy Enterprises manufactures zinc-based battery energy storage systems for grid-scale and commercial applications. Its flagship Z3 battery is positioned for mid-duration applications, and the FY2025 10-K links it particularly to AI and data center markets; the company emphasizes safety and the use of earth-abundant materials. The product set also includes the Znyth BESS core system, the earlier-generation Gen 2.3 BESS, the DawnOS controls architecture with module-level battery management, and Indensity, a system configuration sold with a nameplate-for-life guarantee and no augmentation. Management groups buyers into four types — energy providers and independent power producers, regulated utilities, large energy consumers such as data centers, and "Assurance" customers in defense and critical infrastructure — and summarizes the shift as "Buyers are no longer procuring just a storage system, they're procuring hours."

The company manufactures in Western Pennsylvania and reports a single operating and reportable segment, so there is no product-line revenue breakout. The 10-K property table lists Turtle Creek, Pennsylvania (254,575 sq ft), Warrendale, Pennsylvania (432,000 sq ft), Edison, New Jersey (63,000 sq ft, research and development) and a Pittsburgh office (41,000 sq ft). The Thorn Hill facility that is central to the 2026 calls and press releases is not named in that table, and the source treats its relationship to the listed sites as inferred. In May 2026 the company also became a minority equity holder, technology supplier and long-term service counterparty in Frontier Power USA, a project-development and investment platform; management describes the long-term service agreements as covering up to 25% to 30% of total capital expenditure over a 20-year life.

Business Segments

Competitive Landscape

The FY2025 10-K places Eos against two groups: lithium incumbents — Fluence Energy, Panasonic, Samsung Electronics, LG Chem, Tesla, BYD, Sungrow and Contemporary Amperex Technology — and longer-duration competitors ESS Inc., Enervenue, Ambri and Form Energy. The company positions itself on duration, cycling durability, safety and domestic content, describing the 8-hour-plus band as "the duration band where our economics separate from incumbent technologies," and it points to independent fire testing and a below-25-year service attach. The evidence reads this as a differentiated but unproven-at-scale position rather than a demonstrated moat: the fire-testing result, the cycling data and the domestic-content alignment are real points of difference, while the revenue base is small and margins are negative. The source also flags a conflict — Fluence appears as a competitor in the 10-K's own language and as a customer in a lower-confidence wiring-map link.

Supply Chain

Eos buys components, materials and contractor work for a zinc battery and sells into utilities, independent power producers, large energy consumers and defense buyers. No neighbor transcript in the source set names Eos.