Zinc bromide has a real place: a different job, not a cheaper LFP

The niche is small next to the market and large next to Eos. It serves buyers that lithium cannot serve well, where fire code, footprint, climate or FEOC rules take LFP off the table. The open question is whether those rules hold, not whether zinc bromide can beat LFP on $/kWh.

Desk note on Eos Energy (EOSE) · 20 November 2026

Small for the market, large for Eos

The US market adds over 110 GWh a year by 2030. The non-lithium-addressable niche is 10 to 15 GWh of that. Each square is 1 GWh.

Niche, low case (10 GWh)Up to the high case (15 GWh)Rest of 2030 annual installs

Market: SEIA/Benchmark ESMO, "over 110 GWh" annual by 2030. Niche: this report's estimate. Eos: Turtle Creek annualized rate in June 2026, Q2 2026 call.

97%of 2026 US installs are lithium-ion, 92% of those LFP
$2.5–3.75Ba year: the 10–15 GWh niche at about $250/kWh battery-level prices
55% → 75%non-FEOC cost share a storage project needs, 2026 to 2030
Dec 24 · Dec 31DOE bulk-power rules and Treasury FEOC tables are due

Executive summary

  1. The verdict: yes, it has a place. Zinc bromide is a viable niche technology, not a failed LFP rival. Its buyers are sites where lithium is ruled out or penalised, so it competes on fit, not on $/kWh. Its durable edge is being American-made and non-flammable while the rules favour exactly that.
  2. Where it wins. Fire-code-constrained indoor and urban sites, weak-grid 8 to 12 hour duty, extreme climates, and procurement that needs US-made, non-FEOC content.
  3. Where it doesn't compete. Plain 4-hour arbitrage on open land is LFP's market, won on cost and 85–92% efficiency, with sodium-ion coming. On full-depth cycles Eos trails by about 6–13 points against its 79% fleet average. That gap decides LFP's market and matters much less in the niche.
  4. How big: small for the market, large for Eos. 10–15 GWh a year by 2030 is about a tenth of a 110+ GWh market, but roughly $2.5–3.75B a year at battery-level prices and 7 to 10 times Eos's mid-2026 production run-rate of about 1.5 GWh.
  5. Price and efficiency. Eos's battery price, about $250/kWh by third-party estimates, is close to US LFP turnkey ($219/kWh, a broader scope) and far below the $449 10-hour benchmark. The efficiency gap narrows on partial and long-duration cycles, the duty that defines the niche.
  6. The real risks. Whether the rules hold, and whether Eos wins the niche against other non-lithium options (iron-air, vanadium flow, later US-made sodium-ion). Out-competing LFP is not the test. The compliance premium is likely strongest in 2027 to 2029, then contested.
  7. Next catalysts. DOE's EO 14421 rules (Dec 24) and Treasury's FEOC safe-harbor tables (Dec 31) decide how wide the moat is. BAAJA, which passed the Senate 62–33 on Nov 19, speeds the buildout but does not touch FEOC.
Use caseFitWhy
Indoor or urban, NFPA 855 boundFitsDensity, no propagation
Weak grid, 8–12 hFitsNo mid-life augmentation
−20 to +50 °C, no coolingFitsLower auxiliary load
Non-FEOC, US-made procurementFitsClears tax and procurement rules
Open-land 4 h arbitrageLFP's marketCost and efficiency decide

The thesis: a viable niche built on safety and compliance, not on cost

Zinc bromide works. It does a different job from LFP, and its niche lasts as long as the constraints and rules that create it.

Zinc bromide has a real niche. It's not a fool's errand, but it's a narrow one — and the width of that niche is the whole bet.

Zinc bromide is a compliance product wearing a technology costume. On chemistry alone it loses to LFP and sodium on every metric that matters for cost. Its only durable edge is that it's American-made and non-flammable in a market where the rules are being written to favor exactly that. So the bet isn't on zinc — it's on whether FEOC and bulk-power restrictions stay tight through twenty twenty-nine. If they do, the niche is real and worth a position. If they loosen, it's a dead end.

The verdict on whether to wait: the niche is real but it's a safety-and-compliance niche, not a cost niche. That means it grows with regulation and urbanization, not with falling cell prices. If FEOC thresholds and bulk-power rules tighten as expected through twenty twenty-nine, the addressable market for non-lithium US-made storage expands. If they loosen or get challenged, Eos loses its moat and competes purely on chemistry — where it loses on efficiency and cost. Gross margin at negative fifty-three percent means they're still subsidizing adoption to build the reference base. The question isn't whether zinc bromide works. It does. It's whether the compliance premium persists long enough for Eos to reach positive margins before sodium and LFP scale closes the gap from below.

Where it fits, and where it does not compete

Four jobs where lithium's virtues stop mattering, and one where they decide everything. The niche is defined by the buyer's constraint, not by a price race.

Exhibit 1. Four buckets add up to 10–15 GWh a year by 2030

Non-lithium-addressable US storage demand in 2030, GWh per year. Solid bars are the low case; hatched tops extend to the high case.

Bucket2030, GWh/yrBasis

Estimates built from SEIA/Benchmark ESMO Q3–Q4 2026, BNEF US storage outlook (Oct 2026), published C&I and microgrid forecasts, and state LDES procurements. Dollar value at about $250/kWh battery-level pricing.

Where zinc bromide is a clear choice: indoor or urban sites where NFPA 855 fire separation kills lithium density. A lithium four-hour system might need two to three times the footprint once you add spacing, suppression, and blast walls. Eos's five hundred megawatt-hours per acre claim matters most exactly there. Second, sites with no grid or weak interconnection where you need eight-to-twelve-hour duration and the mid-life augmentation cost of LFP — replacing a third of your cells at year ten or twelve — compounds into a real penalty. Third, extreme climates, minus twenty to plus fifty with zero active cooling, which is auxiliary load lithium can't match without burning power to stay warm. Fourth, and this is the one that ties back to everything we discussed: US-made, non-FEOC, IRA-compliant domestic content with a manufacturing credit behind it. In a market where Chinese cells are being legislated out, a Pennsylvania-made non-lithium system clears procurement and tax hurdles that Korean LFP still has to navigate.

Where it's not: pure four-hour energy arbitrage on open land in Texas or California. There, LFP at two hundred dollars per kilowatt-hour with ninety-plus percent efficiency and a mature supply chain wins on pure math, and sodium is coming to undercut it further. Eos's seventy-nine to eighty-one percent round-trip efficiency is a structural handicap in any application where every point of efficiency is a point of revenue.

Exhibit 2. Eos packs two to four times the energy per acre of incumbent systems

Stored energy per acre of site, MWh.

Eos Indensity brochure (Jan 2026): incumbents about 250 MWh/acre, Indensity target 1 GWh/acre.

Efficiency, degradation and price: the gap is smallest where the niche lives

Field data supports the niche. The efficiency gap is widest on full-depth short cycles, LFP's home turf, and shrinks on the duty cycles that define the niche. Eos's price sits near US LFP, far below the 10-hour zinc benchmark.

Round-trip efficiency:

Fleet data (full cycles): ~79% average on the Z3 fleet (up from 78% prior quarter). This includes systems not yet fully on DawnOS.

Field examples under more standard 20-80-20% SoC cycling show averages around 84.6%, with best cycles reaching the high 80s to low 90s (one 12-hour example at 90.9%). Company materials claim mid-80s to low 90s depending on depth of discharge and duration — higher when you avoid full deep cycles or stretch duration.

LFP system-level AC round-trip efficiency is typically 85–92%. Eos trails on aggressive 4-hour full-depth cycles by roughly 5–10 points. The gap narrows (and can reverse on a TCO basis) in longer-duration or partial-cycle duty cycles where LFP’s cooling load and degradation matter more. Eos’s near-zero auxiliary cooling and claimed ~3.6% degradation over 25 years are the offsets.

Exhibit 3. The efficiency gap is widest in LFP's market and smallest in the niche's

Eos round-trip efficiency (red dot) against the typical LFP AC range of 85–92% (grey band). The scale runs from 60% to 100%.

60%70%80%90%100%

The gap is the LFP range minus the Eos value. Against the 79% fleet average it is 6–13 points across the full LFP range. Eos values come from Eos's Q3 2025 release and its Q1–Q3 2026 calls; "about 90%" is Eos's description of its best long-duration cycles. LFP range: typical AC specification.

Exhibit 4. Every Eos efficiency figure, with its duty cycle

MeasureRTEDuty cycleWhen

Eos earnings calls (Q1–Q3 2026), Eos Q3 2025 release, CEC-500-2023-052.

Exhibit 5. Eos's battery price sits near US LFP, far below the 10-hour benchmark

$/kWh. The bars measure different scopes: read the note below the chart.

PNNL-33283 (2022): fully installed 10 MW/10 h zinc system, in 2021 dollars. Eos: Iceberg Research, 1Q21–2Q24, battery (cube) level. LFP: BNEF 2025 turnkey system averages. A cube price leaves out the power conversion and installation that a turnkey price includes, so Eos's real gap to US LFP is wider than the bars show.

Exhibit 6. In niche duty, these offset most of what remains

FactorEos Z3LFPBasis

The hidden line item: the niche is priced in fire code, not in $/kWh

The cost comparison that matters does not appear in $/kWh. It shows up in fire code, insurance and interconnection.

Eos's real competitor isn't lithium, it's the insurance and fire-code cost of lithium. If a utility can show a risk assessor a non-propagating, non-flammable system, the avoided suppression infrastructure and the avoided interconnection delays are worth real money that never shows up in a dollar-per-kilowatt-hour comparison. That's the hidden line item that could make zinc bromide the rational choice in exactly the places where it looks expensive on paper.

How big is the niche? A tenth of the market, many times Eos

The market is huge, growing faster than forecast, and almost all LFP. Ten to fifteen GWh a year is small next to 110+, but it is roughly $2.5–3.75B a year and 7 to 10 times Eos's mid-2026 output.

Total US storage is past two hundred gigawatt-hours installed, with SEIA and Benchmark now forecasting seven hundred eight gigawatt-hours cumulative by twenty thirty, and WoodMac sees seven hundred fifteen by twenty thirty-one. Annual additions hit roughly seventy-four gigawatt-hours this year. But ninety-seven percent of twenty twenty-six installs are lithium-ion, ninety-two percent LFP. Non-lithium is a rounding error today.

The niches break into four buckets. First, long-duration energy storage — eight hours and up. Flow batteries are the fastest-growing category in the US at twenty-three point six percent CAGR through twenty thirty-two, but from a tiny base. Vanadium flow is over seventy percent of installed US flow capacity, and Dairyland Power just ordered forty-three megawatt-hours of Invinity units in August. California's LDES mandates and Texas utility contracts are pulling duration up, but the absolute numbers are still low single-digit gigawatt-hours annually. Second, commercial and industrial behind-the-meter — the fastest-growing segment overall at twenty-six point five percent CAGR, going from three hundred seventy million to nearly four billion in capex by twenty thirty-five. Urban, indoor, rooftop, and constrained-site deployments live here, where fire code and footprint dominate. Third, data-center on-site storage — BNEF sees four point nine gigawatts and seven point four gigawatt-hours this year, rising to six point three gigawatts and eight point four in twenty twenty-seven, with hyperscalers like Google explicitly experimenting with zinc and iron-air for multi-day backup. Fourth, microgrids and off-grid — growing at twenty-two percent CAGR, the smallest but most chemistry-agnostic bucket.

Add it up generously and the non-lithium-addressable niche is maybe ten to fifteen gigawatt-hours per year by twenty thirty, out of a hundred-plus. That's a real market — tens of billions over five years — but it's not a mass market. It's the size of a handful of large utility contracts.

Exhibit 7. Annual US installs have nearly tripled since 2023

GWh installed per year; 2026 and 2030 are forecasts.

SEIA/Benchmark ESMO Q1 and Q4 2026. 2030 is "over 110 GWh".

Exhibit 8. Every 2026 forecast revision went up

SEIA/Benchmark forecast of cumulative US storage by 2030, by report edition, GWh.

ESMO Q1–Q4 2026 press releases. Q1 and Q2 were stated as "over 600" and "over 610".

Exhibit 9. Non-lithium is 3% of 2026 installs: room to grow from a small base

Share of 2026 US installs by chemistry.

LFP, about 89%Other lithium-ion, about 8%Non-lithium, about 3%

ESMO Q4 2026: Li-ion over 97%, LFP 92% of Li-ion (so about 89% of the total).

Exhibit 10. Data-center storage grows sixfold in two years

US on-site data-center battery storage, GWh per year.

BloombergNEF US energy storage outlook, Oct 2026.

Exhibit 11. The four buckets, side by side

BucketSize or growthStageWhy zinc bromide fits

History says niches persist, and this one is big enough for one company

Non-dominant technologies survive where their defect stops mattering. The test was never beating lithium, only whether the niche can carry a company.

History says niches persist. Lead-acid still owns automotive starting batteries even though lithium-ion took EVs — same energy storage domain, different job, never displaced. Nickel-cadmium survives in aviation where temperature extremes and cycle life beat lithium. Vanadium flow batteries have held a long-duration niche for twenty years without ever threatening lithium on cost. Sodium-sulfur ran grid-scale in Japan since the nineties. Pumped hydro is still ninety percent of global storage capacity. Thin-film solar never matched silicon efficiency but carved out utility-scale land-rich markets where temperature coefficient and manufacturing simplicity mattered more.

The pattern is consistent: the dominant technology wins the mass market, and the non-dominant one survives where its specific defect is irrelevant or its specific virtue is decisive. Zinc bromide's defect — efficiency — is irrelevant in a backup or long-duration safety application. Its virtue — non-flammability, US manufacturing, no augmentation — is decisive exactly where lithium's virtues don't matter.

The risk isn't that niches disappear. It's that the niche stays too small to carry a public company. Eos needs the compliance-driven urban and indoor market to be tens of gigawatt-hours, not five. That's the number worth watching.

Exhibit 12. Six survivors, one rule: they found the job where their defect stops mattering

TechnologyLost the mass market toNiche it keptDefect made irrelevantVirtue made decisive

Each technology kept the job where its weakness stopped mattering.

Policy: the moat is written in Treasury and DOE rules, not in Congress

A Democratic Congress cannot override a veto alone, permitting reform leaves FEOC untouched, and the rules that matter tighten on a fixed schedule.

On the political side, the picture is more interesting than it looks. Democrats took both chambers — fifty-six to forty-four in the Senate, roughly two hundred fifty-eight to one hundred seventy-three in the House, their largest majority since two thousand eight. But Trump is still president through January twenty twenty-nine with the veto, so nothing passes without bipartisan support or a veto-proof margin Democrats don't have.

BAAJA — the Bipartisan American Affordability and Jobs Act — passed the Senate sixty-two to thirty-three on November nineteenth and is heading to the House in the lame-duck session. Fifteen Democrats voted yes, including Schumer. It's permitting reform: it overhauls NEPA, the Clean Water Act, the Endangered Species Act, and historic preservation law to speed up solar, transmission, gas pipelines, and data centers. It includes "forever permits" that lock approvals in place. Progressive Democrats — Huffman, Tlaib, Wyden, Merkley — opposed it and want to wait for full Democratic control in January. Trump endorsed it. If it passes, it accelerates the buildout of everything including storage, but it doesn't touch domestic-content rules or FEOC thresholds directly.

The FEOC and bulk-power restrictions are the ones that matter for zinc bromide, and those are executive and statutory, not easily reversed by Congress. OBBBA's storage credit runs full value through twenty thirty-three construction starts, with the material-assistance threshold rising from fifty-five percent in twenty twenty-six to seventy-five by twenty thirty. Section forty-five-X manufacturing credit is intact with its twenty thirty-to-twenty thirty-two phase-out. EO fourteen four twenty-one's DOE implementing rules are due December twenty-fourth — about a month out — and the FAQ explicitly says the key definitions, covered entities, and what counts as foreign-produced are still undecided. Treasury's FEOC safe-harbor tables are due December thirty-first.

Democratic priorities for twenty twenty-seven center on rolling back tariffs, oversight of the administration, and affordability. That cuts toward loosening trade restrictions, which would pressure the tariff wall that protects domestic non-lithium makers. But national-security framing around Chinese grid equipment has bipartisan support, and the bulk-power order is an emergency declaration Trump can renew. The realistic read: FEOC thresholds keep tightening on schedule through twenty thirty regardless of Congress, the bulk-power rules likely land in December and bite Chinese suppliers, but a Democratic Congress with tariff-rollback ambitions could create pressure to ease the broader trade wall by twenty twenty-eight. The compliance premium for US-made non-lithium is probably strongest twenty twenty-seven through twenty twenty-nine, then contested.

Exhibit 13. The 120th Congress: large Democratic majorities, no veto override

Seats from January 2027 to January 2029, against the two-thirds needed to override a veto. Policy that passes needs Republican votes, so the FEOC and bulk-power rules stay in executive hands.

AP. House: 257 Democratic seats called as of Nov 19, a handful still being counted.

Exhibit 14. The FEOC bar rises every year whatever Congress does

Minimum share of storage project cost from non-prohibited sources, by construction-start year (%).

OBBBA §7701(a)(52); Treasury Notice 2026-15. The 45X manufacturing credit phases down to 75%, 50% and 25% of full value in 2030–2032 under existing law.

Exhibit 15. Two deadlines in the next six weeks set the width of the moat

    Conclusion: a viable niche, and the rules decide how big it gets

    Zinc bromide has a place. It is not a cheaper LFP and does not need to be. It serves buyers lithium cannot serve well, in a niche that is small for the market and large for Eos.

    Zinc bromide does not compete with lithium head-on. Its market is a defined set of sites where fire code, footprint, climate and FEOC rules make LFP's cheaper $/kWh irrelevant, and that set can grow into tens of GWh a year. The policy calendar favours the thesis in the near term: the FEOC thresholds are statutory and rise on schedule, and the bulk-power order is an emergency power the administration controls. The real competition is inside the niche: iron-air for multi-day duty, vanadium flow where footprint is cheap, and later US-made sodium-ion. The longer-dated risk is a trade-policy thaw from 2028.

    Price and efficiency. At the battery level Eos's price is close to US LFP, and its efficiency deficit is largest exactly where it does not compete, on full-depth 4-hour arbitrage. In the niche's duty cycles (partial cycling, long duration, no cooling), the gap narrows to a few points or less, and lower auxiliary load and claimed flat degradation offset part of what remains. That strengthens the economic case at the margin, but the moat is still the rules.

    Exhibit 16. The compliance premium window

    How strong the premium for US-made non-lithium is likely to be, by year, given the policy calendar. Qualitative.

    What to watch, and what each outcome means

    SignpostWhenSupports the thesisBreaks the thesis

    Sources