Authority, Not Obligation: The Defense Battery Demand Signal Has No Floor

Begin at the cell spec. It carries more of the argument than a table of numbers usually does.

The Defense Industrial Base Consortium's June 2026 cylindrical-cell solicitation asks industry to stand up domestic annual production within two years. The bands:

Three hundred cycles describes a consumable. Automotive packs get specified at a thousand cycles and upward because the vehicle has to outlive its warranty; stationary storage runs higher still, since the asset only earns by cycling. A 300-cycle cell at 270 Wh/kg has bought energy per kilogram and paid for it in life. That is the right trade for expendable munitions, loitering systems, and dismounted power. It is the wrong trade for nearly everything that sells commercially at volume. The 1,000-cycle objective would put the cell back inside commercial territory, but it is the objective, and a first line gets built to the threshold.

Now read the accompanying Q&A. Award count and value depend on FY2027 appropriations, on what industry submits, and on government need. The government is providing "no guaranteed offtake." Commercial applications are expected to form part of the industrial model. Assembled, those three statements ask capital to underwrite a defense-shaped product on the strength of commercial demand the document never identifies, while declining to commit to the defense volume.

That reads as candor. It is an accurate description of the market's structure and the most useful line in either file. It also frames a question narrow enough to answer: what would a committed defense battery demand signal consist of, and where does the visible record sit against it?

Demand signals attach to different legal objects

The mandate architecture, meaning the statutory push toward non-Chinese cell content in defense systems, is the reason demand should theoretically exist, and its compliance date lands inside the same window a new domestic line would need for qualification. The rulemaking mechanics belong to another piece. What matters here is that nobody lends against a prohibition. Lenders lend against instruments, and instruments differ in what they attach to. Statutory authority is one object. An executed contract is a second. A budget-justification forecast is a third. Money appropriated to build the line is a fourth. Discussion of defense battery demand treats the four as one signal, and they carry entirely different weight in a credit file.

Munitions procurement is the clean comparison, and not by coincidence. It spent decades solving the problem cell manufacturers now face, which is how to get a private party to build capacity only the government will consume. PAC-3 MSE puts four separable objects in public view.

Authority with a ceiling quantity. Section 1244 of the FY2023 NDAA authorized multiyear procurement of up to 3,850 missiles. Authority permits. It does not obligate.

An executed contract naming a quantity. In September 2025 the Army announced a $9.8 billion FY2024–FY2026 multiyear award covering 1,970 missiles and associated hardware, incrementally funded subject to annual appropriations.

A published rate structure. The Army's FY2026 missile procurement justification records a minimum sustaining rate of 144 missiles per year, an economical rate of 185, and a maximum of 550. The minimum sustaining rate is the point below which the line is understood to shed its workforce and its supplier base. Publishing it declares what the government intends to defend.

Advance procurement. The same exhibit records $252 million obligated ahead of production, split evenly in support of FY2025 and FY2026 units. The government bought long-lead material before the production year opened.

Individually, none of the four guarantees anything. Together they let a lender construct a floor: a quantity named in a signed instrument, a rate the program has institutional reason to protect, and demonstrated willingness to fund inventory before need. No advanced domestic cell program in the public record puts all four objects in view. Most put none.

Ranking the signals by what a lender can underwrite

Every citation in this piece is a primary document: solicitation, budget exhibit, FAR clause, SAM.gov notice, contract announcement, SEC filing, issuer earnings release. Nothing rests on trade reporting. Several 2026 announcements carry the Department of War banner; I write DoD throughout to stay consistent with the procurement records.

Ranked by what can actually be underwritten:

1. Multiyear contract with named quantities plus a published sustaining rate.
2. Validation-contingent minimum annual quantity: a commitment that converts to a production floor on passing qualification.
3. Long-term indefinite-delivery contract sitting on an established consumption history. Rungs two and three trade against each other instead of nesting cleanly. Two is the stronger legal object resting on a conditional event; three is the weaker legal object resting on demonstrated behavior.
4. Indefinite-delivery ceiling with a nominal minimum and no history. Bankable to the minimum and no further.
5. Fully obligated, non-recurring production order with no disclosed quantity or rate. Funds the work. Commits to nothing past it.
6. Prototype or consortium agreement with no obligation beyond the prototype.
7. Announced industrial-base interest.

The DIBC solicitation lands below rung seven. It is a pre-award request for proposals issued under a consortium agreement and contingent on FY2027 appropriations. The anchoring document of this analysis sits beneath the bottom of the ladder it prompted.

What the legacy battery vehicles actually obligate

DoD does buy batteries on long-term vehicles, so the comparison is available rather than theoretical. Three May 2023 DLA awards for lead-acid 6T vehicle batteries carried three-year bases plus two option years at announced values of roughly $10.25 million to $13.94 million. A 2026 sole-source vehicle for Clarios lists estimated annual demand of 31,009 batteries against a potential value near $46.9 million. Delivery orders flow against these vehicles year over year.

Read the commitment language, not the ceiling. Under FAR 16.504, an indefinite-quantity contract has to state a minimum and a maximum, and the government's obligation runs to the minimum only. Everything above it is forecast, and forecast is not consideration. A July 2026 DLA Aviation solicitation for nonrechargeable aircraft, submarine, and ground batteries lists estimated annual demand across seven stock numbers totaling 15,769 units, with the contract minimum set at 5 percent of total annual-demand value. The Common Wearable Battery vehicle projected as many as 944,787 batteries against an initial minimum ordering quantity of 200 first-article-test units per awardee. Discrete lot buys behave identically: Navy notices for BA-5390A/U primary batteries specify lots of 2,800 and 2,000 units, which are purchases, not rates.

What finances a 6T plant, then, is reorder history: a stock number consumed for a decade or more against fielded platforms. The guaranteed minimum is not doing that work.

That is an inference about how capital providers read the record rather than a documented finance term, and I label it as inference. It survives the obvious test. The guaranteed minima on these vehicles are nowhere near large enough to underwrite a plant, and the plants exist. Something other than the minimum is carrying the credit.

Advanced domestic cells have neither leg. No guaranteed floor, and no consumption history, because the platforms that would consume them are not yet fielded at volume. That is a structural condition. Exhortation does not move it.

Packet Digital sits at the top of the advanced-cell record

The most production-shaped advanced-cell instrument in the public record is the Navy's April 6, 2026 order to Packet Digital: $26,999,999 firm-fixed-price, SBIR Phase III and APFIT, covering pilot prototyping, raw materials, pilot production, LRIP-1 and LRIP-2 pouch cells, testing, and integration into multiple pack designs through April 2028. The full announced amount was obligated at award. A May 4 order adds $9,799,991 for equipment commissioning, raw materials, pack testing, and pilot- and low-rate cells for the Navy Advanced Family of Batteries through November 2027.

Roughly $36.8 million, fully obligated, reaching low-rate initial production. That is a real instrument, and it advances the pathway this publication traced in Issue 1. Both figures also stop just short of round numbers, which reads as award sizing derived from APFIT and Phase III conventions rather than from any count of cells the Navy needs.

The omissions carry the observation. Neither notice discloses a cell count, pack quantity, MWh figure, LRIP lot size, option quantity, or annual rate, and neither does the USAspending record for the April order. Nothing is redacted. The quantities are simply absent at every level of disclosure.

That absence is the finding, because Phase III and APFIT get read habitually as production commitments and are structurally something else. Phase I buys a feasibility study and Phase II buys prototype development; neither is a procurement instrument in any form. Phase III is contracting authority, which lets an agency buy from a prior performer without further competition. It obliges no one to buy anything. APFIT applies procurement dollars to production-ready technology in one-time awards of $10 million to $50 million, and its own program guidance states that an award stands alone and guarantees neither follow-on APFIT funding nor program procurement. A September 2024 Navy award handed Intramicron $7.62 million to procure high-power battery modules, with the number of modules and the rate unstated. Same shape.

My claim that essentially all publicly visible advanced-cell activity sits on rungs five and six is an inference from the survey behind this piece: DoD daily contract announcements, SAM.gov solicitation and award notices, USAspending prime-award and IDV records, and issuer filings, through July 31, 2026. It excludes a systematic count of Phase I and Phase II battery awards. Adding one lengthens the bottom of the ladder. It moves nothing up it.

Qualification cost, which an earlier piece in this series treated as the binding constraint on conversion, is therefore incurred against a revenue stream of undisclosed magnitude. That is a financing problem before it is an engineering one.

Rung two is not hypothetical

In May 2026 DoD announced a prospective arrangement under which Castelion, following successful test and validation, would receive a two-year multiyear procurement contract for a minimum of 500 Blackbeard missiles annually, options potentially extending to five years, with firm-fixed material unit costs for 2027 through 2029. The announcement reports no executed production award, no obligation, and no value at that stage. I would not treat it as one.

What it establishes is availability of form. A minimum annual quantity conditioned on validation is a structure a program office can offer a new entrant. Whether the form maps onto cells raises a genuine sequencing question, since cell qualification runs longer and more conditionally than missile flight test. The existence of the form is not in doubt.

The Korean makers' own disclosures are the diagnostic

Forge Nano's June 2026 SEC-filed disclosure describes a proposed North Carolina cylindrical-cell facility at 3 GWh, $300–330 million expected investment, a $100 million DOE grant, and a 2028 manufacturing target, with Samsung SDI providing technical support and conditionally purchasing cells beginning in 2028. Volume: unstated. Defense and critical infrastructure appear as intended markets, not as named counterparties. A conditional private purchase carrying no quantity is rung five at best, and it is a commercial arrangement rather than a defense one. Issue 6 reached the same classification on the compliance side. Nothing in the quantity record has moved since.

The diagnostic is what these firms disclose when they do hold volume. LG Energy Solution's January 2026 earnings release quantified more than 300 GWh of 46-series cylindrical backlog and roughly 140 GWh of stationary-storage backlog. SK Innovation's Q3 2025 investor presentation disclosed up to 7.2 GWh to a U.S. storage developer across four years from 2026. Committed volume in GWh is house practice in these materials. Across quarterly investor presentations, earnings releases, and annual reports for the preceding eighteen months, neither company named a defense customer, contract value, capacity allocation, or volume.

The absence sits inside a disclosure practice that would surface the number if the number existed, which is what makes it evidence rather than a gap in my reading. It says nothing about confidential negotiations and nothing about intent. The defense pivot is real as strategy. It has not appeared as quantity.

Sizing the gap, assumptions attached

Those are DoD prime-contract obligations coded to nonrechargeable and rechargeable batteries. Query terms, for anyone reproducing it: USAspending spending-over-time, DoD prime contract obligations only, PSC 6135 and PSC 6140.

The $150–300/kWh cell-level band is an assumption for scale rather than a quoted price, and the inference is mine. At that band, 3 GWh of annual output implies annual cell revenue between one and a half and three times every dollar DoD visibly obligated for batteries of any chemistry in FY2025. The 50 MWh threshold converts to the same order of magnitude as the instruments actually observed, Packet Digital's included.

Two limitations bound both figures. The obligation number excludes cells embedded in platform contracts and industrial-base awards coded elsewhere, so read it as a floor on visible spend; public sources do not resolve battery content inside platform awards, and I will not infer volumes from program values. And the comparison converts energy to dollars through an assumed price rather than a disclosed one.

Neither limitation changes the direction. The threshold is fundable out of the demand base that visibly exists. The objective is not.

Three artifacts would change the reading

• A solicitation for an advanced domestic cell stating a guaranteed annual minimum quantity instead of a percentage of estimated demand.
• A validation-contingent minimum of the Blackbeard form applied to a cell line.
• A follow-on order against Packet Digital's LRIP work that discloses a quantity.

None has appeared. Absent one, the solicitation is asking capital to fund 3 GWh of capacity against a signal whose only unambiguous term is that no offtake is guaranteed, for a cell whose 300-cycle threshold leaves the commercial fallback largely notional.

The disclosure was accurate. That is the part to take seriously.