ALMU: The $30 Million CHIPS Deal Has an Equity Price

The 60-second read

Five things to know

1. The CHIPS proposal has an equity price. July 29 terms call for equity securities equal in value to the final award; up to $30M is conditional and is not cash already held.
2. Commercialization spending accelerated before revenue. Q4 revenue was $0.582M while research and development expense reached $2.274M; GAAP loss widened to $4.013M.
3. The cash increase came from capital markets. FY2026 public-offering proceeds net of costs were $43.478M, against $3.912M of computed free-cash-flow use.
4. Count the new owners. June-end basic shares rose 21.4% year over year; the September 11 cover reports 19,265,671 shares, above the denominator implied by provider capitalization.
5. Controls and commercial proof both remain unfinished. The annual report still describes an unresolved material weakness, while management describes customer NRE negotiations rather than binding high-volume revenue commitments.

Three scenarios, one probability-weighted number

Analyst 12-month value scenarios; enterprise values discount longer-term commercialization. These are not management forecasts.

Signposts: what would change our mind

These are Charged Alpha thresholds, not company guidance. This is our first ALMU packet; there are no previous channel signposts to grade. Dates are review checkpoints unless the issuer has announced an event.

The tape and the denominator

The reference price is $11.10 from a saved FMP after-hours trade at 19:58:09 Eastern Daylight Time on September 16, 2026, or 2026-09-16T23:58:09+00:00. The trade size was 443 shares. That is preliminary, potentially illiquid pricing, not a completed next-session close. The regular-session close was $13.43. The computed decline between those observations is 17.35%; the regular session itself rose 2.28% before this release. A viewer should not confuse that earlier gain with the earnings response.

Price is only half of capitalization. The annual-report cover identifies 19,265,671 common shares outstanding on September 11. At our reference price, basic equity value is $213.85M. Subtracting June-end cash of $56.006M gives $157.84M of enterprise value before lease adjustments. We use the latest reported cash balance, not an invented September balance. The resulting enterprise value is 35.38 times FY2026 revenue, despite the stock's sharp early decline.

FMP's saved profile reports a $245.84M market capitalization at $13.43, using a smaller implied denominator than the new annual cover. We do not carry that stale denominator into per-share valuation. Provider data remains useful for actual trade observations, volume and price history, but company filings control ownership. Likewise, the provider employee count is outdated; the annual report's dated workforce figures supersede it.

The stock's history spans a wide range, with a saved 52-week high of $31.79 and low of $10.24. That range does not create a floor or imply a return to the high. A small business funded for a large addressable market can trade on changing probabilities rather than current earnings. Our task is to make those probabilities and ownership assumptions visible, while keeping the observed trade separate from the valuation judgment.

The print: spending outruns sales

Fourth-quarter revenue fell to $582,000 from $1.317M a year earlier and $1.222M in the preceding quarter. The annual revenue total was $4.461M, down 4.37% from $4.665M. These are small absolute revenues for a company pursuing a large photonics opportunity. Government development contracts can be uneven, so one weak quarter does not establish permanent technological failure. Equally, an uneven contract base does not justify treating future commercial adoption as already achieved.

Gross profit, computed as revenue less cost of revenue, was $149,000 in the quarter, a 25.60% margin. Research and development expense was $2.274M and general and administrative expense was $2.247M. Together with cost of revenue, operating expenses totaled $4.954M. The operating loss of $4.372M was partly offset by $359,000 of interest income, producing a $4.013M GAAP net loss. Income-tax expense was zero in the table. The quarter's basic and diluted loss was $0.22 per share.

The provider calendar now agrees with the filing's $582,000 revenue. Its $575,000 estimated revenue implies a small 1.22% beat, which is much less informative than the year-over-year contraction and spending increase. Provider EPS actual of negative $0.15 matches the release's adjusted EPS, not GAAP EPS. The estimated negative $0.075 figure is therefore not compared with negative $0.22 as if both used the same accounting basis. A surprise percentage would exaggerate precision on a very small and potentially mixed-basis consensus.

The source tables list June 2026, March 2026 and June 2025 quarter columns before the two annual columns. We verified those headers before extracting values. Some machine-readable taxonomy tags swap operating expense and operating loss in older and current filings; the visible issuer statements take precedence. The accompanying extraction audit preserves those discrepancies rather than smoothing them away.

One segment, several unproven markets

Aeluma reports one operating and reportable segment. Management evaluates the business as a whole; the company does not publish separate profits for AI datacom, mobile sensing, defense, aerospace or quantum applications. We therefore show total segment revenue and operating results, while treating those applications as development opportunities. Dividing the enterprise into invented profitable divisions would give a false impression of established commercial breadth.

The platform aims to combine compound-semiconductor performance with larger-diameter substrates and manufacturing methods associated with mainstream microelectronics. Its potential attraction is scale, integration and economics for photodetectors, lasers and related devices. Those are engineering and business claims that must survive customer tests. A qualified manufacturing flow, acceptable reliability, usable yields and repeat orders matter more to valuation than a broad addressable-market estimate.

The annual report describes small-volume wafer and chip sales, engineering samples and non-recurring engineering work. It warns that early orders may not become production orders. The revenue concentration note says all customers have been government agencies, with two customers accounting for 59% and 14% of FY2026 revenue. This wording limits what can be claimed about already established commercial demand, even as management describes expanding prospective-customer engagement.

Tower Semiconductor and Sumitomo Chemical Advanced Technologies relationships may improve access to production capabilities. They do not by themselves commit end customers to buy Aeluma products or disclose a guaranteed commercial margin. Equipment purchases and additional wafer capacity can prepare the business for demand; they can also increase spending before demand is proven. We look for qualification milestones and binding NRE agreements with concrete amounts and acceptance conditions. Until those arrive, the opportunity is best represented as a funded development program with upside, rather than a scaled AI component supplier.

Profit bridge: the adjustments are real costs

The quarterly GAAP loss of $4.013M becomes a $2.705M non-GAAP loss after adding back $1.308M of stock compensation. Adding $126,000 of depreciation and amortization and subtracting $359,000 of interest income produces adjusted EBITDA of negative $2.938M. This is a transparent reconciliation, but it does not make the excluded stock compensation economically free. The business uses ownership interests to attract and retain people, and investors must assess the resulting dilution.

For the year, the $9.159M GAAP loss includes $4.519M of stock compensation. The adjusted net loss is $4.640M; after depreciation and interest adjustments, adjusted EBITDA is negative $5.216M. That reverses the prior year's small positive $186,000 adjusted EBITDA result. The expanding engineering organization and manufacturing-readiness effort explain part of the change, but the investment case still requires a credible path from higher spending to funded customer demand.

Historical GAAP comparisons need care. In Q3 FY2025, a $2.577M favorable derivative-liability remeasurement helped turn an operating loss into a GAAP profit. That was not evidence of recurring operating earnings. Conversely, the following comparisons contain different valuation marks and convertible-note discount amortization. The packet retains the full below-the-line and non-GAAP tables. We do not splice older definitions of adjusted profit into a smooth, purportedly comparable operating trend.

Cash interest is also a genuine benefit of the larger funding balance, yet it is not evidence that photonics products have become profitable. A business can report less net loss because investors supplied cash that earns interest. For evaluating commercialization, operating loss and cash spending must remain visible beside the net result. Our valuation assumes an eventual cash-generating business; it does not capitalize current interest income at a semiconductor growth multiple.

Earnings quality and unfinished controls

The most serious reporting-quality issue is the annual report's unresolved material weakness. Management says accounting and SEC-reporting expertise were insufficient for some nonroutine and complex transactions, and disclosure controls were ineffective at June 30, 2026. Hiring a chief financial officer and adding accounting staff are constructive steps, as is engagement of a third-party controls firm. The company explicitly says remediation requires implementation, an adequate operating period and testing. We do not present those plans as completed remediation.

That weakness does not itself prove the statements are wrong. Rose, Snyder & Jacobs LLP issued an opinion that the audited financial statements present fairly, in all material respects, the reported financial position and results. The smaller-reporting/emerging-growth framework does not include an auditor attestation on internal-control effectiveness. A clean financial-statement opinion and a management-identified controls weakness can coexist. Investors should understand both before relying heavily on fine-grained forecasts.

Stock compensation was 224.74% of quarterly revenue and 101.30% of annual revenue. Such percentages become extreme because the revenue base is small, but the dollar cost and ownership claims still matter. Receivable days, computed from year-end receivables divided by annual sales times 365, improved from 75.3 to 27.8. This is an endpoint ratio, not a customer-by-customer aging schedule. Government-contract milestone timing can drive it, so the decline does not alone demonstrate durable collection efficiency.

Operating cash use was smaller than the GAAP loss partly because compensation and depreciation are noncash and working capital supplied some cash. With losses on both sides, a conventional positive-profit cash-conversion ratio is not meaningful. Inventory days and minority leakage are not invented where the statements provide no corresponding reported balances. Nor do we fabricate Beneish or Altman outputs from missing or unsuitable inputs. The quality judgment rests on transparent checks, explicit missing data and the distinction between development finance and earned profitability.

Cash strength came with more shares

The June 30 balance sheet contains $56.006M of cash and cash equivalents. There is no certificate-of-deposit balance at that date; the prior year had $3.628M of cash plus $12.112M of certificates. Comparing only the cash line would overstate the economic increase in liquid resources. Total cash and certificates increased by $40.266M, while financing activities supplied $44.178M during FY2026. Operations used $3.266M and equipment purchases used $646,000.

Computed free cash flow is therefore negative $3.912M. Fourth-quarter cash flow is derived from annual totals less the first nine months: operating cash use of $1.623M and equipment spending of $207,000, or negative $1.830M of free cash flow. This does not forecast next year's burn. Management plans more people, MOCVD tools, qualification work and expanded facilities; a trailing cash-use rate can substantially understate future requirements.

The largest financing source was $43.478M in net public-offering proceeds across the year. During May and June, the ATM program issued 830,484 shares, generating $20.7M gross and approximately $20.1M net. The reported average sale price was $24.87. Raising equity at that price improved the cash resources per surviving share relative to funding at a much lower price, but it still increased ownership claims. June-end common shares rose 21.39% from the previous June.

The latest cover count is 19.265671M. Outstanding options, RSUs and warrants at June 30 total approximately 2.070M gross underlying shares, although option exercise prices and treasury-method effects matter. The 80,507 warrant count included 805 expiring July 1 at $2 and 79,702 with a $3 strike. Current option grants span exercise prices including amounts above $11.10; using the $8.40 weighted-average strike for every option would misstate treasury-method dilution; individual exercise-price and vesting details are required. The combined 1,989,119 option-and-RSU total is not an extra award pool to add again. Our base uses 20.5M model diluted shares, clearly distinguished from both basic shares and GAAP loss-period diluted shares. This is an assumption, not a cap on future issuance. The remaining ATM capacity was approximately $29.3M, and the CHIPS proposal could create additional government equity. Liquidity is a strength; fixed ownership is not promised.

Three routes, plus the price-implied hurdle

Our base fair value is $7 per share, supported by two commercialization routes at $7.33 and $6.84, whose midpoint is $7.08. A third, separately calculated balance-sheet route tests downside support; it is not averaged into a going-concern value. These are judgment tools for a development-stage business, not calibrated forecasts with statistically reliable probabilities. Very High uncertainty calls for a substantial margin of safety: our illustrative 35% discount to base value implies $4.55, and even that price would require the operating thesis to remain intact. A falling price alone cannot turn a failed commercialization case into a good investment.

Route one values a future cash-generating business. Assume FY2031 revenue of $50M, a 20% free-cash-flow margin and a terminal value of 25 times free cash flow. Discount the $250M terminal enterprise value for five years at 18%, add $56.006M current cash and subtract $15M of present-value near-term cash consumption. Divide by 20.5M shares. This produces $7.33 per share. The high discount rate and burn allowance reflect development risk; the revenue, margin and terminal multiple remain analytical assumptions.

Route two uses a commercialization revenue multiple. The $20.37M FY2028 revenue figure in the saved provider estimate snapshot is used only as a scale reference; our model deliberately pushes that amount to FY2029 to reflect timing risk. Applying an assumed eight-times sales multiple, discounting three years at 18%, and adding the same cash less burn yields $6.84 per share. Eight times sales is a generous scenario assumption, not an observed peer benchmark. It requires attractive eventual margins and further growth to make economic sense.

Route three is a balance-sheet downside cross-check. Start with the audited $56.006M cash balance, deduct all $3.732M of recorded liabilities and a further assumed $15M of development cash use, then divide the remaining $37.274M by 20.5M model shares. The result is $1.82 per share. We assign no recovery to equipment, other noncash assets or intellectual property. The $15M is incremental to settling the liabilities already deducted, including recorded leases, so the same obligation is not charged twice. This is an analytical residual-cash case, not a guaranteed liquidation floor: timing, unrecorded future obligations and larger losses can reduce it. Its proximity to the $2 bear scenario provides a measurable downside anchor without pretending the operating valuation is independently confirmed.

At $11.10 and the same 20.5M model shares, reversing route one's arithmetic requires about $85.35M of FY2031 revenue at the same 20% cash margin and 25-times terminal multiple. The challenge is the path from today's $4.461M annual revenue to that scale, while protecting per-share economics. The bull case can exceed that hurdle, but the price already demands meaningful success. Our subjective 35% bear,45% base,20% bull weights produce $7.85, above base but still below the reference trade.

Wall Street: date every comparison

The saved provider consensus carries one Buy rating and a $25 target. That is a useful record of a bullish view, but it is not a fresh reaction to this earnings release. The latest dated grade in the retrieved series is March 3, 2026, showing Benchmark maintaining Buy. Earlier records also show Benchmark Buy observations in September and December 2025. The limited sample does not justify describing broad institutional agreement or a robust distribution of independent targets.

The target-news feed creates an additional attribution problem. Its March 3 article title references Benchmark and identifies David Williams, while the analyst-company field says Williams Trading. We preserve the raw record and do not confidently attach that target to a firm based on the inconsistent field. The presentation therefore separates the provider's $25 target consensus from the independently labeled dated grade. No target revision after September 16 is claimed.

Revenue estimates in the annual snapshot are $7.095M for FY2027, $20.370M for FY2028 and $79.168M for FY2029. Revenue analyst counts are two, three and one respectively. Those numbers illustrate the market's expectation of an eventual step change, but the distant estimate is extremely thin. Some accompanying SG&A and profit fields are internally inconsistent with the revenue scale, so they are excluded from our valuation inputs rather than copied into an impressive-looking forecast table.

Management does not supply an equivalent numerical FY2027 revenue target in the verified release. It emphasizes technology and product development, customer engagement, equipment capacity and prospective NRE agreements. Comparing that qualitative plan with a provider revenue estimate is informative only if the difference in status is explicit. We are more cautious than the saved bullish target because we charge for time, burn, stock compensation and dilution, and we require binding commercial evidence before underwriting the later-year ramp.

Management: financing skill is not product proof

Management deserves credit for obtaining capital ahead of the heavier spending phase. The ATM sales were completed at a reported average price well above the current reference trade, and the year-end balance sheet provides room to execute engineering and qualification work. That matters for a small technology company: running out of cash before a customer can qualify a product can destroy otherwise promising economics. Liquidity reduces that immediate financing pressure.

Capital allocation must still be graded on the eventual return. The release describes six new contracts totaling $5.3M and a transition from early government R&D toward scale and commercialization. These are not interchangeable with recognized annual revenue or unconditional backlog. The annual report distinguishes $5.4M of remaining obligations under obligated government contracts from an additional $5.7M of contingent obligations. We do not add those amounts to the CHIPS proposal and label the sum secured commercial demand.

The CHIPS LOI is especially important. The July 29 release says final documents would involve issuing equity securities to the Department of Commerce with aggregate value equal to the award amount. Some funding is contemplated upfront and the remainder through milestones tied to eligible costs and technical progress. Definitive documents, due diligence and approvals are still required. At an illustrative $11.10 issue price, $30M would correspond to approximately 2.703M new shares and 12.30% of the enlarged basic base. That is arithmetic, not disclosed final pricing or security terms.

Management must also finish the accounting work that accompanies a larger organization. A new finance leader, more accounting staff and external support are inputs. Tested effective controls are the outcome. Our scorecard rewards financing preparedness while withholding a commercial-scale or controls-remediation pass. The right next questions concern the economics of signed customer contracts, the timetable and criteria for qualification, future burn, government equity terms and the tests needed to close the material weakness.

1. Which customer agreements now commit money rather than express interest?
2. What reliability and yield gates remain before production qualification?
3. What cash spending is planned before the first repeat commercial production order?
4. What security, price and milestones will apply to government funding?
5. When will management have enough tested operating history to conclude controls are effective?

Risks and the strongest opposing case

The central risk is commercialization delay. A semiconductor platform can work in a lab and still fail a customer's performance, reliability, cost or supply-chain requirements. A design win can remain too small to absorb the fixed cost base. Qualification cycles are long, counterparties may be confidential, and large established competitors can respond. Our base does not assume technological failure; it assumes value takes longer to convert into distributable cash than a simple AI-market narrative suggests.

The second risk is ownership dilution. Current liquidity helps, but additional ATM sales, equity compensation and a possible government investment can enlarge the denominator. A proposed cash award should not be added to equity value while ignoring the shares issued in exchange. Conversely, equity financing is not automatically destructive: attractive pricing or value-creating development can more than compensate existing holders. The relevant question is the change in value per share, not the existence of fundraising alone.

The strongest bull case is that a scalable domestic photonics platform becomes strategically important as optical interconnect demand expands. Manufacturing partners and government support could shorten qualification and financing constraints; successful customer-funded development could validate the technology, then lead to production orders. A small starting revenue base allows very large percentage growth if even a few programs convert. Our $20 bull scenario preserves that possibility and is not a claim that today's revenue defines the permanent ceiling.

The bear case is persistent development spending with little repeat demand, a shrinking cash cushion and additional ownership issuance at unfavorable prices. Government concentration, contingent funding and weak controls amplify uncertainty. The investment conclusion is therefore cautious but not categorical: SELL at the current reference against a $7 model base, low conviction and a wide range. A strong operating milestone can change the distribution rapidly. That possibility is why the position should not be framed as a certain decline or as a simple short-sale recommendation.

Catalysts: distinguish dates from promises

The next quarterly filing is the first practical check on the sharper fourth-quarter spending trajectory. We set a November 30, 2026 review deadline for revenue, operating cash use and the updated cover share count; this is our monitoring checkpoint, not an announced earnings date. Revenue above $1.5M with evidence of recurring commercial contribution would improve the case more than a temporary government-contract milestone alone. Two further quarters below $0.6M would challenge the scale-up timetable.

A definitive CHIPS agreement would resolve several uncertainties, but the headline award amount is not the only relevant field. We would need funding dates, eligible costs, technical milestones, security type, issue price or valuation mechanism, restrictions and any clawback provisions. A favorable agreement could accelerate development. An agreement with costly restrictions or dilution could produce a different per-share result even at the same nominal award amount. Our December 31 checkpoint is a research date, not a deadline promised by the government.

The issuer has announced its annual stockholder meeting for November 19, 2026, with exact time and virtual-attendance details to follow in proxy materials. Governance disclosures can add ownership, compensation and financing context. The August lease amendment also adds approximately 3,400 square feet expected to become available October 1, with about $437,000 of additional payments over sixty months. This is evidence of planned expansion, not evidence that expanded capacity is already generating revenue.

We will also watch for signed NRE agreements, customer acceptance, manufacturing qualification and an explicit controls-remediation conclusion. None should be represented as complete based merely on encouraging language. The packet's signposts are designed to make our next assessment accountable: better evidence should move the model, while continued delays should reduce probability or increase cash consumption. Tracking only the stock price would miss the business developments that determine whether the equity financing ultimately creates value.

Appendix: data, sources and method

The dataset preserves nine fiscal quarters through June 2026 and three fiscal years of cash flow, with source URLs attached to observations. Current-quarter results are taken from the visible earnings-release tables. When a fourth-quarter cash-flow amount is not separately published, it is computed as the full fiscal year less the first nine months. EPS is never calculated by subtracting annual and cumulative per-share figures because the weighted share denominators differ.

Quarterly revenue and operating results are shown in millions of U.S. dollars. Earlier releases used dollars, while newer tables use thousands; the extraction records those units and favors the latest clearly labeled comparative presentation. Some annual and quarterly comparisons consequently reflect rounding. Historical non-GAAP definitions also changed, so adjusted figures are shown with their actual source definitions rather than represented as one uninterrupted uniform series.

The annual-report structure is a U.S. common-stock issuer with one reportable segment. There is no ADS ratio or VIE framework to apply. Preferred shares are authorized but not outstanding in the reported balance sheet. Dividends have never been paid and are not expected in the foreseeable future under the stated policy. The auditor is Rose, Snyder & Jacobs LLP, which has served since 2021. The packet does not infer the outcome of future controls testing or fill missing product margins with estimates.

All valuation scenarios, discount rates, revenue pathways, cash margins, enterprise values and share allowances are Charged Alpha assumptions. Historical facts and computed arithmetic are distinguished from those judgments. A low-confidence SELL is a valuation view at a dated trade, not individualized financial advice or an assertion of fraud. Subsequent filings, agreements or trading can invalidate this snapshot. Source access and the model should be revisited before drawing conclusions from an older packet.

• September 16 earnings release, SEC Exhibit99.1
• FY2026 Form10-K: financials, share count, controls and structure
• July29 CHIPS LOI release: award conditions and equal-value equity
• FY2024 annual filing
• FY2025 annual filing
• SEC quarterly reports and earnings releases for all nine periods are archived with SHA256 and retrieval timestamps.
• FMP saved Sep16/17 snapshots: dated trades, prices, earnings, targets, grades, float and estimates. API credentials are omitted from source receipts.