Phase A — Understand the business
Lens 1 · Company Overview
Amprius designs, manufactures and markets silicon-anode lithium-ion battery cells for mobility — primarily aviation (drones/UAS, high-altitude pseudo-satellites "HAPS", and eVTOL), with secondary ambitions in light electric vehicles (LEV) and, eventually, EVs. The pitch: replace the graphite anode (theoretical ceiling ~355 mAh/g, per management) with an engineered silicon anode (~3,400 mAh/g, ~10x), as a drop-in that runs on existing Li-ion lines and supply chains.
Two product platforms:
- SiCore — the commercial volume engine, developed with Berzelius (Nanjing) Co., Ltd. (a former affiliate) and launched commercially Jan 2024. Built almost entirely through outsourced contract manufacturing — Amprius has access to >2.0 GWh/yr across a South-Korean consortium (the "Amprius Korea Battery Alliance") in pouch/cylindrical/prismatic formats. Specs: up to 450 Wh/kg energy design, or 360 Wh/kg power design at up to 10C, cycle life 150–1,000 depending on design.
- SiMaxx — the in-house, highest-performance platform, made at the Fremont, CA pilot line. Prototype demonstrated >500 Wh/kg / >1,300 Wh/L (March 2023, third-party verified); a 390 Wh/kg cell passed the MIL-PRF-32383 nail-penetration (military) test. SiMaxx EV A-samples to USABC hit 360 Wh/kg at beginning-of-life, beating USABC's 275 Wh/kg end-of-life target.
Customers: >500 cumulative engagements since inception; >4.2M cells shipped. Named: AALTO (Airbus subsidiary), AeroVironment, BAE Systems, Kraus Hamdani Aerospace, Nokia Drone Networks, Nordic Wing, Teledyne FLIR, and the U.S. Army. Single reportable segment (the battery business).
Contract structure / payment terms: mix of recurring battery sales plus non-recurring customization/design-service revenue and government grants. No take-or-pay. Concentration is severe — see Lens 4/10. SiCore is a reseller-style model (Amprius buys finished cells from contract manufacturers and resells), which is why gross margins are thin vs a vertically-integrated cell maker.
Coverage-rule note (MenFem): Amprius is substantially defense-exposed — UAS/HAPS for the U.S. Army and allied militaries, MIL-spec qualification, a DIU contract, and (per Q1) batteries "incorporated into end products used in the defense industry by customers in jurisdictions experiencing military conflict". Per the MenFem hard exclusion (NO DEFENSE), this name is dossier-only intelligence on the energy/silicon-anode frontier; it is not a candidate for a MenFem position. Flagged explicitly in the Position seed.
Lens 2 · Supply Chain
Map: silicon-anode material / finished cells (upstream) → Amprius (designer + brand + some assembly) → integrators/OEMs → end users (drones, HAPS, LEVs).
- Upstream — silicon anode material: Berzelius (Nanjing) Co., Ltd. is the proprietary silicon-anode-material source for SiCore, under a Nov 2023 Exclusive Supply Agreement giving Amprius exclusive rights to buy Berzelius material in the US, Canada and Mexico. This is a single-source dependency AND a related party (former CEO/director Kang Sun sits on Berzelius's board). The short report alleges Berzelius was ~36% of COGS in 2025 and is itself ~80% owned through a Hong Kong entity tied to "Amprius Inc." at Amprius's own Fremont address — unproven, but material if true (Lens 10/13).
- Manufacturing — outsourced cells: the Amprius Korea Battery Alliance (a consortium of South Korean firms across the Li-ion value chain) provides >2.0 GWh/yr of SiCore capacity. The short report names alleged partners Eurocell (Korea, factory reportedly seized over fraud) and Nanotech Energy (US, abandoned plans) as evidence the network is "subscale" — Amprius does not name these in the 10-K; treat as allegation.
- In-house: Fremont, CA pilot line (SiMaxx + quick-turn SiCore prototypes), being expanded to 10 MWh, partly funded by the DIU contract (now $18.1M).
- Downstream: integrators build the named end products (AALTO Zephyr HAPS, AeroVironment/Teledyne FLIR drones, etc.). Heavy EMEA delivery, including a large slug to Ukraine (Lens 4).
Chokepoints / single-source: (1) Berzelius for anode material — concentrated, related-party, China; (2) Korean contract manufacturers for cell volume — Amprius owns the IP/recipe but not the fab, so capacity and quality are at arm's length; (3) China export controls on Li-ion batteries/materials/equipment (enforcement suspended to ≥Nov 2026) and NDAA "foreign entities of concern" rules both threaten the China-sourced supply against the U.S. defense demand — a direct collision in the business model.
Lens 3 · Competitive Advantages (moats)
- Performance moat in aviation: management argues that for aviation specifically — where you need simultaneous high specific energy, high power density, low-temperature operation and fast charge — silicon-anode is "the only battery solution currently available and suitable for broad aviation adoption". The AALTO Airbus Zephyr flying 67 days continuously at ~70,000 ft on Amprius cells (May 2025) is the proof point. This is a genuine, narrow, defensible niche — the energy-density frontier where graphite simply can't go.
- First-mover + reputation: >500 validated customers, design wins and recurring orders with tier-one aerospace/defense names; qualification cycles in aviation are long and sticky (switching cost moat once designed-in).
- IP: 86 patents total — 74 issued (35 US, 39 international across EU/Korea/Japan/China/Taiwan/Israel; expirations 2030–2039), 10 pending, plus 2 licensed from Stanford; 10 trademarks; trade secrets over 15 years of R&D. Real but not impregnable — many well-funded silicon players exist.
- Bargaining power: weak on both sides today. Upstream, Amprius is dependent on a single related-party material source and arm's-length Korean fabs (asset-light = low capex but low control). Downstream, revenue is wildly concentrated (one customer = 37% of FY2025 — Lens 4), so individual buyers hold leverage. The moat is the physics/qualification in aviation, not commercial bargaining power.
- Durability question: the silicon-composite field (Enovix, Sila, Group14, Nexeon, Enevate, StoreDot, BTR, Shanshan) is crowded and better-capitalized. Amprius's edge is being aviation-first and already flying, not being the only silicon technology.
Lens 4 · Segments
Single reportable segment (battery business), so the meaningful cut is geography + customer concentration:
Revenue by geography (Q1 2026, $000):
| Region | Q1 2026 | Q1 2025 |
|---|
| North America | 6,058 | 1,923 |
| EMEA | 16,523 | 6,302 |
| APAC | 5,955 | 3,059 |
| Total | 28,536 | 11,284 |
- EMEA is the engine (~58% of Q1'26), and within it, shipments to Ukraine-based customers were $10.0M in Q1 2026 (vs $4.6M Q1'25) — i.e. ~35% of total Q1 revenue is Ukraine. This is battlefield-drone demand: high-growth, high-margin, but ceasefire-fragile and defense-coded. Management explicitly warns "any cessation… of these conflicts could impact future sales".
- Customer concentration: FY2025 — one customer = $27.1M of $73.0M (37%). Q1 2026 — three customers = 49% of revenue (improving from 66% in Q1'25); two customers = 64% of AR. Diversifying, but still dangerous.
- Trend: revenue accelerating (FY2025 +202%; Q1'26 +153% YoY, +13% QoQ), and the mix is broadening toward China LEV (the $21M SiCore order, Lens 8) which would shift the geography toward APAC and dilute the Ukraine dependence — but at likely lower LEV margins.
Phase B — Measure performance
Lens 5 · Earnings Result (FY2025 + Q1 2026)
FY2025 (10-K, audited by BDO USA — clean opinion, no going-concern qualification), all $000:
| Line | FY2025 | FY2024 | Δ |
|---|
| Revenue | 73,011 | 24,167 | +202% |
| Cost of revenue | 64,747 | 42,497 | +52% |
| Gross profit (loss) | 8,264 | (18,330) | first profit |
| Gross margin | 11% | (76)% | +87pt |
| R&D | 9,430 | 7,344 | +28% |
| SG&A | 22,956 | 18,726 | +23% |
| Impairment & other | 22,524 | 1,862 | +1110% |
| Loss from operations | (46,646) | (46,262) | flat |
| Net loss | (44,024) | (44,671) | −1% |
| EPS (basic=diluted) | (0.35) | (0.45) | improving |
The story: revenue tripled and gross margin flipped positive for the first time (driven by +$48.1M SiCore battery sales). But the operating loss was masked flat by a $22.5M impairment — $19.1M for the abandoned Brighton, Colorado lease (a 774,000 sq ft factory plan scrapped in favor of the asset-light contract-manufacturing strategy) plus equipment retirement. On Jan 30, 2026 Amprius paid $20.0M to terminate the Brighton lease (hits Q1'26).
Q1 2026 (latest print, 10-Q), $000:
- Revenue $28,536 (+153% YoY, +13% QoQ); gross profit $5,740 (20.1% GM) vs $(2,361) loss — margin is climbing toward the 25% FY target.
- Net loss $(5,046) vs $(9,371); EPS $(0.04) vs $(0.08).
- The $20M Brighton settlement does not show as a new P&L hit — the impairment was pre-booked in FY2025 and the cash settlement netted against existing ROU/lease balances (ROU assets $19.5M→$6.0M; non-current lease liab $35.2M→$5.4M).
Balance-sheet flags:
- Cash $62.4M (down from $90.5M at YE — ~$28M consumed by the $20M lease payment + operating burn). Zero debt.
- Accounts receivable $35.3M (up from $23.7M at YE, and from $5.6M a year earlier) — receivables are growing far faster than revenue and now equal ~123% of a quarter's revenue / ~31% of trailing-12m. This is the single most important accounting flag and the crux of the bill-and-hold short thesis (Lens 10/13).
- Inventory $8.2M (rising). Equity $109.4M, accumulated deficit $223.4M.
Market reaction: despite the revenue beat, Q1 was framed as an EPS miss, and the stock fell ~15% in the week after the May 7 print. Roth Capital nonetheless raised its PT to $22 (Buy). The tape reaction says the market is now scrutinizing quality of earnings, not just growth.
Guidance (raised): FY2026 revenue ≥$130M (up from ≥$125M), net loss <$8.0M, and positive non-GAAP adj. EBITDA ≥$4.0M — i.e. the first quasi-profitable year — with a 25% gross-margin target.
Lens 6 · Earnings Calls (sentiment trend)
No transcripts/ on the research layer; this lens is ``.
- Trajectory of tone: management's message has shifted from "scaling, pre-profit" (2024) to a confident "inflection to profitability" narrative in 2026 — explicitly guiding to positive adjusted EBITDA and a 25% margin, raising revenue guidance on "strong Q1, increased visibility from customer awards, and a significant China LEV purchase order".
- What they emphasize now: the asset-light contract-manufacturing model ("rapid capacity expansion with minimal capital investment"), the DIU/defense path (NDAA-compliant cells; a hoped-for "Program of Record"), and margin improvement in 2H as logistics/customer mix evolve.
- What they've stopped saying: the Brighton/Colorado vertical-integration build-out (quietly killed and written off). The strategy pivot from "build our own gigafactory" to "outsource everything" is the biggest narrative change and a credit to capital discipline — but it also concentrates the supply-chain risk (Lens 2/13).
- Sentiment read: genuinely improving fundamentals + promotional framing of the order book. The May 2026 short report directly attacks the credibility of that order/manufacturing narrative — so the next 2–3 calls are a referendum on whether guidance converts to GAAP cash.
Lens 7 · Comps
Silicon-anode / next-gen battery peers. Multiples are ``, dated; where I cannot source a clean figure I mark n/a. Note the peers are not pure comparables — Enovix is consumer-electronics-led, Sila/Group14 are EV-materials-led, while Amprius is aviation-led; valuation reflects different end-market growth.
| Company | Ticker | Mkt cap | EV/Sales (TTM) | P/E | Div yield | 5-yr avg ROE | Note |
|---|
| Amprius | AMPX | ~$2.1–2.3B | ~38–40x P/S | n/m (loss) | 0% | negative | Aviation silicon-anode; $73M FY25 rev |
| Enovix | ENVX | n/a (was ~$15.5/sh Jul-25 ) | n/a | n/m (loss) | 0% | negative | 9M'25 rev $20.6M; GAAP GM 18%; cash $648M |
| Sila Nanotechnologies | private | n/a — private | n/a | — | — | Moses Lake ~10 GWh target 2026 | |
| Group14 Technologies | private | n/a — private | n/a | — | — | Materials supplier; 80+ customers; SK JV, BASF | |
| Nexeon / Enevate / StoreDot | private | n/a — private | n/a | — | — | Silicon-anode developers | |
| CATL / LG Energy / Panasonic / Samsung SDI | incumbents | tens of $B | n/a | varies | varies | Graphite tier-1; the "good-enough" alternative outside aviation | |
Read: AMPX trades at ~38–40x trailing P/S vs a peer average ~17.5x — the most expensive name in the silicon cohort, and on ~17–18x FY2026 guided sales ($130M). The premium prices in (a) the aviation moat, (b) the swing to profitability, (c) defense optionality. The peer that matters most for capitalization risk is Enovix: ~$648M cash vs Amprius's $62M — Amprius is fighting a capital-intensity battle with a fraction of the balance sheet, mitigated only by its asset-light model.
Lens 8 · Stock-Price Catalysts (moves >5%)
Mostly ``. Pattern over the last ~18 months:
- 2024 — SiCore commercial launch + Amprius Holdings dissolution (Oct 2024 removed the 58.6% parent overhang, freeing the float).
- May 2025 — AALTO Airbus Zephyr 67-day stratospheric flight on Amprius cells: flagship validation event.
- ~$500M cumulative U.S. military orders + DIU contract increases to $18.1M — defense demand re-rating.
- Mar 18, 2026 — 52-week high on FY2025 results + bullish 2026 guidance; +26.7% move on the guidance raise + drone deals.
- Mar 25, 2026 — $21M China LEV order announced, but stock fell ~10.2% — a "sell-the-news"/China-margin-skepticism reaction. Tells you the market is wary of low-margin China volume.
- May 7, 2026 — Q1 print: revenue beat, EPS miss → stock −15% on the week.
- May 20, 2026 — Manatee Research short report → shares slip; Block & Leviton "investigation" solicitation follows.
- Jun 2026 — recovery; ~$15.23 (Jun 17), +3.2% intraday Jun 15.
What the tape actually reacts to: (1) marquee aviation validation (Airbus), (2) defense/DIU order flow, (3) the swing-to-profit guidance — up; and (4) revenue-quality / margin / related-party doubts (Q1 EPS miss, China-order skepticism, short report) — down. This is a name that rallies on the moat story and falls on the accounting story. Forward catalysts: a DoD Program of Record, EV-prototype data (claimed 10x TAM), the warrant-exchange dilution clearing, and whether 2H delivers the 25% margin.
Phase C — Judge people & books
Lens 9 · Management
- CEO: Tom Stepien (effective Jan 1, 2026; joined as President May 2025). ~35 yrs in tech; prior CEO of South 8 Technologies, co-founder of Primus Power, leadership at Applied Materials. A professional operator/scaler brought in precisely for the commercialization-and-margin phase.
- Founder transition: Dr. Kang Sun retired as CEO end-2025; remains a director and Executive Advisor. Critical caveat: Sun also sits on the board of Berzelius, the related-party single-source supplier, and (per the short report) holds an indirect ownership stake in it. The founder's overhang on the supply chain persists even after he left the CEO seat.
- Track record: Sun took Amprius from R&D to a $73M-revenue, first-gross-profit company and got it flying on Airbus's Zephyr — a real technical achievement. The asset-light pivot (writing off Brighton to avoid a capex sinkhole) is a disciplined call. Capital-allocation history is otherwise thin (no buybacks/dividends; pre-profit).
- Skin in the game: post-Amprius-Holdings dissolution, insider ownership is dispersed. Stepien holds ~625,625 shares + ~615,625 RSUs and 36,914 via trust, and has been a net seller (small sells $42K–$897K in 2026). The short report alleges insiders sold >$79M Nov-2025→May-2026 — heavy insider selling into the rally is a real governance yellow flag, even granting that much of it may be the post-dissolution distribution unwinding.
- Archetype: transitioning from founder-scientist (Sun) to professional manager (Stepien) — appropriate for the stage, but the related-party supply structure tied to the founder is the governance question that overhangs the whole thesis.
Lens 10 · Forensic Red Flags
Acting as a forensic analyst. Ground: filings/ + the regulatory file.
- Receivables outrunning revenue (highest-priority flag): AR went $5.6M (YE'24) → $23.7M (YE'25) → $35.3M (Q1'26) while quarterly revenue is ~$28.5M. DSO has ballooned. Combined with negative operating cash flow ($(31.1)M in FY2025 vs a $(44.0)M net loss ), this is the classic "earnings without cash" divergence. This is exactly what the short report attacks as bill-and-hold revenue (recognizing sales on goods not yet shipped/collected). The filings do not disclose bill-and-hold explicitly, but the receivables trend warrants close watch on the next 10-Q's cash conversion.
- Related-party single source: Berzelius — exclusive anode-material supplier, board-linked to founder/director Kang Sun, alleged ~36% of COGS and an opaque ownership chain back to "Amprius Inc." at Amprius's own address. The 10-K discloses the relationship and the exclusive agreement but does not quantify Berzelius as a % of COGS — a disclosure gap the short report exploits.
- Impairment / strategy-reversal: $22.5M FY2025 impairment (Brighton + equipment) and the $20M lease-termination cash-out — not an accounting trick (it's a write-off of a bad plan), but it signals execution churn and flatters the forward P&L by removing the drag.
- SBC: $7.4M FY2025 (≈10% of revenue) — non-trivial; the FY2026 "positive adjusted EBITDA ≥$4M" target leans on adding SBC + D&A back, so GAAP will still be a loss (guided <$8M net loss). Watch that "adjusted" framing.
- Dilution machinery: funded entirely by equity — $97.5M cumulative ATM completed, ongoing option/warrant exercises; shares 116.9M (YE'24) → 134.5M (YE'25) → 139.3M (Q1'26), with 18.59M warrants outstanding at Q1'26 and a May 2026 warrant-exchange issuing more shares. Chronic dilution is structural until the model self-funds.
- Auditor / controls: BDO USA, clean opinion, auditor since 2022; emerging-growth + smaller-reporting company → reduced disclosure and no SOX 404(b) auditor attestation on internal controls. The reduced-disclosure status is itself a (legal) risk-amplifier for a name now facing fraud allegations.
Regulatory findings (required sub-section). Sources: regulatory/regulatory-findings.md (SEC EDGAR EFTS — LR + AAER), web search, and 10-K Item 3.
- SEC Litigation Releases: none found naming Amprius (2021-06-18→2026-06-18).
- SEC AAERs: none found.
- 10-K Item 3 (Legal Proceedings): the 10-K (filed 2026-03-06, pre-dating the May short report) discloses no material pending litigation of note — standard for the period.
- Non-SEC / web: No FTC/DOJ/FDA enforcement actions, consent decrees, fines or penalties found. However, a plaintiff-firm "securities fraud investigation" (Block & Leviton) and the Manatee Research short report (2026-05-20) allege exaggerated orders, hollow/"Potemkin" manufacturing (citing a Korean partner "Eurocell" factory reportedly seized over fraud and US partner "Nanotech Energy" abandoning plans), bill-and-hold revenue, and undisclosed related-party economics with Berzelius. These are unproven allegations and solicitations — not government actions or proven findings — but they are material to the thesis and corroborate the independently-observed receivables flag. Net: no regulatory enforcement on record; an active short-seller fraud allegation and plaintiff solicitation pending, unadjudicated, as of 2026-06-18.
Phase D — Project & stress-test
Lens 11 · Forward Projection
Built bottom-up from FY2025 actuals + management guidance; EPS lines are `` with arithmetic shown. No forecast.ts create is run (watchlist/unattended rule).
Revenue:
- FY2026: management guide ≥$130M; base $135M.
- FY2027: $185M base.
- FY2028: $240M base.
Margins / opex: 25% GM target FY2026; assume GM 24% (FY26) → 27% (FY27) → 30% (FY28) as outsourced scale + mix matures. Opex ~$48M FY26 growing ~15%/yr. SBC ~$8M/yr. Share count ~145M FY26 → ~155M FY27 → ~163M FY28 (continued dilution + warrant exercises).
EPS (non-GAAP-ish, illustrative — company is GAAP-loss):
- FY2026: Rev $135M × 24% GM = $32.4M GP − ~$48M opex + ~$2.5M other ≈ $(13)M GAAP op loss; net loss guided <$8M → GAAP EPS ≈ $(0.05). Adjusted EBITDA ≈ +$4–6M (adds back ~$8M SBC + ~$3M D&A).
- FY2027 (base): Rev $185M × 27% = $50M GP − ~$55M opex ≈ ~$(3)M op; near GAAP breakeven → EPS ≈ $0.00 ± 0.03.
- FY2028 (base): Rev $240M × 30% = $72M GP − ~$63M opex ≈ +$9M op → EPS ≈ +$0.05 — first sustained GAAP profit.
Bull path: Program-of-Record + EV qualification → FY2027 rev $220M+, GM 30%, EPS positive a year early.
Bear path: Ukraine demand fades / short-report damages order book / China-LEV margins disappoint → FY2027 rev $140M, GM stuck ~18%, continued losses + dilution.
The number that matters: does the model reach self-funding before the $62M cash + warrant proceeds run out? At ~$30M/yr operating burn pre-breakeven, cash funds ~2 years; warrant exercises ($18.6M warrants @ $11.50–12.50, well in-the-money at ~$15) plus possible ATM refresh bridge it. Management asserts 12-month sufficiency. Base case: reaches adj-EBITDA-positive in 2026 and GAAP-profit ~2028, with one more capital raise likely. Falsifiable forecast (not logged): AMPX FY2026 revenue ≥ $130M AND positive non-GAAP adjusted EBITDA — p≈0.65.
Lens 12 · Bull vs Bear
Bull case. Amprius owns the one battery niche where physics, not price, decides — aviation/HAPS/high-end drones — and it's already flying on Airbus's Zephyr and qualified to MIL-spec. Revenue tripled to $73M, gross margin flipped positive, and 2026 guides to the first profitable (adj-EBITDA) year on ≥$130M with a 25% margin. The asset-light contract-manufacturing model means it can scale GWh without a gigafactory's capex. Optionality is large and cheap-to-acquire-for-an-acquirer: a DoD Program of Record (multi-year defense revenue), EV qualification (USABC targets already beaten — a 10x TAM), and an 86-patent silicon-anode estate that makes it an M&A target. Secular tailwind: drones, HAPS connectivity, and electrified mobility all need exactly the energy density Amprius sells.
Bear case (permanent-impairment risks). (1) Revenue quality — receivables ($35M) are sprinting ahead of revenue and operating cash flow is deeply negative; if the short report's bill-and-hold/"exaggerated orders" allegation has substance, reported revenue is overstated and the whole growth narrative cracks. (2) Related-party / China supply — a single, founder-linked Chinese supplier (Berzelius, ~36% of COGS) feeding U.S. defense demand, into the teeth of NDAA "foreign-entity-of-concern" rules and China export controls (suspended only to Nov 2026) — a structural collision that could sever either the supply or the defense market. (3) Concentration — ~35% of revenue is Ukraine battlefield drones; a ceasefire is a revenue cliff. Pre-mortem (18 months out, thesis broke): the short report's allegations gained traction (or an SEC inquiry opened), a key contract manufacturer failed or was sanctioned, Ukraine demand normalized, and the China-LEV "growth" came in at near-zero margin — leaving a ~$120M, low-margin, cash-burning company that had been priced at $2B+. Multiples: at ~38x trailing / ~17x forward sales, expectations are demanding; any of the three bear vectors compresses the multiple violently. Contrarian view the market is refusing to see: the quality of the order book and the legitimacy of the related-party supply chain matter more than the headline growth rate — the market is paying a hyper-growth multiple for revenue whose cash conversion and provenance are exactly what a credible short seller is now contesting.
Lens 13 · Devil's Advocate (short-seller)
Dismantling the bull case. The short thesis here isn't hypothetical — Manatee Research already wrote it (2026-05-20), and it's well-aimed:
- Where revenue is concentrated → what breaks it: ~37% from one customer (FY25), ~35% from Ukraine (Q1'26). A single design-loss or a ceasefire halves growth. Bill-and-hold allegation: if revenue is being pulled forward against rising receivables, the "+202%/+153%" growth is partly accounting, not demand.
- Why the moat is weaker than bulls think: Amprius doesn't make the cells (outsourced to Korea) or the key material (Berzelius/China) — it owns IP and a brand. That's a recipe, not a fortress; better-capitalized silicon players (Enovix at $648M cash, Sila, Group14+BASF) can out-spend it, and graphite tier-1s (CATL/LG/Samsung SDI) own everything outside the narrow aviation window.
- Most dangerous competitor bulls underestimate: Group14 + the materials-supplier model — if silicon-carbon material becomes a merchant commodity (Group14 claims 80+ customers, 95% of battery production), Amprius's "proprietary" SiCore edge erodes to a design-house margin.
- Worst capital-allocation / governance: the Berzelius related-party structure tied to the founder (alleged ownership back to Amprius's own address), the $79M alleged insider selling into the rally, the Brighton write-off churn, and chronic equity dilution (116.9M→139.3M shares + 18.6M warrants).
- Assumptions that must hold for $15+/$2B: 25% gross margin actually lands; Ukraine demand persists; the short allegations are false; the China supply survives NDAA; and equity dilution stays modest. If growth disappoints 20–30% (rev ~$95–105M, GM stuck ~18%), this is a sub-$1B company — i.e. ~50%+ downside on multiple compression alone.
- Single scenario that permanently impairs: an SEC inquiry or audit re-statement validating the bill-and-hold / related-party allegations. Plausibility: low-to-moderate (no agency action yet, clean BDO opinion), but non-trivial given the receivables math is independently visible in the filings. This is the binary that defines the name.
Lens 14 · Management Questions (ordered by information value)
- What exact percentage of COGS did Berzelius represent in 2025 and Q1 2026, what is the full ownership chain of Berzelius, and what related-party economics flow to Dr. Kang Sun or any insider? (Directly tests the central short allegation.)
- What portion of recognized revenue in FY2025 and Q1 2026 was bill-and-hold or not yet shipped, and what is your revenue-recognition policy for cells held at contract manufacturers? (Tests the receivables/earnings-quality gap.)
- Reconcile the rise in accounts receivable to $35M against $28.5M quarterly revenue — what are DSOs by customer cohort, and how much is past due?
- Name your contract manufacturers and their qualified, audited capacity — specifically address the Eurocell and Nanotech Energy claims in the Manatee report.
- How do you keep a U.S.-defense (DIU/Army) battery business compliant with NDAA "foreign-entity-of-concern" rules while sourcing anode material from a Chinese related party? What is the NDAA-compliant second source and its timeline?
- What share of 2026 revenue do you expect from Ukraine/conflict-driven demand, and what is the revenue plan if there is a ceasefire?
- What gross margin does the $21M China LEV order carry, and what is the blended-margin trajectory as China LEV scales vs aviation?
- What is the realistic timeline and revenue scale for a DoD Program of Record, and what milestones gate it?
- Bridge "positive adjusted EBITDA ≥$4M" to GAAP net loss <$8M for 2026 — what exactly is added back, and when does GAAP profitability arrive?
- What is the total remaining dilution from warrants, RSUs and any planned ATM, and at what share count does the model self-fund without new equity?
- On EV: what cycle-life, cost, and volume gaps remain before SiMaxx is commercially competitive, and what capital does that require?
- How defensible is SiCore if silicon-carbon material (Group14/BASF, Sila) becomes a merchant commodity — what is proprietary in the cell vs the material?
- What was the total insider selling Nov-2025 to May-2026, and how do you reconcile it with confidence in the guidance?
- Why retain emerging-growth/smaller-reporting reduced-disclosure status (no SOX 404(b) attestation) given the scrutiny — will you opt into full controls attestation?
- What is the single biggest risk to the 2026 plan that is not in the risk factors, and how are you mitigating it?