Array Technologies

Phase A — Understand the business

Lens 1 · Company Overview

Array makes the steel-and-software skeleton that points solar panels at the sun. Its core product is a single-axis tracker — motors, gearboxes, drivelines, torque tubes and controllers that rotate rows of PV modules east-to-west across the day, lifting energy yield and lowering the Levelized Cost of Energy (LCOE) versus a stationary "fixed-tilt" mount. The flagship is DuraTrack HZ v3 (launched 2015), whose patented design lets one motor drive multiple linked rows — the central structural claim of the moat (Lens 3). Around it sits a product family: OmniTrack (uneven terrain, less grading), SkyLink (grid-independent PV-powered controls, wireless, no trenching), the STI H250 dual-row tracker (from the 2022 Spanish STI acquisition), SmarTrack software (hail/snow/diffuse-weather stow algorithms), and — new in 2025 — APA foundation and racking systems (ground screws, helicals, c-piles, ballast).

Plain model: Array sells to solar developers, IPPs, utilities and EPCs under master supply agreements / multi-year procurement contracts, usually drop-shipped from outsourced suppliers straight to job sites. Revenue is recognized over time (cost-to-cost) on most projects — Q1-2026 was 77% over-time / 23% point-in-time. Since founding it has shipped >96 GW of trackers; FY2025 split 81% US / 19% rest-of-world. HQ: Albuquerque, New Mexico; ~283,000 sq ft domestic plant plus a new ~176,000 sq ft Bernalillo County facility coming online H1-2026, plus STI manufacturing in Spain (~67k sq ft) and Brazil (~610k sq ft).

Contract structure / payment terms. Contracts run "hundreds of thousands to tens of millions of dollars," with delivery from days to several months — i.e. short-cycle, not take-or-pay. Array can decline fixed-price or commodity-indexed multi-year contracts it deems too risky ("Structured Cost Management"), accepting that this pushes some volume to competitors. Customer concentration is real but moderate: in FY2025 the two largest customers were ~13.7% and ~12.2% of revenue; top-5 = ~29.8% of receivables.

Lens 2 · Supply Chain

Upstream inputs → Array → end customer, named where the filings name them:

• Raw materials: steel tubing, steel supports, aluminum extrusions (clamps, U-joints, bearing housings), electric motors, gearboxes, bearing assemblies, drivelines, electronic controllers. Steel and aluminum are the dominant cost drivers; Array does not hedge raw-material prices. This is the single most important supply-chain fact: margins ride bare on steel/aluminum and tariffs.
• Suppliers/manufacturing model: Array deliberately outsources capital-intensive, low-value-add fabrication (steel tubing, supports, drivelines, bearings, gearboxes, motors, controllers ship direct from suppliers to job sites). It assembles module clamps, center structures and motor-controller assemblies in-house at Albuquerque. Named related-party leases (Ohio/Connecticut) came with APA — five leases are with related parties owned by APA management (flagged again in Lens 9/10).
• 45X benefit-sharing: Array negotiates with suppliers of torque tubes and structural fasteners to split the Section 45X advanced-manufacturing PTC; it also self-manufactures certain 45X-eligible parts. The 45X economics are a live margin lever and a live risk (OBBB FEOC limits).
• Logistics chokepoints (named): Red Sea / Suez shipping disruption (Asia routes), Strait of Hormuz risk from the March-2026 US/Israel–Iran conflict driving commodity/freight costs, and Brazilian Real depreciation hammering the STI segment.
• Downstream chokepoint NOT in Array's control: high-voltage breakers, switches and transformers for grid interconnection are in shortage and gate when projects actually get built. Array can ship a tracker; if the developer can't get a transformer, revenue slips a quarter.

Single-source / concentration risk: Array claims redundancy ("identified certain alternative vendors") and the ability to in-source, but the outsourced steel/aluminum chain means AD/CVD and Section 232 metal tariffs flow almost directly into cost per watt (CPW). The chain is US-centric for the Legacy segment (a deliberate tariff hedge) and Spain/Brazil-centric for STI.

Lens 3 · Competitive Advantages (moats)

The moat is real but narrow, and it has a clock on it.

1. The linked-row patent. Array's core US patent on a single-motor, linked-row rotating-gear-drive system "does not expire until February 5, 2030." Competitors must use designs Array argues are inherently less efficient (e.g. one motor per row). IP estate: 74 issued US patents, 174 non-US, 216 pending; US patents expire 2030–2045. This is a genuine, datable switching/design-around advantage — but the flagship's foundational claim lapses in 2030, ~4 years out.
2. 100% domestic-content capability — the 2026 moat that matters most. Array (and Nextracker) can deliver 100% US-content trackers, letting customers claim the IRA/OBBB domestic-content bonus credit and dodge the FEOC "prohibited foreign entity" trap. Sell-side notes Array has booked 100%-domestic-content orders and is cutting tariff-exposed BOM to <14%. In a post-OBBB world where Chinese supply is being cut off, this is a structural tailwind to relative share even as the absolute US market faces a credit cliff.
3. 45X benefit capture — Array converts torque-tube/fastener 45X credits into margin via supplier-sharing deals (Lens 2). Durable only as long as 45X survives and FEOC rules don't disqualify the parts.
4. Bargaining power: weak-to-balanced. Array sells to large, sophisticated buyers (Tier-1 utilities, IPPs) who often dictate the tracker spec; it buys commodity steel/aluminum it can't hedge. It is a price-taker on inputs and a spec-taker on the demand side — the moat is product design + domestic footprint, not pricing power. Its willingness to walk from bad fixed-price contracts is a discipline signal, not a moat.

Versus rivals on perceived value: Array competes on LCOE, reliability, warranty duration and US-content. It is not the share leader (Lens 7) — Nextracker out-innovates (robotics/AI division via OnSight) and out-scales it ~3:1. Array's edge is the patent + a credible domestic-manufacturing story, not brand dominance.

Lens 4 · Segments

Two reportable segments. segments.csv is empty, so all figures are from the filings' segment tables.

By segment — FY2025 vs FY2024:

By segment — Q1-2026 vs Q1-2025:

The story the segments tell: Array is now a US business with a smoldering crater where Brazil used to be. The Legacy segment (with APA bolted on Aug-2025) is growing volume and holding ~28–30% gross margin. STI — the Spanish/Brazilian operation — has imploded: −93% revenue and a negative 37% gross margin in Q1-2026, driven by Brazilian PPA economics collapsing under a weak Real and pricing pressure. Array has already (a) taken a $29.5M one-time inventory valuation charge on STI in 2025, (b) recorded $102.6M of STI goodwill impairment in 2025 (on top of $236.0M in 2024 — see Lens 10), (c) approved a plan to phase out the non-SmarTrack H250 and resize international operations, and (d) signaled intent to push the flagship DuraTrack through STI's channel. Translation: the entire STI thesis is being rebuilt mid-air, and until it is, it is a margin drag, not a contributor.

By geography: FY2025 = 81% US / 19% RoW. The mix is becoming more US-weighted as STI shrinks and APA (US) grows — which is strategically defensible given OBBB/FEOC but concentrates Array into a single policy regime.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print = Q1-2026, period ended 2026-03-31)

[All figures research-layer: filings/10-q-2026-q1.md unless labeled.]

• Revenue $223.4M, −26% YoY (vs $302.4M) — but this is seasonal + STI collapse, not core US weakness (Legacy +2%). Q1/Q4 are structurally Array's weak quarters (Lens 8 seasonality).
• Beat vs consensus: revenue $223.4M beat the ~$205.9M Street estimate; adjusted net income/diluted share $0.06 beat the −$0.05 consensus loss.
• Gross profit $63.0M; GAAP gross margin 28.2%, UP from 25.3% YoY — the headline positive. Adjusted gross margin 30.7%. Margin is expanding even as revenue shrinks, because the mix shifted to higher-margin US Legacy and 45X benefits landed.
• Income from operations $7.1M (down from $27.3M — flattered last year by no impairment this quarter; the YoY op-income drop is the revenue decline, not a cost blowout).
• GAAP net income $2.0M. BUT — the punchline — after $15.5M of preferred dividends & accretion, net loss to common = $(13.5)M, EPS $(0.09) (vs +$0.02 PY). Adjusted EPS was +$0.06. The gap between $2.0M net income and a $(13.5)M loss to common is the Series A preferred (Lens 10 / Devil's Advocate).
• Adjusted EBITDA $28.8M.
• Order book / backlog — TWO numbers, both true, do not conflate:
• GAAP Remaining Performance Obligations = $544.0M as of 2026-03-31, 94% recognizable within 12 months. This is the binding, revenue-recognizable backlog. (Was $400.6M at Q3-2025 — a healthy sequential build.)
• "Total executed contracts and awarded orders" = record $2.4B. This is the IR non-GAAP "order book," includes awarded-but-not-yet-binding orders, and is the figure the press repeats. 2x quarterly book-to-bill, 1.3x TTM book-to-bill.
• Balance-sheet flags: operating cash flow NEGATIVE $(29.4)M (vs −$13.1M PY) — Q1 is a working-capital drain quarter; cash fell to $200.7M (from $244.4M at YE2025). Inventory rose to $168.0M (from $150.4M) and unbilled receivables to $110.5M — receivables/inventory building ahead of the Q2/Q3 seasonal ramp (consistent with guidance) but worth watching. IRA vendor-rebate receivable $164.7M (45X economics owed by suppliers) is a large, policy-dependent asset.
• Guidance / tone (reaffirmed at Q1): FY2026 revenue $1.4–1.5B, adjusted EBITDA $200–230M, adjusted net income/share $0.65–0.75. Q2-2026 guided to $300–320M revenue — i.e. management is telling you the seasonal ramp is intact and H2-weighted.
• Market reaction: modestly positive (up ~1.7–4% after the print; one source cited +10.6% after-hours). The stock had risen ~57% over the prior year off the 2025 trough.

Unusual vs its own history: margin expanding while revenue contracts (good), but a GAAP net-loss-to-common despite GAAP net income (structural preferred drag) and negative operating cash flow (seasonal). The single most "unusual" item is the STI segment at −37% gross margin — that is not normal and is the active wound.

Lens 6 · Earnings Calls (sentiment trend)

transcripts/ is empty on the shelf; this lens is `` + filing-derived MD&A tone.

• Tone arc, FY2024 → Q1-2026: From defensive (2024 was dominated by two consecutive STI goodwill impairments, a sliding stock hitting 52-week lows, and IRA-guidance uncertainty) → to cautiously confident in 2025–2026. The recurring phrases now: "record order book," "100% domestic content," "margin expansion," "disciplined international expansion," "APA integration progressing well".
• What management started saying: domestic-content as a competitive weapon; APA synergies; book-to-bill > 1.
• What they stopped saying: the aggressive Brazil/international growth story — replaced by "resize," "realign cost structure," "phase out" the non-SmarTrack H250. The pivot from "grow STI" to "shrink and fix STI" is the clearest sentiment shift.
• Management focus (stated): "high-quality domestic opportunities while pursuing disciplined international expansion" — CEO Hostetler. Read: defend the US franchise, stop bleeding overseas.

Lens 7 · Comps

Peers from the 10-K's own stock-performance peer group (Enphase, SolarEdge, Shoals, FTC Solar) plus the only true direct tracker peer, Nextracker (now "Nextpower," NXT). Multiples are ``; where not sourced, marked n/a.

Read: Array trades at roughly half Nextracker's multiple on both EV/Sales (1.29x vs 3.84x) and EV/EBITDA (9.3x vs 17.6x). That discount is partly earned — Nextracker is ~2.8x the revenue ($3.56B vs $1.28B), has a >$5.25B backlog vs Array's $2.4B order book / $544M GAAP RPO, FY2026 adj EBITDA $810–830M vs Array's $200–230M, adj EPS $4.26–4.36, GAAP-profitable, and out-innovating (AI/robotics division). The bull's argument is the gap is too wide given Array's comparable domestic-content position and expanding margins. The bear's argument is that the discount reflects real inferiority: smaller, slower, no GAAP common earnings, a crater segment, and a preferred overhang Nextracker doesn't carry. The comp is not "cheap Nextracker" — it is "the clear #2 priced as the clear #2." Enphase/SolarEdge/Shoals are different sub-sectors (inverters/eBOS) and only loosely comparable; FTC Solar (EV/Sales ~0.14x) is the sub-scale failure case and a reminder that scale is destiny in this industry.

Lens 8 · Stock-Price Catalysts (what moves ARRY >5%)

Mostly ``, cross-checked to filing events.

• 52-week range ~$4.52 – $12.23; currently ~$7.93–$9.08. A stock that has nearly 3x'd off its trough and round-tripped inside a year — extreme policy-beta.
• Pattern of >5% moves:

1. Policy/legislative headlines dominate. The mid-2025 OBBB signing (Jul-4-2025: ITC termination for projects placed in service after 12/31/2027 unless construction begins before 7/4/2026) and the cascade of Treasury notices, FEOC rules, and AD/CVD determinations drove sector-wide double-digit swings. Array is a policy-beta name first, a fundamentals name second.
2. Earnings prints — Q1-2026 beat (revenue + adj EPS + record order book) → modest pop; 2024's STI impairments + lowered outlook → 52-week lows.
3. Tariff/trade rulings — the Feb-2026 SCOTUS ruling that IEEPA does not authorize tariffs (struck the reciprocal/fentanyl tariffs; collection ceased 2/24/2026), the temporary Section-122 10% surcharge (expires 7/24/2026), and the new Section-232 full-customs-value metal-tariff scheme (Apr-2026) all move solar-supply-chain sentiment.
4. Macro rates — developer financing economics swing with the Fed; "customers delay installations awaiting rate cuts."

• What the market actually reacts to for ARRY: US solar policy and tariffs, then book-to-bill, then margin. It is not an idiosyncratic-execution name — it trades with the policy tape. That is the single most important fact for sizing a position: your timing risk is a Treasury notice or a court ruling, not a product cycle.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO: Kevin Hostetler (since April 2022). Career industrials operator, 25+ years; previously CEO of Rotork plc (UK flow-control), where he ran a "Growth Acceleration Program" lifting margins and capital efficiency. Archetype: professional turnaround/operations manager, not founder. That fits Array's stage — this is a margin-and-execution story, and Hostetler's resume is margin-and-execution. The 2025–2026 playbook (resize STI, integrate APA, push domestic content, expand gross margin) is textbook industrial-operator behavior.
• Track record at Array (quantified): Took gross margin from a 2024 trough through two brutal impairment years and delivered FY2025 +40% revenue and Q1-2026 margin expansion to 28.2% GAAP / 30.7% adj — credible. But also oversaw the STI value destruction (two goodwill impairments totaling $338.6M across 2024–2025, a $29.5M inventory charge) — though much of STI's deterioration is Brazilian macro, the original STI thesis (bought Jan-2022) has been a value-destroyer on his watch.
• Skin in the game / ownership: Insiders hold only ~0.41% of the company; institutions ~83%. This is a PE-incubated public company: Oaktree Capital was the IPO parent and remains a ~10% owner; Chairman Brad Forth is ex-Oaktree MD / Neos Partners. Low direct insider ownership = limited founder-style alignment; the alignment that exists is institutional/PE.
• Capital-allocation history: Mixed-to-deteriorating on ROIC. (a) STI acquisition (2022) → $338.6M of impairments — a capital-allocation black mark. (b) APA acquisition (Aug-2025, $185.4M) — too early to judge; contributed $14.9M revenue but a $10.2M operating loss in Q1-2026 (loaded with deferred comp + amortization); thesis is foundation/racking cross-sell + synergies. (c) Debt management is the genuine bright spot: in 2025 they refinanced out of the Term Loan via the 2031 convertible notes (2.875%), repurchased $100M of 2028 notes at a gain, and in Feb-2026 expanded the revolver from $166M to $370M and pushed maturity to 2031 (Fifth Amendment, Goldman as agent) — de-risking the near-term balance sheet materially. (d) Capped calls raise the 2031 conversion price to $12.74 and the 2028 to $36.02, limiting dilution.
• Red flags (governance): Five APA leases are with related parties owned by APA management, and the principal APA seller still runs APA; the APA earnout/deferred consideration is tied to continued employment (compensatory) — related-party complexity to monitor, not yet a scandal. No dividend ever paid on common. Heavy reliance on a TRA (tax receivable agreement) owed to a former indirect stockholder.
• Net: competent industrial operator with a real margin story and excellent recent balance-sheet stewardship, carrying the scar tissue of a value-destroying international acquisition and the structural awkwardness of a PE-legacy cap table with thin insider ownership.

Lens 10 · Forensic Red Flags

Forensic-analyst lens. All figures research-layer: filings/ unless noted.

• The headline GAAP-vs-common gap (most important). Net income of $2.0M (Q1-26) becomes a $(13.5)M loss to common because of $15.5M/quarter of preferred dividends + accretion on the Series A Redeemable Perpetual Preferred (400,000 shares, $1,000 liq pref = $400M face, issued 2021 for ~$395.4M). It accrues at 6.25% PIK until the 5th anniversary — Aug 11, 2026 — then must be paid in CASH at 5.75%+ (stepping up 50bps on the 5th/6th/7th anniversaries, 100bps on the 8th–10th). Accrued unpaid dividends already total $98.5M. This is the dominant forensic feature: a large, compounding, soon-to-be-cash claim that sits senior to common and converts operating profit into common-shareholder losses. It is classified as temporary (mezzanine) equity, not debt — so leverage screens understate the true senior claim on the business.
• Earnings vs cash flow divergence. Q1-2026 GAAP net income +$2.0M but operating cash flow −$29.4M — explained by a $50.6M working-capital outflow (seasonal inventory + unbilled-receivable build ahead of the Q2/Q3 ramp). FY2025 OCF was +$101.8M on a GAAP net loss of $52.2M — the loss being non-cash impairment. So the divergences are explainable (impairment one way, seasonal WC the other), not obviously manipulative — but the rising inventory ($168M) and unbilled receivables ($110.5M) deserve monitoring; if the H2 ramp disappoints, that working capital becomes trapped.
• Goodwill / impairment history (a genuine flag). $338.6M of STI goodwill+asset impairment across 2024–2025 ($236M + $102.6M goodwill, plus $91.9M long-lived in 2024, plus the $29.5M inventory charge). Goodwill still carried at $135.2M (all Array Legacy) post-impairment; intangibles net $224.9M. The repeated impairments show the STI purchase price was badly wrong; the remaining Legacy goodwill looks defensible (no 2025 Legacy impairment indicators) but is not immune if US demand cracks.
• Non-GAAP flattery. The spread between adjusted EPS +$0.06 and GAAP EPS −$0.09 (Q1-26) is driven by add-backs for amortization of acquired intangibles, acquisition/deferred-comp expense, and — critically — adjusted metrics that sit above the preferred dividend line. "Adjusted net income per common share $0.65–0.75" guidance is before the full weight of the soon-to-be-cash preferred. SBC and acquisition deferred-comp ($5.0M in Q1) are real costs being normalized away.
• Tax volatility. Effective tax rate 51.6% in Q1-26 (vs 28.1% PY) and a swing to $23.0M tax expense in FY2025 despite a pre-tax loss — driven by a goodwill-impairment reversal for tax purposes (+$22.3M) and a Brazil valuation allowance. Tax is noisy and jurisdiction-mix-driven; not a red flag per se but reduces earnings-quality predictability.
• Related-party leases / TRA — see Lens 9. Monitor.

Regulatory findings (required sub-section).

• SEC Litigation Releases / AAERs: None. Per regulatory/regulatory-findings.md (fetched 2026-06-18, EDGAR EFTS LR+AAER, 2021–2026): 0 SEC LR, 0 AAER naming Array Technologies.
• Item 3 Legal Proceedings (10-K, quoted/labeled and updated ):
• Plymouth securities class action (filed May-2021, SDNY): alleged §10(b)/§11 misstatements/omissions tied to the Oct-2020 IPO and 2020–2021 secondary offerings. Dismissed with prejudice May-2023; the Second Circuit AFFIRMED the dismissal with prejudice on March 24, 2026 — i.e. this multi-year securities overhang is now resolved in Array's favor. A material removal of risk in the latest quarter.
• Derivative suits: the SDNY derivative action was voluntarily dismissed April 28, 2026 following the Plymouth affirmance; the Delaware Court of Chancery derivative cases remain stayed.
• SWSS arbitration (served Sep-2025): contractual/negligence claim by Sterling & Wilson Solar Solutions over a Bickleton, WA project; Array is vigorously defending, stay requested Apr-2026, no material loss contingency accrued, damages unknown but characterized as early-stage. Not material on current facts.
• Non-SEC enforcement (web search): No material FTC/DOJ/FDA/CFPB enforcement, consent decree, fine, or penalty against Array Technologies surfaced. The relevant "regulatory" exposure is trade/tariff (AD/CVD on imported cells/modules from SE Asia + India/Indonesia/Laos; the First Solar Section-337 TOPCon exclusion-order case that could constrain module supply to Array's customers) — these are industry trade actions, not enforcement against Array.
• Conclusion: No material regulatory or enforcement findings against the company — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-18. The historic IPO securities litigation is now resolved (dismissed with prejudice, affirmed Mar-2026).

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028 EPS)

Bottom-up from FY2025 actuals + management's reaffirmed FY2026 guidance. All outputs ``; the preferred drag is the swing factor — I project both adjusted (pre-full-preferred) and a common-shareholder GAAP-style view.

Anchors:

• FY2025 actual: revenue $1,284M, GAAP net loss $(52.2)M, GAAP loss to common ~$(118)M-equiv after ~$59M preferred (4 quarters × ~$14.5–15.5M), adj EBITDA ~$155M-range [implied; web Q-by-Q].
• FY2026 guidance: revenue $1.4–1.5B, adj EBITDA $200–230M, adjusted net income/share $0.65–0.75.
• Share count ~153M; preferred dividends rising (turns cash Aug-2026), ~$62–66M/yr and climbing.

FY2026 (base): Take guidance midpoint — revenue ~$1.45B, adj EBITDA ~$215M, adjusted EPS ~$0.70. On a common-shareholder GAAP basis, after $64M preferred, real D&A ($110M/yr intangibles + PP&E) and interest, GAAP EPS to common is likely roughly breakeven-to-slightly-negative (~$(0.05)–$0.10). The honest read: adjusted ~$0.70; GAAP-to-common ~flat.

• Bull FY2026: revenue $1.5B, adj EBITDA $230M, adj EPS $0.75; safe-harbor pull-forward + 100%-content share gains + STI stops bleeding.
• Bear FY2026: revenue $1.35B (safe-harbor demand pulls into early-2026 then air-pocket; project delays from transformer/interconnection + AD/CVD module shortages), adj EBITDA ~$180M, adj EPS ~$0.55, GAAP-to-common clearly negative.

FY2027 (base): This is the OBBB air-pocket year. Projects must begin construction before Jul-4-2026 to keep the ITC; demand likely pulls forward into 2026 and decelerates in 2027 unless technology-neutral credits + data-center load offset. Base: revenue ~$1.35–1.45B (flat-to-slightly-down), adj EBITDA ~$200–215M, adj EPS ~$0.65–0.70. The preferred turning cash-pay in Aug-2026 is a real FCF hit from FY2027 on.

• Bull FY2027: AI/data-center power demand + 100%-content moat sustain US volume; STI recovers to breakeven; adj EPS ~$0.85.
• Bear FY2027: post-safe-harbor demand cliff is severe; revenue −15–20%; adj EPS ~$0.40, GAAP-to-common deeply negative, FCF pinched by cash preferred.

FY2028 (base): Two-sided. The core linked-row patent expires Feb-5-2030 (still ahead), but the ITC fully terminates for projects placed in service after 12/31/2027 — FY2028 is when post-cliff demand reality bites and when "solar is cost-competitive without subsidy" gets tested. Base: revenue ~$1.4–1.55B (recovery if data-center/merchant solar fills the policy gap), adj EBITDA ~$210–235M, adj EPS ~$0.70–0.80.

• Bull: unsubsidized US utility-solar + AI load = secular growth resumes, adj EPS ~$1.00+.
• Bear: subsidy-cliff demand destruction + patent-cliff approaching = adj EPS ~$0.45, structural de-rating.

Forecast NOT logged — per --watchlist unattended rules, forecast.ts create is skipped (only log a Brier forecast on genuine committed conviction in a /thesis pass). The base call for tracking would be: "ARRY FY2026 adjusted net income/share ≥ $0.65" (p≈0.60, resolves 2027-02) — recorded here for the analyst, not written to the tracker.

Lens 12 · Bull vs Bear

Bull case. Array is the #2 in a structurally growing, oligopolistic industry (top-3 US suppliers = >90% share) trading at half the #1's multiple. It owns a datable patent moat to 2030 and — more importantly for this policy cycle — a 100%-domestic-content capability that makes its trackers the FEOC-safe, bonus-credit-eligible choice exactly as OBBB cuts off Chinese supply. Margins are expanding (28.2% GAAP / 30.7% adj, up YoY) even as revenue dips on seasonality + STI. The balance sheet was just de-risked (Term Loan repaid, revolver expanded to $370M / 2031, converts pushed to 2028/2031 with capped calls). The order book is at a record $2.4B with 2x book-to-bill, and the IPO securities litigation just got dismissed with prejudice (affirmed Mar-2026), removing a multi-year overhang. AI/data-center power demand is a fresh secular tailwind to US utility-scale solar. If STI merely stops bleeding and APA synergies land, the EBITDA base ($200–230M) supports meaningful equity value at 9x EBITDA. Earnings surprise vector: safe-harbor-driven 2026 demand pull-forward + margin beats.

Bear case (2–3 permanent-impairment risks).

1. The preferred is a permanent tax on common. $400M face compounding at 6.25%→cash from Aug-2026, $98.5M already accrued, $15.5M+/quarter — it converts operating profit into common-shareholder losses and will drain FCF in cash from FY2027. Unless redeemed/refinanced (expensive at this stock price), common holders sit behind a perpetual senior claim. This is not cyclical; it is structural.
2. The OBBB safe-harbor cliff is a demand time-bomb. Projects must begin construction before Jul-4-2026 to keep the ITC; the credit terminates for placements after 12/31/2027. The most likely shape is 2026 pull-forward → 2027 air-pocket. A tracker vendor with a $544M GAAP RPO that's 94% recognizable within 12 months has almost no multi-year visibility through the cliff — the $2.4B "order book" includes non-binding awarded orders that can evaporate if developer economics break.
3. Permanent #2 status + a fixable-but-unfixed crater. Nextracker out-scales (~2.8x revenue, >$5.25B backlog), out-earns (GAAP-profitable, $810M+ EBITDA), and out-innovates (AI/robotics). STI is at −37% gross margin and being rebuilt mid-air. Array is a price-taker on un-hedged steel/aluminum with no pricing power.

Pre-mortem (it's Dec-2027, the thesis broke — what happened?): The 2026 safe-harbor rush pulled demand forward; 2027 US utility-solar bookings fell off a cliff as the ITC expired and merchant/data-center demand didn't fill the hole fast enough. STI never recovered. Array's $2.4B "order book" shrank as awarded-but-non-binding orders were cancelled when developer IRRs collapsed. The preferred went cash-pay, FCF turned negative, and the stock de-rated to ~5x trough EBITDA. The 9x multiple was not "cheap" — it was the right multiple for a sub-scale #2 facing a policy cliff with a senior-claim overhang.

Are multiples too high? No — at ~9x EBITDA / 1.29x sales, Array is not expensively priced; the risk is earnings, not multiple. A multiple this low can still fall if EBITDA falls.

Contrarian view (what the market is refusing to see): The bears are over-anchored on the OBBB cliff and under-weighting that OBBB simultaneously hands Array a relative-share gift — by enforcing FEOC and domestic content, it kneecaps the Chinese-supplied competition and makes 100%-content trackers the only credit-safe option. The contrarian bet is that Array's share of a shrinking-then-stabilizing US market rises faster than the market shrinks, and AI-data-center load arrives sooner than the consensus 2027-air-pocket model assumes. The non-contrarian risk that everyone should see but the price doesn't fully reflect: the preferred turning cash in Aug-2026 is a near-certain, datable FCF event, not a tail risk.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the money machine? Demand is 100% policy- and rate-dependent, recognized over-time on short-cycle contracts with no multi-year backlog visibility ($544M GAAP RPO, 94% within 12 months). A single Treasury notice, court ruling, or rate surprise re-rates the whole sector — and Array trades as pure policy-beta (52-week range $4.52–$12.23 = a ~2.7x peak-to-trough). The short thesis doesn't even need Array to do anything wrong; it needs the 2027 ITC air-pocket to be as bad as the mechanism implies.
• Where is revenue concentrated, and what if it shifts? Now ~81% US and rising — a single regulatory regime. Top-2 customers ~26% of revenue. If two large IPP customers push projects right (transformer shortages, financing), a quarter craters (see Q1-26 STI −93%).
• Why is the moat weaker than bulls think? The flagship's core patent expires Feb-2030; design-arounds are already in market (Nextracker, GameChange). "100% domestic content" is matched by Nextracker — it is table stakes among the US top-3, not a unique edge. There is no pricing power (un-hedged commodity inputs, spec-taking customers).
• Most dangerous competitor bulls underestimate: Nextracker. It is not just bigger — it is GAAP-profitable, has a >$5.25B backlog (10x Array's GAAP RPO), and just launched an AI/robotics division. In an oligopoly, the scale leader compounds its lead (procurement, R&D, balance sheet). Array risks being permanently out-invested.
• Worst capital-allocation moves: the STI acquisition → $338.6M impaired. That is the tell on management's M&A judgment — and they just did another $185M deal (APA) that's currently running an operating loss with related-party leases and an employment-tied earnout. Pattern risk.
• What must hold for today's ~$9 price? FY2026 adj EBITDA $200–230M and a market willing to look through the 2027 cliff and the preferred not forcing a dilutive/expensive refinancing and STI stabilizing. That's a lot of "ands."
• If growth disappoints 20–30%: FY2027 revenue ~$1.0–1.15B, adj EBITDA ~$120–150M, GAAP-to-common deeply negative, FCF pinched by the now-cash preferred. At a de-rated 6x trough EBITDA → EV ~$720M–900M, less ~$680M net debt and the $400M+ preferred claim → equity value compresses violently. This is a high-operating-and-capital-structure-leverage name.
• Single scenario that permanently impairs: The 2027 ITC air-pocket + a stalled merchant/data-center bridge + the preferred going cash. Plausibility: moderate-to-high on the demand cliff (it's legislated), lower on it being permanent (solar economics + AI load argue for eventual recovery). The permanent impairment is to the common stake specifically, via the preferred — not necessarily to the enterprise.

Lens 14 · Management Questions (ordered by information value)

1. The Series A preferred turns cash-pay in August 2026 at 5.75%+, with $98.5M already accrued. What is the explicit plan — redeem, refinance, let it accrue at the default rate, or convert — and at what cost to common, given the current share price? (This single answer most changes the view — it's the structural overhang.)
2. Your GAAP RPO is $544M (94% within 12 months) but you report a $2.4B "order book." Walk me through the bridge: how much of the $2.4B is binding vs awarded-but-cancellable, and what cancellation rate have you actually seen when developer IRRs compress?
3. Model the 2026 safe-harbor pull-forward vs the 2027 ITC air-pocket. How much of FY2026's $1.4–1.5B guide is demand borrowed from 2027, and what is your internal FY2027 US-volume assumption?
4. STI ran a −37% gross margin in Q1-2026. What is the concrete path and timeline to breakeven — and at what point do you exit Brazil/Spain entirely rather than fund another impairment?
5. After $338.6M of STI impairment, why was APA the right $185M use of capital — and what specific, quantified synergies justify it by when? Why is it running an operating loss two quarters in?
6. Nextracker is ~2.8x your revenue, GAAP-profitable, with a >$5.25B backlog and a new AI/robotics division. Where do you structurally out-compete them in the next 3 years rather than just hold #2?
7. The core linked-row patent expires February 2030. What protects the franchise post-2030 — what's in the 216 pending applications that extends the moat, and how do you quantify the design-around threat today?
8. You don't hedge steel or aluminum and have no pricing power on inputs. With Section-232 metal tariffs now on full customs value, what's the realistic 2026–2027 CPW trajectory, and how much can 45X + domestic-content actually offset?
9. If 45X eligibility or the domestic-content bonus is altered by future Treasury rulemaking or FEOC reinterpretation, what's the dollar impact to gross margin? How much of your $164.7M IRA vendor-rebate receivable is at risk?
10. Operating cash flow was −$29.4M in Q1 with inventory and unbilled receivables building. What's the full-year FCF bridge, especially once the preferred is cash-pay, and do you stay FCF-positive in a 2027 demand-down scenario?
11. Insider ownership is ~0.41% and Oaktree remains a ~10% holder from the IPO. How is management's compensation tied to per-common-share value creation rather than adjusted metrics that sit above the preferred line?
12. Five APA leases are with related parties owned by APA's management, and the earnout/deferred consideration is employment-tied. What governance guardrails prevent related-party leakage as you integrate?
13. The First Solar Section-337 TOPCon case could trigger a general exclusion order constraining US module supply. How does a module-supply shock to your customers flow through to tracker demand — is it a net headwind or, via domestic-content, a relative tailwind?
14. AI/data-center load is the new secular bull narrative for US solar. How much of your current order book is explicitly tied to data-center-adjacent projects, and how durable is that demand if the AI capex cycle cools?
15. What single development over the next 18 months would most change your view of this business — and what's the leading indicator we should watch for it?

---