ARRY — Array Technologies, Inc. · 2026-09-06 · Verdict: WATCH · Conviction 3

Price $4.60 (screen close, 2026-09-05) · Mkt cap $708.4M · EV $1,058.9M (screen basis, preferred excluded) · EV/EBIT n/m (CY2025 EBIT -$29.0M) · FCF yield 11.3% · Net debt $350.4M, or $856.8M including the Series A preferred at liquidation preference · ADV20 $27.2M Sources read: 10-K filed 2026-02-25 (Items 1, 1A, 7), 10-Q filed 2026-08-05 (Q2 2026 MD&A), DEF 14A 2026-04-07, 8-Ks 2026-05-06, 2026-07-16, 2026-08-05, 2026-08-20, 2026-08-31 with EX-99 exhibits, Form 4 summary (12m), triage note 2026-09-04. No transcript in the pack.

1. What the business actually does

Array makes single-axis solar trackers, the motorised mounts that rotate panel rows to follow the sun, plus fixed-tilt racking, foundations, control software and field services for utility-scale solar. Its patented linked-row design drives multiple rows from one motor, with the core U.S. patent running to February 5, 2030; competitors are Nextpower (formerly Nextracker), PV Hardware and GameChange Solar, and 81% of 2025 revenue was U.S. (10-K 2025, Item 1). Two segments: Array Legacy Operations, which now includes APA Solar, bought August 2025 for $185.4M, and STI Operations in Spain and Brazil, bought January 2022. On August 31, 2026 Array closed the $203M purchase of Affordable Wire Management, paying about $165M cash (8-K 2026-08-31, Item 8.01).

2. Why it might be mispriced: the edge case

The honest label is misunderstood segment plus a screen artifact. Trailing CY2025 GAAP EBIT is negative $29.0M, but that is a $102.6M STI goodwill impairment and a $29.5M STI inventory charge sitting on a legacy business that earned $300.0M of gross profit on $1,070.5M of revenue (10-K 2025, Item 7). Consolidated revenue then fell 15% in H1 2026, entirely STI, down 83% to $27.8M, while Array Legacy plus APA grew 6% to $537.7M at a 30.5% gross margin (10-Q Q2 2026, MD&A). STI is now under 5% of revenue with its goodwill written off, so the melting piece is nearly melted. Below the line, $31.4M of H1 preferred dividends and accretion turned $26.3M of net income into a $5.1M loss to common (8-K 2026-08-05, EX-99.2).

The offset is that the same capital structure makes the screen's $1,058.9M EV wrong: it omits $506.4M of Series A Redeemable Perpetual Preferred liquidation preference (10-Q Q2 2026), which ranks ahead of the common forever. I cannot document a forced seller. The register is passive-heavy, BlackRock 10.8% and Vanguard 10.2%, with Hill City Capital at 9.3% (DEF 14A 2026-04-07). This looks like momentum abandonment after a 42% twelve-month decline, not a dislocation.

3. Unit economics and growth

Q2 2026 revenue $342.1M, down 6%; gross margin 29.1% versus 26.8%; adjusted EBITDA $63.3M at 18.5% versus $63.6M at 17.5% (8-K 2026-08-05, EX-99.1 and EX-99.2). Legacy gross margin reached 30.7%. That recovery matters because FY2025 legacy margin collapsed from 41% to 28% on a 22% rise in cost per watt: 13 points from tariffs, 6 from lower 45X amortisation, 3 from inflation (10-K 2025, Item 7). Section 45X credits are therefore a live, policy-dependent input to gross margin, now subject to foreign-entity-of-concern limits Treasury is still defining (10-Q Q2 2026, MD&A).

Growth quality is weaker than the headline. Legacy Q2 revenue rose 10% on a 25% ASP increase against a 12% volume decline; H1 was 15% ASP against 7% lower volume (10-Q Q2 2026, MD&A). Watts shipped are shrinking while dollars per watt rise on APA foundation content and tariff pass-through, so the screen's +40.2% CY2025 growth flag is stale. Cash conversion is lumpy: H1 free cash flow $76.7M, being negative $36.9M in Q1 and positive $113.6M in Q2 (8-K 2026-08-05, EX-99.2).

4. Balance sheet and capital allocation

At June 30, 2026: cash $307.3M, long-term debt $657.7M (2028 converts $321.5M at 1.00%, due December 1, 2028; 2031 converts $336.3M at 2.875%), Series A preferred carried at $498.2M against a $506.4M liquidation preference, total stockholders' equity negative $202.1M (10-Q Q2 2026). The February 2026 Fifth Amendment lifted the revolver to $370.0M and extended it to 2031, and other debt was fully repaid in Q2. Then $165M went out for AWM on August 31.

The dated fact that decides the framing: on or before August 10, 2026 Array could pay the Series A dividend by accruing it to the liquidation preference at 6.25%; after that date dividends are payable only in cash, and unpaid dividends accrue at the Cash Regular Dividend Rate plus 200 basis points (10-Q Q2 2026, MD&A). That anniversary passed three weeks before the screen date, so about $31M a year of preferred service is now a cash claim ahead of the common. Management's own August slide lists "Pref Service as % of FCF" and "Managing convert maturities" as focus areas while setting the leverage target on "core operational" net debt at or below 2.5x, a definition excluding the preferred (8-K 2026-08-20, EX-99.2). Insiders own 985,864 shares, under 1% as a group, with zero open-market purchases in twelve months (DEF 14A 2026-04-07; Form 4 summary, 12m).

5. Management: said versus did

Hostetler raised FY2026 adjusted EBITDA to $210-230M from $200-230M and adjusted EPS to $0.68-0.75 at Q2 (8-K 2026-08-05, EX-99.1); flat adjusted EBITDA on 6% less revenue is real execution. Against that, STI has absorbed $102.6M of goodwill impairment in 2025 on top of $327.9M of goodwill and long-lived impairment in 2024 (10-K 2025, Item 7), and the response has been $388M of further acquisitions, AWM at 8.8x trailing EBITDA and guided to be at least high-single-digit accretive to adjusted EPS in year one before synergies (8-K 2026-07-16, EX-99.1).

6. Valuation

The guide requires a Q4 larger than any quarter in company history by half again. FY2026 revenue of $1.4-1.5B less H1 actual $565.5M less Q3 guidance of $310-330M leaves $525-605M for Q4, against roughly $225M in Q4 2025 and a best-ever quarter near $394M (derived: FY2025 $1,284.1M less H1 $664.6M less an implied Q3 near $394M from the disclosed 18.3% adjusted EBITDA margin on $72.2M). The $2.5B orderbook, up 37% with a 1.5x trailing book-to-bill, supports it (8-K 2026-08-05, EX-99.1). But the ITC terminates for facilities placed in service after December 31, 2027 unless construction began before July 4, 2026, and Notice 2025-42 removed the 5% safe harbor for utility-scale projects (10-Q Q2 2026, MD&A). An orderbook that peaked into a begin-construction deadline is not obviously a run rate.

Base: 2026 lands at the guide midpoint and AWM adds about $23M of EBITDA (implied by $203M at 8.8x), so run-rate adjusted EBITDA is roughly $240M; pro forma EV including the preferred at liquidation preference and $165M less cash is about $1.73B, or 7.2x, and 7.5x gives about $6.30 a share. Bear: Q4 disappoints and 2027 EBITDA reverts to $160M in a post-deadline air pocket; at 5.5x the residual behind the preferred and converts is nil to $1.00. Bull: the orderbook is share gain, 2028 EBITDA reaches $300M at the high-teens margins guided for APA (8-K 2026-08-20, EX-99.2), and 8x gives about $10.50. Weighting 50/25/25 gives roughly $5.55 against $4.60. Reverse DCF: at $4.60 the market pays $708M for a residual behind $506M of perpetual preferred and $658M of converts, implying about $113M of steady-state free cash flow to common capitalised near 16%, or that cash flow at a 12% cost of equity declining about 4% a year forever.

7. Catalysts and timeline

Q3 2026 results in early November: the first quarter carrying mandatory cash preferred dividends, the first with AWM, and the point where the Q4 shape is reaffirmed or cut. Refinancing of the 2028 converts, needed well before December 1, 2028. Any redemption or refinancing of the Series A. Treasury's proposed 45X foreign-entity rules.

8. Risks and pre-registered kill criteria

Policy is the dominant risk, customers are concentrated (10-K 2025, Item 1A), and the equity is a levered stub. Kill criteria: 1. Q3 2026 revenue below the $310M guidance floor, or FY2026 revenue guidance cut below $1.4B at the Q3 print. 2. Orderbook below $2.2B at any quarter end, or trailing-twelve-month book-to-bill below 1.0x. 3. Array Legacy Operations gross margin below 27% for two consecutive quarters (30.7% in Q2 2026). 4. Array elects not to pay the Series A dividend in cash for any quarter after August 10, 2026, letting it accrue at the penalty rate.

9. Verdict and one-paragraph summary

WATCH, conviction 3. Array looks cheap on the screen because the screen omits a $506M perpetual preferred and reads a trailing EBIT loss that is really two STI write-offs; underneath, the operating business grew 6% in the first half at a 30.5% gross margin with a record $2.5B orderbook, guides to $210-230M of 2026 adjusted EBITDA, and throws off roughly $113M of free cash flow to common after debt service, capex, tax and preferred dividends, a 16% yield on a $708M market cap. Two things stop it being an idea today. First, hitting even the low end of the revenue guide needs a fourth quarter of $525M against roughly $225M last year and a best-ever $394M, and the orderbook supporting it was built into the July 4, 2026 ITC begin-construction deadline, so pull-forward and backlog are the same object until Q3 separates them. Second, the August 10, 2026 anniversary just turned about $31M a year of preferred dividends into a mandatory cash claim ahead of a common stock that already carries negative $202M of book equity and a $321.5M convert maturing in December 2028. Revisit after the Q3 print confirms the fourth quarter, and note that no insider has bought a share on the open market in a year.

Research for discussion, not investment advice. Positions and sizing are the reader's decision.