Nextpower (ex-Nextracker)

Refresh delta (since 2026-06-20)

What changed

1. FY27 Q1 printed 30 Jul 2026 and beat: record revenue $935.2M (+8% YoY, +6% QoQ), GAAP diluted EPS $1.07, adjusted EPS $1.20, adj. EBITDA $232.6M (24.9% margin), adj. FCF $105M.
2. Gross margin jumped to 35.9% GAAP / ~37% adjusted (from 32.6% a year ago) — but on refunds, not operations. Tariffs fell to $3.9M from $10.8M and the 45X credit rose to $103.3M. CFO Chuck Boynton attributed the beat to IEEPA tariff recoveries and called them "near-term benefits, not structural margin drivers."
3. An ~$830M acquisition wave — a step-change from FY26's ~$149M of tuck-ins. Prevalon Energy (BESS, up to $365M, closed 17 Jul 2026); Zigor/Apex Power (inverters, up to $80.5M, closed 30 Jul 2026); Zimmermann PV-Steel Group (Germany, up to €330M / ~$378M, announced 22 Jun 2026, expected to close H2 FY27). Two of the three are part-funded in stock — new for this company.
4. The policy clock moved the right way, and the prior dossier's cliff framing was wrong. On 6 Jun 2026 the US District Court for DC vacated IRS Notice 2025-42 in Oregon Environmental Council v. IRS, restoring the 5% safe harbor four weeks before the 4 Jul 2026 begin-construction deadline. Projects that began construction by that date get a four-calendar-year placed-in-service runwayRunwayHow long the cash lasts at the current rate of spending. It shortens the moment spending rises, which is why a figure taken from a quiet quarter flatters. (i.e. into 2030); only projects starting after 4 Jul 2026 face the 31 Dec 2027 deadline.
5. The mix deteriorated underneath the headline. US is now 83% of revenue (69% a year ago); Rest-of-World fell 40% YoY; over-time (project) revenue actually declined ~1% YoY — 100% of growth came from point-in-time components/software ($96.0M vs $17.0M). Revenue growth was price, not volume: MD&A says higher ASP "offset with lower GW delivered."
6. The stock de-rated hard through a beat-and-raise. ~$125 at the prior dossier → $95.35 on 28 Jul (−8.1%, after a −7.6% downgrade day on 13 Jul) → $103.53 on 9 Aug 2026. That is −17% since the boundary and −34% from the $156.40 ATH close (29 May 2026). Forward P/E is now 21.55x vs the ~26.5x in the prior dossier.

What held

The structural thesis stands, with one leg amended. Net cash ($1.21B) and zero funded debt — intact and stronger. Category leadership — 55% US / 30% global share — intact and slightly better than the prior dossier's ">50% US / ~30% global." Founder-CEO Dan Shugar in seat; asset-light manufacturing; backlog growing (>$5.5B, from >$5.25B). The 45X dependence is unchanged and slightly worse — $103.3M of COGS relief is 11.0% of revenue, above FY26's 10.7%. The TRA drag persists ($373.8M liability). The $500M buyback remains entirely undrawn.

The leg that changed: the prior dossier's core bear mechanic — a flat "placed-in-service cliff 31 Dec 2027" — was an oversimplification. The binding date was the 4 Jul 2026 begin-construction deadline, which has now passed with the 5% safe harbor restored. The demand air-pocket therefore likely sits further out (FY29–31) and is now conditional on an appeal, not on a fixed 2027 calendar date.

Phase A — Understand the business

Company Overview — REFRESHED (materially touched)

The 10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited.'s self-description has changed. NXT is no longer presented as a tracker company: it is "a leading global provider of solar and energy technology solutions for utility-scale power plants… an integrated suite of structural, electrical, and digital solutions across the full lifecycle."

In eight weeks the company added three genuinely new businesses:

Prevalon's stated economics: management guided $200–300M of FY27 revenue and >$300M of incremental backlog. Zimmermann's: ~€300M annual revenue and ~€45M adj. EBITDA — a 15% EBITDA margin against NXT's ~25%, i.e. accretive to EPS (≈7.3x EV/EBITDA on the headline price) but dilutive to margin. That trade-off is the single most important thing the new NXT has bought.

Note the honest correction: Prevalon was announced 28 May 2026 — three weeks before the prior dossier — and the prior dossier missed it, and consequently quoted the superseded 13 May FY27 guide ($3.8–4.1B / $825–900M) rather than the 28 May raise ($4.0–4.4B / $845–930M / adj EPS $4.30–4.73). Lens 5 and Lens 11 of that dossier were built on stale guidance.

New product launches 17 Jun 2026: NX Gemini 2P tracker and NX Anchor foundation system.

Supply Chain — unchanged, see the previous dossier

One flag for the next pass rather than a re-run: Zimmermann is owned German steel fabrication, which is a genuine departure from the ">100 contract-manufacturing facilities, close to no capital investment" model the prior dossier documented. If that deal closes, the capex-light characterisation needs re-testing.

Competitive Advantages — REFRESHED (moat shift)

• Share improved: 55% US / 30% worldwide (2025 basis, cited on the Q1 FY27 call), vs the prior dossier's ">50% US / ~30% global."
• The company is now litigating its moat. On 1 Jun 2026 NXT sued GameChange Solar in the US District Court for the District of Delaware, alleging infringement of three patents (issued 2022 and 2024) covering self-powered tracker technology and the TrueCapture™ control system, over GameChange's Genius Tracker systems and software. NXT seeks injunctive relief and damages; GameChange "denies the allegations and intends to mount a full and vigorous defense."
• Read: the prior dossier named GameChange "the most dangerous competitor bulls underestimate." NXT evidently agrees. Suing your fastest-growing private competitor over the software patents is a tell that the differentiation NXT most fears losing is TrueCapture, not steel. An injunction would be a genuine moat event; a loss would confirm the bear's commoditisation case. Binary, and unresolved.
• Moat surface widened by acquisition — BESS + inverters + European product lines + agriPV/carport/floating are new adjacencies, but each is a market where NXT is a new entrant, not the incumbent. Buying adjacency is not the same as owning it.
• Moat verdict: unchanged in shape — wide in the US on share + domestic content + installed base, narrow abroad. The US leg got more concentrated (83% of revenue) and therefore more policy-levered, not less.

Segments — REFRESHED (major mix moves)

Still one reportable operating segment; no segment EBITDA disclosed. The disaggregations now in our figures:

Three things this table says that the headline does not:

1. The project business shrank. Over-time revenue — the percentage-of-completion project line that is the tracker franchise — fell ~1% YoY. Every dollar of growth, and then some, came from the point-in-time line. That line is real (eBOS, components, software; record eBOS bookings, >$100M annual run-rate, foundations +50% YoY ) but it is being asked to carry the whole story.
2. Growth is price, not volume. MD&A: revenue rose on "a higher average selling price resulting from increasing costs per watt coupled with increased customer demand in the U.S, offset with lower GW delivered." The prior dossier's FY26 finding was the exact opposite — "+20% revenue on +13% GW, i.e. growth is volume + mix, not price (ASP under pressure)." That has inverted in one year. Pricing power returning is good; shipping fewer gigawatts while doing it is not, and the two together look like cost pass-through, not share gain.
3. Rest-of-World is in free-fall, not drift. −11% in FY26 became −40% in the quarter (Latin America and the Middle East down; Europe up). The international diversification that is supposed to offset US policy risk is going backwards — which is the clearest commercial rationale for buying Zimmermann.

---

Phase B — Measure performance

Earnings Result — FY27 Q1, reported 30 Jul 2026

P&L:

vs consensus: a beat on both lines — revenue $935.2M vs ~$932.3M consensus; adj. EPS $1.20. One widely-syndicated figure claims a "+71% beat vs a $0.70 estimate" — treat that as unreliable; a $0.70 consensus is incompatible with the $4.42–4.73 full-year guide and with the $1.04 Zacks-style estimate quoted the same week. The honest statement is: a modest beat, not a blow-out.

The margin story is the whole quarter, and it is borrowed. Gross margin rose 331 bps. The MD&A bridge: a $10.1M increase in the 45X credit (to $103.3M) and a $6.9M decrease in tariffs, net of refunds (to $3.9M, from $10.8M), partly offset by higher freight and logistics "as a result of the U.S.-Iran war" and acquisition headcount. Two consequences:

• 45X is now 11.0% of revenue and gross margin ex-45X is ~24.9%. The subsidy is not a rounding item; it is roughly the entire gross-margin premium over a commodity fabricator.
• The tariff relief traces to the 20 Feb 2026 Supreme Court ruling that IEEPA tariffs were unlawful, with refunds "with interest" expected but "timing and precise coverage… uncertain," plus the 6 Apr 2026 Section 232 revision that cut exposure on certain imported components. The CFO said the recoveries are not structural, with more expected in Q2. A gross margin that improves because a court struck down a tariff is not operating leverage.

Meanwhile operating margin fell: SG&A +36% (sales-organisation expansion, +$20.7M) and R&D +106%. Operating income grew 2.5% on 8.2% revenue growth. The company is spending its subsidy-and-refund windfall on opex, which may be the right call for a platform build-out — but it means there was no operating leverage in the quarter at all.

Balance-sheet flags — mostly benign, two to watch:

• Cash $1,213.9M (from $1,095.0M); no funded debt; interest expense $0.25M (facility fees only); investment-grade. Total equity $2,557.1M.
• The 45X receivable finally converted: down to $311.6M from $352.6M. This was the prior dossier's #1 quality-of-earnings worry and it improved. OCF $121.1M vs $81.3M a year ago; OCF/NI 0.73 (from 0.52) — still <1, but moving the right way.
• Contract assets rose $74.1M to $607.4M — revenue recognised ahead of billing, growing 13.9% QoQ against 6% sequential revenue growth. On an 89.7%-POC revenue base this is the line to watch; it is the classic place where POC optimism shows up before cash does. Not alarming yet; the trend is worth a stopwatch.
• TRA liability $373.8M, essentially flat; no TRA payment made in the quarter.
• Customer re-concentration: Customer A = $152.1M = 16.3% of revenue. The prior dossier's finding — "no single customer >10% in FY26, deconcentrating" — has reversed.

Guidance — raised twice, but the second raise was floor-only (now in our figures):

The shape of that raise matters more than the fact of it. Only the floor moved; the ceiling has been $4.4B / $930M / $4.73 since May. An analyst asked why, and got "a cautious response on maintaining visibility over three remaining quarters"; Zimmermann is excluded from guidance on closing-timing uncertainty. And the arithmetic is unflattering: Q1 adj. EPS $1.20 annualises to $4.80 — above the top of the guide. At the $4.575 midpoint, Q2–Q4 must average $1.125, below Q1. Management is guiding a declining EPS run-rate through a year in which revenue steps up from $935M toward ~$1.1B/quarter. That is a deliberate signal that the Q1 margin does not repeat.

Market reaction: the print landed well — an after-hours pop on the beat-and-raise — but it followed $95.35 on 28 Jul (−8.1%) and a −7.6% day on 13 Jul on an analyst downgrade. At $103.53 (9 Aug) the stock is still ~17% below the prior dossier's ~$125 and 34% below the 29 May ATH. The market took the beat and the raise and still paid less.

Sourced conflict, unresolved: the prior dossier recorded an all-time-high close of $156.40 (29 May 2026); stockanalysis.com now shows a 52-week range of $52.61–$163.13. Most likely a close-vs-intraday difference, but I have not verified it — do not treat either as the ATH without checking.

Earnings Calls — sentiment shift

transcripts/ is still empty and I could not fix it this run. The Q1 FY27 transcript (30 Jul 2026) exists on Seeking Alpha (paywalled — skipped per protocol); fool.com has Q4 FY26 but not Q1 FY27 (404 on the expected URL); Insider Monkey and Benzinga are not indexed for it. This lens is therefore built on syndicated call highlights, not a primary transcript. Treat it as secondary.

Call cast: Sarah Lee (IR), Dan Shugar (CEO/founder), Howard Wenger (President), Chuck Boynton (CFO). No management change surfaced this quarter — the same team, and Boynton signed the 8-KForm 8-KA filing for something that happened between reports and matters enough to tell shareholders now..

The shift vs the prior three-to-four calls is real and it is defensive-turned-acquisitive:

• The prior dossier read the FY25→FY26 arc as a pivot from "growth" vocabulary to "resilience / supply-chain flexibility / platform," with an AI-and-data-centre narrative overlay it judged "partly a defensive move to change the subject from the solar-ITC cliff." This quarter, management stopped narrating the pivot and bought it — $830M of deals whose press releases lead with "AI data center power supply," "GPU AI workload smoothing," "inertia support, grid stabilization." The story is now backed by assets, which is a genuine upgrade in credibility.
• Notably candid on margin. Boynton volunteering that tariff recoveries are "near-term benefits, not structural margin drivers" is the opposite of promotional. Management had every incentive to let a 37% adjusted gross margin speak for itself and did not. That earns trust and should be weighted as such.
• Notably cautious on guidance. Raising only the floor after a beat, and excluding a signed €330M acquisition from the outlook, is conservative behaviour from a team the market has taught to beat-and-raise. Either genuine caution about H2 conversion, or sandbagging. The prior dossier's watch-item — "watch whether non-tracker actually delivers +40%" — reads on track: non-tracker is ~14% of revenue, eBOS at a >$100M run-rate with record bookings, foundations +50%.
• What they are not saying much about: Rest-of-World, down 40%. The GW-delivered decline. Neither featured in the highlights; both are in the filing.

Sentiment read: more credible than a quarter ago on the platform story, more honest than average on margin quality, and quietly more cautious on the near-term. The tone gap between the press release (record revenue, record backlog) and the filing (fewer GW, RoW −40%, over-time revenue down) is the widest it has been.

Comps — brief update (multiple materially moved)

Read: NXT's own forward multiple compressed from ~26.5x to ~21.6x — a ~19% de-rating in eight weeks, against rising guidance. The prior dossier's central valuation complaint ("a demanding multiple… already double-to-triple its own 2023–25 ~6–8x EV/EBITDA average") is materially less true today. It is still not a cheap multiple for a business with 83% US revenue and 11%-of-revenue subsidy dependence. Peer multiples were not re-sourced this run — the ~3x premium-to-Array claim from the prior dossier is now stale and should not be repeated without a refresh. Sell-side consensus PTConsensus price targetThe average of what published analysts think the share price should be. An opinion poll, not a forecast. $142.52 across 28 analysts, "Strong Buy"; TD Cowen cut to $118 from $135, Hold, on valuation; BMO upgraded to Outperform, PT $132.

ROE: GAAP NI $165.4M on $2,557.1M equity → ~25.9% annualised. Unchanged in character — a genuinely high-return, low-capital model.

Stock-Price Catalysts

>5% moves since the boundary:

The catalyst set has changed shape. The prior dossier's four buckets (beat-and-raises; US solar policy; tariffs; sector/rate beta) still hold, but two are now inverted and one is new:

1. Tariffs flipped from a headwind to a (temporary) tailwind — the 20 Feb 2026 SCOTUS IEEPA ruling and the 6 Apr 2026 Section 232 revision cut the tariff line to $3.9M and generated refunds. Watch for the end of refunds, and for the threatened new Section 301 tariffs across "a broad range of countries" — that is the live negative.
2. The safe-harbor litigation is now the single biggest policy swing factor, replacing the 2027 calendar. If the government's appeal of the Oregon Environmental Council vacatur succeeds with retroactive effect, projects that safe-harboured on the 5% method before 4 Jul 2026 could be disqualified. Practitioners are explicit: "market participants should not treat the decision as final."
3. New: M&A execution and integration. Three deals in eight weeks, one still unclosed, two using stock. Analysts named integration risk and near-term margin dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. as concerns on the call. The acquisition push was cited as a reason for the de-rating.
4. New: the GameChange patent suit — an injunction or a loss are both step-changes for the US competitive structure.

Pattern: the market has stopped paying for beat-and-raise and started discounting the quality of the beat. That is a regime change in how this stock trades and the most important thing in this lens.

---

Phase C — Judge people & books

Management — REFRESHED (capital allocation materially changed)

Team unchanged: Shugar (CEO/founder, since 2013), Wenger (President), Boynton (CFO). No departures, no governance events surfaced.

Capital allocation is where this refresh bites. The prior dossier praised "disciplined and shareholder-friendly… ~$117M of tuck-in M&A at sensible prices… $500M buyback authorized." Eight weeks later:

• M&A spend went up ~5.5x. ~$830M of announced/closed consideration (Prevalon $365M + Zigor/Apex $80.5M + Zimmermann ~$378M) vs $149M for the whole of FY26. Cash at close is more modest ($150M + $46M + €180M ≈ $400M), with the rest in stock and earn-outs — so the balance sheet is not strained. But the pace is now aggressive, and the 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes.'s own caveat that acquired companies "may not have an adequate system of internal controls" applies to three more businesses.
• The buyback is dormant while the share count rises. Zero shares repurchased in the quarter; $499.6M of the $500M authorisation still available — unchanged since the January 2026 authorisation. Meanwhile shares outstanding rose 149.39M → 151.65M on option exercises and RSU vesting, and diluted shares rose 2.8% YoY. Add ~€105M of Zimmermann stock and $50M of Prevalon stock still to be issued. Management is using equity as currency and not buying it back at a 34%-off price. They may be right to preserve cash for deals; it is nonetheless the opposite of what the prior dossier's "positive signal" framing implied, and shareholders are being diluted, not bought in.
• SBC $29.6M in the quarter (3.2% of revenue), up 33% YoY — added back to every adjusted metric.
• The good version of this story: buying a BESS business with 1.3 GW of firm hyperscaler contracts, a US inverter line, and a European steel platform — while the stock the acquirer is issuing is one of the few US-listed solar names with net cash and investment grade — is exactly how a category leader converts a policy-levered hardware franchise into something more durable. Judgement should be reserved until Zimmermann closes and the integration shows up in margin.
• TRA unchanged: $373.8M liability, no payment in the quarter. Still 85% of realised tax benefits to Flex/TPG affiliates; still a structural drag on owner economics.

Forensic Red Flags — REFRESHED

Every figure unless noted.

1. 45X dependence deepened — the top item, unchanged in rank. $103.3M of COGS relief = 11.0% of revenue; ex-45X gross margin ≈ 24.9%. Credit steps down 25%/yr from 2030 and ends after 2032; torque tubes $0.87/kg, structural fasteners $2.28/kg through 2029. FEOC disqualification of suppliers remains an open, unquantified risk — Treasury's Notice 2026-15 (12 Feb 2026) is interim guidance only and NXT says it "is currently evaluating itself and its supply chain partners."
2. Margin quality — new this quarter. The GM beat came from tariff refunds the CFO declined to call structural. Any model that extrapolates 35.9% GAAP / 37% adjusted gross margin is extrapolating a court ruling.
3. Contract assets +$74.1M (+13.9% QoQ) to $607.4M on 6% sequential revenue growth, with 89.7% of revenue on POC. Deloitte's Critical Audit Matter is POC cost estimation. Nothing here is an exception — but the gap between revenue recognition and billing widened, and OCF/NI is 0.73. Monitor.
4. Customer re-concentration — reversal of a prior-dossier positive. Customer A = 16.3% of quarterly revenue; a year ago no customer exceeded 10% for the full year.
5. Litigation — status changed. NXT is now an active plaintiff (patent suit vs GameChange, D. Del., 1 Jun 2026). No material litigation against the company disclosed. Regulatory findings: SEC EDGAR EFTS returned 0 Litigation Releases and 0 AAERs for 2021-08-10 → 2026-08-10 (regulatory/regulatory-findings.md, regenerated 2026-08-10). No non-SEC enforcement surfaced in web search.
6. Goodwill $489.0M — flat in the quarter, and about to jump. Three closings land in FY27 Q2–H2; intangibles rose only $6.0M gross. Future impairment risk scales with the deal wave.
7. Deferred tax asset $509.0M + TRA $373.8M — the coupled, policy-sensitive pair, essentially unchanged.
8. Steel still unhedged; Section 232 at 50% on covered steel/aluminium/copper full customs value, with threatened new Section 301 tariffs.

Conclusion: books remain clean — unqualified Deloitte opinion, ICFR effective, no regulatory findings, net cash. The forensic story has shifted from "is the 45X receivable collectible?" (improving) to "how much of this quarter's margin is non-repeating, and is POC running ahead of cash?"

---

Phase D — Project & stress-test

Forward Projection — FY27/FY28/FY29 adj. EPS

Model status — read this before the table. model.xlsx was rebuilt this run (our model --force). Integrity flags, honestly:

• Quarters of history: 5 (2025-Q2 → 2026-Q2) — up from 0; opening balance sheet sourced from our figures: yes.
• "No computed values." our model reports 51 formulas and no cached results — the workbook must be opened in Excel and saved before any output exists. Value per share, EV, FY+3 EPS and the balance check are all blank.
• Share price is blank; CAPM inputs unset.
• Three backfilled quarters carry null capex/FCF/depreciation (cash-flow statements not read).

Therefore I cite no model numbers. Everything below is `` with the arithmetic shown, anchored on filings and guidance.

Anchors: FY26 adj. net income $687.5M → FY26 adj. EPS ≈ $4.52. FY27 guide (30 Jul): revenue $4.1–4.4B, adj. EBITDA $870–930M, adj. EPS $4.42–4.73 (mid $4.575) — excluding Zimmermann. FY27 Q1 actual adj. EPS $1.20. Implied FY27 conversion: adj. NI ≈ $714M on adj. EBITDA $900M → adj. NI / adj. EBITDA ≈ 0.79.

The shape that matters — and it is different from the prior dossier. That dossier's central claim was that "FY27's strength is partly borrowed from FY28" via a 31 Dec 2027 placed-in-service cliff. The mechanic was mis-specified. The binding date was 4 Jul 2026 begin-construction, and projects clearing it get four calendar years to be placed in service — into 2030. Combined with the 5% safe harbor being restored on 6 Jun 2026, the pull-forward window closed more favourably than the prior dossier assumed, and the demand air-pocket moves out to roughly FY29–FY31.

What replaced the cliff is worse in one respect and better in another: better, because the runway is years longer and the safe-harboured cohort is large; worse, because the risk is now binary and legal — an appellate reversal with retroactive effect — rather than a knowable calendar. You cannot underwrite a court.

Against consensus: the prior dossier cited FY28 consensus ~$5.69. My base FY28 of ~$4.90 is ~14% below that. The disagreement is specific and testable: consensus models margin re-expansion; I model margin dilution, because (a) the two closing acquisitions carry ~15% EBITDA margins against NXT's ~25%, (b) the buyback is dormant while share count rises ~1.5–2.8%/yr, and (c) this quarter's margin came from tariff refunds management said will not repeat. If FY28 adj. EBITDA margin holds above 22% while revenue clears $4.8B, I am wrong.

Valuation at $103.53: 22.6x base FY27, 21.1x base FY28. Base FY28 at a 20x multiple ⇒ ~$98; bull FY28 $5.80 at 25x ⇒ ~$145; bear FY28 $3.90 at 13x ⇒ ~$51 — and note the 52-week low is $52.61, so the bear case has already been touched inside a year.

Per protocol: no our model create in an unattended run — no committed base case logged.

Bull vs Bear — updated

Bull case (strengthened since June). The category leader — 55% US, 30% global — just posted record revenue and a record >$5.5B backlog, raised guidance twice, and used eight weeks to buy three businesses that convert a solar-tracker company into an energy-infrastructure platform with a real AI-data-centre leg: Prevalon's >6 GWh of deployed BESS and 1.3 GW of firm hyperscaler supply contracts, UL-certified US inverters, and a German steel platform with 20 GW deployed across 15 new countries. The balance sheet funded it without borrowing a dollar: $1.21B cash, zero debt, investment grade, ~26% ROE. The policy clock — the prior thesis's central fear — moved in NXT's favour: the 5% safe harbor was restored four weeks before the 4 Jul 2026 deadline, giving the safe-harboured cohort a runway to 2030. The 45X receivable is converting. Non-tracker is executing (eBOS >$100M run-rate, record bookings; foundations +50%). And the stock is 34% off its high at ~21.6x forward. If the appeal fails and the platform integrates, this is a high-return compounder bought at a de-rated multiple.

Bear case (permanent-impairment risks).

1. The margin is rented. 11.0% of revenue is a 45X credit that steps down from 2030 and dies after 2032, and this quarter's 331 bps of gross-margin expansion came from IEEPA tariff refunds the CFO explicitly called non-structural. Ex-45X gross margin is ~24.9%. Strip both and this is a mid-20s-gross-margin steel-and-electrical fabricator, priced as a platform.
2. The core business is not growing. Over-time (project) revenue fell ~1% YoY; GW delivered declined; revenue grew on ASP driven by "increasing costs per watt" — i.e. cost pass-through. RoW −40%. US is now 83% of revenue, so the diversification that was supposed to hedge policy risk is going backwards while the policy exposure concentrates.
3. The cliff became a lawsuit. The safe-harbour runway rests on a district-court vacatur the government is expected to appeal, where reversal could be retroactive. That is an unhedgeable, binary risk sitting under FY28–30 US volume.
4. Dilution instead of return. Zero buyback against a $500M authorisation while shares rose 2.8% YoY and two acquisitions pay in stock — at a price 34% below the high.

Pre-mortem (18 months out — it is early 2028 and the thesis broke). The appeal landed for the government in mid-2027 with retroactive effect. Developers who had safe-harboured on the 5% method in June 2026 lost eligibility; the US pipeline that looked locked through 2030 re-priced overnight. NXT's backlog — always "not take-or-pay," always subject to cancellation "for convenience" — converted at 70 cents on the dollar. Meanwhile the three acquisitions arrived on schedule: Zimmermann's 15% EBITDA margin and Prevalon's integration costs pulled blended adj. EBITDA margin to 18%, and the goodwill from $830M of deals took a first impairment. Rest-of-World never stopped shrinking. Tariff refunds ended and new Section 301 duties landed. FY28 adj. EPS printed $3.90 against a consensus that had been $5.69, the multiple went to 13x, and the stock revisited the low $50s. Nothing was fraudulent; the earnings were simply rented from the US Treasury and a district court, and both landlords gave notice.

Are multiples too high? Much less so than eight weeks ago. ~21.6x forward, down from ~26.5x, on raised guidance. It is a fair multiple for a #1-share, net-cash, ~26%-ROE franchise — and a demanding one for a business with 83% US revenue, 11%-of-revenue subsidy dependence, a declining core project line, and a legal binary under its demand runway. Fairly valued on my numbers; cheap only on consensus numbers I cannot source.

Contrarian view (what the market is refusing to see). In June the market's error was ignoring the cliff. Today the error may be the mirror image: the market de-rated NXT ~17% partly because of the acquisition push — treating $830M of deals as empire-building — at the exact moment the deals were buying the one thing the bear case said the company lacked (non-US, non-tracker, non-subsidised revenue). Simultaneously, the sell side is still modelling FY28 margin re-expansion that the acquisition mix arithmetically prevents. Both the price and the estimates look wrong, in opposite directions — which is why the stance is watch, not act.

Devil's Advocate (short-seller) — updated

• The tell is in the disaggregation, not the headline. A company reports "record revenue, +8%" while its percentage-of-completion project revenue declines, its gigawatts delivered fall, and 100% of growth comes from a point-in-time line that sextupled off a $17M base. That is a business whose core is flat-to-shrinking, dressed in a mix shift. The prior year's growth was volume; this year's is price. Price-led growth in a commoditising product with falling volumes is late-cycle, not early.
• The margin will not survive contact with a normal quarter. Management said so. 45X is 11% of revenue and legislated to die; tariff refunds are one-time; freight is rising on the U.S.–Iran war. Ex-45X gross margin ~24.9% is fabricator economics.
• Why the moat is weaker than bulls think. NXT is suing GameChange over software patents — an implicit admission that the steel is not defensible and TrueCapture is what is left. If the suit fails, the bear's commoditisation thesis is confirmed by NXT's own choice of battlefield. Meanwhile GameChange takes the growth markets while NXT's RoW falls 40%.
• The acquisitions are the new risk, not the new moat. Three deals in eight weeks — one BESS (a market with brutal Chinese cell-driven pricing), one inverter line (entering against Sungrow/SMA), one German steel fabricator at 15% EBITDA margins. Each dilutes margin; each consumes management bandwidth that a 106% R&D ramp already claims; each adds goodwill. Analysts raised exactly this on the call. Buying four adjacencies at once is what companies do when the core is decelerating.
• Governance/allocation: TRA still routes 85% of realised tax benefits to Flex/TPG; the buyback sat at zero while the stock fell 34% and the company issued stock to sellers. Owner economics are leaking at both ends.
• Assumptions that must hold for ~$103: (1) the Oregon Environmental Council vacatur survives appeal, or replacement guidance is no worse; (2) 45X survives FEOC scrutiny across a >100-facility contract-manufacturing base; (3) three acquisitions integrate without margin or goodwill accidents; (4) point-in-time revenue keeps compounding fast enough to offset a flat-to-down project line; (5) RoW stops falling. If (1) fails retroactively, base EPS goes to the bear path (~$3.90 FY28) and a 13x multiple ⇒ a stock near $50 — a level printed within the last 52 weeks ($52.61 low).
• Single scenario that permanently impairs the business: retroactive disqualification of the 2026 safe-harbour cohort coincident with 45X supplier FEOC disqualification. That would strip demand and subsidy simultaneously, and the US-domestic-content premium — the entire non-price moat — would evaporate on both sides at once. Plausibility: low-to-moderate. Still not a fraud story — clean audit, net cash, real market leadership. It is a rented-earnings story.

Management Questions — unchanged, see the previous dossier (15 questions)

Four to add at the top of that list, arising from this quarter:

1. Quantify the non-repeating portion of Q1's gross margin — IEEPA refunds recognised, refunds still expected, and where gross margin settles once they stop.
2. Over-time revenue declined YoY and GW delivered fell while ASP rose. Is the core tracker business growing in volume terms in FY27, and what is the GW guide?
3. What is your legal read on the Oregon Environmental Council appeal, and what share of backlog depends on the 5% safe-harbour method used before 4 Jul 2026?
4. With the stock 34% off its high, why was zero of the $500M buyback used while you issued equity to acquisition sellers?

---