First Solar

Phase A — Understand the business

Lens 1 · Company Overview

First Solar designs and manufactures thin-film cadmium-telluride (CdTe) photovoltaic solar modules and sells them, by the gigawatt, to utility-scale solar project developers and operators — not to homeowners, not to commercial rooftops. It is the only Western module maker at scale that does not use crystalline-silicon (c-Si); its entire technology, cost structure, and moat derive from being the credible non-Chinese, non-silicon alternative.

• Product: Series 7 CdTe module — 2.8 m², up to ~540 W, ~19.7% total-area efficiency ``. Larger steel-back-rail form factor reduces install time.
• What it sells: modules only (it exited the systems/EPC and O&M businesses years ago). Revenue is recognized as control of modules transfers to the customer.
• Customers: US utility-scale developers and IPPs. The US was 96% of 2025 net sales `` — this is, functionally, a US policy company with overseas factories.
• Manufacturing footprint: United States (Ohio ×3, Alabama, Louisiana, + a 6th US "finishing" facility in South Carolina due H2 2026), plus Malaysia, Vietnam, and India ``.
• Contract structure — the crown jewel: a 50.1 GW contracted backlog worth $15.0B aggregate, expected to recognize as revenue through 2030 ``. ~23.2 GW of that carries price adjustments tied to future technology improvements — potential additional revenue of up to $0.6B, mostly 2027–2028. The backlog gives FSLR multi-year revenue visibility almost no other manufacturer in any industry can match. Caveat (see Lens 10/13): the contracts can be amended, cancelled, or repriced, and India volumes with unsecured payment are excluded from the $15.0B.
• The economic engine you must understand: under IRA Section 45X, FSLR earns ~17 cents per watt on each US-made module sold to a third party, recognized as a reduction to Cost of Sales ``. This credit — refundable or sellable — is the single largest driver of reported profit (Lens 5, Lens 10).

Plain-terms summary: A vertically-integrated, single-technology module factory whose product is ~15% less efficient than the Chinese best-in-class but is bankable, domestically-made, tariff-protected, and IRA-subsidized — and whose order book is sold out for years. The business model is "build US capacity → harvest 45X credits → fill a pre-sold $15B backlog."

Lens 2 · Supply Chain

Upstream inputs → First Solar → end customer, named stakeholders along the chain:

• Raw materials (upstream): tellurium (a scarce byproduct of copper refining — the genuine geological chokepoint for CdTe), cadmium, glass (large-format float glass), aluminum/steel framing, encapsulants. Tellurium supply is the structural constraint on how big CdTe can ever scale; the Joule "Roadmap to 100 GWDC" literature flags CdTe supply-chain limits explicitly ``. First Solar has historically secured tellurium via long-term supply deals and its own recycling loop.
• First Solar (the node): wafer-free. Unlike c-Si makers it skips polysilicon → ingot → wafer → cell entirely; it deposits CdTe directly onto glass. This is why it is immune to the polysilicon price cycle and to Xinjiang polysilicon forced-labor (UFLPA) detentions that periodically freeze c-Si imports — a real, underrated supply-chain moat.
• Manufacturing: US fleet (Ohio, Alabama, Louisiana, South Carolina finishing) + Malaysia, Vietnam, India ``.
• Downstream customers (named): large US utility-scale developers/IPPs. The filings name BP Solar / Lightsource bp and Lightsource Renewable Energy Trading as (now-terminated) counterparties ; **Pine Gate Renewables** is a known utility-scale buyer in the ecosystem . Other large utility-scale buyers across cycles have included Intersect Power, Leeward, and the major IPPs serving data-center load.
• Chokepoints / single-source dependencies:

1. Tellurium — finite byproduct supply caps CdTe's total addressable scale.
2. The US policy stack — 45X credit + domestic-content ITC bonus + AD/CVD tariff wall. This is a political single point of failure, not a physical one, but it is the dominant chokepoint.
3. Customer concentration in a single end-market (US utility-scale) exposed to interest rates, interconnection queues, and developer solvency.

This lens passes the "names or it didn't happen" test: tellurium, BP/Lightsource bp, Pine Gate, the US Ohio/Alabama/Louisiana/SC + Malaysia/Vietnam/India fleet.

Lens 3 · Competitive Advantages (moats)

How the product compares: On raw efficiency, FSLR loses — Series 7 at ~19.7% module efficiency sits ~15% below TOPCon (23.2%) and ~20% below heterojunction (24.8%) ``. FSLR does not win on performance-per-watt. It wins on four durable, mostly non-technical moats:

1. The domestic-content / policy moat (the real moat). FSLR is the largest American-owned, American-made module supplier. Its modules qualify customers for the domestic-content ITC bonus, and FSLR itself harvests the 45X manufacturing credit. Under OBBBA's new FEOC rules (2026), Chinese-owned or Chinese-influenced manufacturers are excluded from 45X — which widens FSLR's relative advantage ``. This is a regulatory moat, durable only as long as the policy persists (the bull/bear crux).
2. Bankability. FSLR's 25+ year track record (93 GW of modules sold to date ``), audited financials, US balance sheet, and warranty credibility make its modules financeable by US tax-equity and debt providers in a way that a thinly-capitalized Chinese newcomer's are not. For a 30-year utility asset, bankability is worth a premium.
3. Technology differentiation / IP. CdTe is a fundamentally different physics path — wafer-free, polysilicon-free, better temperature coefficient and spectral response in hot/humid climates, and a proprietary process the Chinese c-Si complex cannot simply copy. Record CdTe cell efficiency is now 23.1% in the lab ``, and the CuRe + perovskite-tandem roadmap (Lens 12/13) is the efficiency catch-up plan.
4. Tariff insulation. As a US producer and a petitioner in the AD/CVD cases (it co-filed the trade complaints against Cambodia/Malaysia/Thailand/Vietnam c-Si imports), FSLR is structurally long the tariff wall ``.

Bargaining power: Strong over customers right now — a sold-out $15B backlog and a scarce domestic-content product mean developers need FSLR more than FSLR needs any single developer (demonstrated by FSLR terminating BP/Lightsource for breach rather than accommodating them — Lens 10). Power over suppliers is moderate (tellurium is the squeeze point). The asymmetry favors FSLR only while US supply is scarce; if domestic c-Si capacity (now 50 GW and rising) floods in, pricing power erodes.

Lens 4 · Segments

First Solar reports as a single operating segment (modules); there is no product-segment P&L to break out (it exited systems). The meaningful disaggregation is geography of sales and revenue by recognition driver:

• Geography: US = 96% of 2025 net sales ``. The remaining ~4% is rest-of-world (India, and historically EMEA/APAC). The trend is increasing US concentration as the IRA pulls volume domestic — a double-edged sword (subsidy capture, but single-policy exposure).
• Revenue trajectory (consolidated, ``):
• FY2023 net sales $3,318.6M, gross profit $1,300.7M (39.2% GM)
• FY2024 net sales $4,206.3M (+26.7% YoY), gross profit $1,857.9M (44.2% GM)
• FY2025 net sales $5,219.4M (+24.1% YoY), gross profit $2,120.3M (40.6% GM)
• Q1-2026 net sales $1,044.2M (+23.6% YoY vs $844.6M), gross profit $486.1M (46.6% GM) ``
• Why GM moves: the 45X credit volume (more US-made GW sold = more credit reducing COGS) and freight. Q1-2026's 6-point GM jump to ~47% was explicitly attributed to "higher qualifying Section 45X volumes and reduced freight" ``. The segment story is the policy story — there is no "high-margin software line" hiding here; the margin is the subsidy plus operating leverage on a sold-out plant.
• Volume: produced 16.1 GW, sold 17.5 GW in 2025 ``.

---

Phase B — Measure performance

Lens 5 · Earnings Result (Q1 FY2026, reported 2026-04-30)

The latest print was a clean beat-and-hold.

• Net sales $1,044.2M, +23.6% YoY (vs $844.6M) — record Q1 revenue ``.
• Gross profit $486.1M → 46.6% gross margin, up ~580 bps from 40.8% . Driver: higher qualifying 45X volumes + lower freight .
• Operating income $345.3M (vs $221.2M), op margin ~33.1%. R&D rose to $66.9M (from $52.4M) — deliberately, to fund CuRe/perovskite (Lens 12) ``.
• Net income $346.6M; diluted EPS $3.22 (vs $1.95, +65%) . **Adjusted EBITDA ~$520M** .
• Guidance — reaffirmed (unchanged): FY2026 volume 17.0–18.2 GW, net sales $4.9–5.2B, adjusted EBITDA $2.6–2.8B, capex $0.8–1.0B ``. Tone: confident, "record start," but explicitly not raised — management is sandbagging into policy uncertainty.
• Balance-sheet flags: net cash position — cash & equivalents $2,363.0M + marketable securities $63.6M vs long-term debt of only $237.2M . Total assets $13,351.1M; total equity $9,878.6M. **Operating cash flow was −$214.9M for Q1** (seasonal working-capital draw; materially better than the −$608.0M prior-year Q1) . Net cash fell to ~$2.0B from ~$2.4B on the South Carolina capex + WC ``. AR $1,374.0M and inventory $893.9M (current) both rising with the build — watch in Lens 10.
• Market reaction / what's priced in: the stock trades ~$257.70 (Jun 20 2026) ``, near the upper end of the analyst range — the beat was largely expected; the share price is now driven more by policy headlines than by quarterly prints (Lens 8).
• Unusual vs. own history: the ~47% gross margin is a peak for FSLR and is structurally subsidy-inflated; do not extrapolate it as a "normal" module margin.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts/ empty), so this is ``. Across the last several calls the management arc has shifted:

• 2024 calls: defensive/uncertain — pre-election, heavy hedging on IRA durability, "constructive dialogue with both parties."
• Mid-2025 (post-OBBBA): relief and confidence — OBBBA "fully retained the IRA's 45X manufacturing credits through 2032" ``, which removed the worst tail risk. Tone pivoted to capacity execution and backlog conversion.
• Q1-2026 call: clearly upbeat — "record" framing, margin-expansion narrative, and a new emphasis on perovskite/CuRe R&D (tying the $67M R&D quarter to the technology roadmap) ``.
• Recurring phrases: "domestic content," "bankability," "differentiated technology," "disciplined capital allocation," "long-term contracted backlog."
• What they stopped saying: the existential "will the IRA survive" hedging — replaced by FEOC/domestic-content offense. The new watch-item phrase is anything softening on ASPs, India, or customer credit (the three pressure points).

Lens 7 · Comps

There is no clean US-listed pure-play CdTe peer; FSLR's "peers" split into (a) US solar-adjacent names with very different business models and (b) the Chinese c-Si module giants that are the real cost competitors. Multiples are `` with date or n/a.

Read: The "solar industry average forward P/E" was cited around ~18x, with one bear note pegging FSLR at a discount to that on forward numbers . On FY26E EPS ~$17.40 and a $257.70 price, FSLR trades ~**14.8x forward** — not demanding if the 45X-fattened earnings are durable, expensive if you haircut earnings to a pre-subsidy base. The honest comp problem: FSLR's P/E is being applied to a number that is ~100%+ tax credit (Lens 10), so the multiple understates the policy risk. EV/Sales, EV/EBIT, and 5-yr ROE were not sourced to a citable figure and are left n/a rather than fabricated.

Lens 8 · Stock-Price Catalysts (last ~5 years)

The defining feature of FSLR's tape: it is a policy/election stock, not an earnings stock. The >5% moves cluster around Washington, not around the print.

• Nov 2024 — Trump election win: −29.3% over the episode (−16.1% intraday) ``. The single largest move in the window, and it was political, not operational.
• Sep 2024 — Harris judged to win the first debate: +15.2% ``. Again, pure policy beta.
• Jul 2025 — OBBBA signed: a relief rally — feared incentive cuts came in "not as negative," and crucially 45X was retained through 2032 ``.
• 2024–26 — recurring rate-sensitivity: FSLR repeatedly sold off on rising Treasury yields (utility-scale solar economics are discount-rate sensitive) even on otherwise positive news (e.g., it "slid as rising Treasury yields pressured rate-sensitive solar stocks despite an India partnership headline") ``.
• AD/CVD tariff determinations (SE Asia c-Si duties of 34%–652%+) acted as positive catalysts for FSLR ``.

What the market actually reacts to: (1) federal solar policy (IRA/45X/ITC/FEOC), (2) the interest-rate path, (3) tariff rulings, and only then (4) the quarterly print. A FSLR position is, first, a bet on US energy policy and rates — that is the dominant variable to underwrite.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO Mark Widmar (CEO since 2016; prior CFO at FSLR). A finance-trained, disciplined operator — credited with the strategic exit from systems/EPC, the pivot to a pure module-manufacturing model, the Series 6→7 transition, and building the contracted-backlog machine. Archetype: professional manager / capital-disciplined operator, not a charismatic founder. Appropriate for this stage — FSLR needs flawless capex execution and contract management, not visionary risk-taking.
• Track record (quantified): revenue $3.3B (2023) → $5.2B (2025), EPS $7.74 → $14.21, gross margin held ~40–44% through a brutal global oversupply that crushed most of the industry ``. Built a sold-out 50.1 GW / $15B backlog. This is a genuinely strong multi-year operating record.
• Capital allocation: conservative and shareholder-aligned. No dividend, minimal debt ($237M long-term), large net-cash buffer reinvested into US capacity (the 6th plant) to capture 45X — i.e., capex is directly ROIC-accretive while the credit lasts. Notably, FSLR has been cautious about buybacks despite the cash pile, preferring growth capex; this is defensible given the policy-window opportunity but means shareholders don't get a cash return.
• Skin in the game / insider activity: insider-transactions.csv not on disk. Per web, Widmar has been a net seller — ~69,980 shares sold over the trailing year, no open-market purchases, mostly routine vesting-driven ``. Not alarming on its own (typical for option/RSU comp), but there is no insider buying signal to lean on, and insider ownership is modest (professional-manager, not founder-owner).
• Red flags (governance): none egregious. The cleanest concern is strategic dependence on a single policy (45X) — a strategy risk more than a governance one. Comp is performance-weighted; no related-party deals surfaced.

Lens 10 · Forensic Red Flags

Forensic equity-analyst lens. FSLR's accounting is clean-looking on the surface — the issue is not fraud, it's what the headline numbers actually represent.

1. THE central flag — earnings are a subsidy. FSLR recognized ~$1.6 billion of Section 45X credits as a reduction to Cost of Sales in 2025 ``. FY2025 net income was $1,528.2M. So the 45X credit ($1.6B) exceeds reported net income — on a pre-credit basis, FSLR's 2025 module business is roughly breakeven to modestly profitable. The ~46–47% gross margins are not an industrial achievement; they are a US fiscal transfer running through COGS. Every valuation multiple in Lens 7 is being applied to a tax-credit-inflated earnings base. This is the most important sentence in the dossier.
2. Tax-credit monetization / counterparty + indemnity risk. FSLR doesn't just earn the credit, it sells it: $857.2M of 2024 credits sold for $818.6M cash, and $701.9M of 2025 credits sold for $668.1M cash . It **indemnifies buyers** if the IRS disallows any portion — a contingent liability that "may have a material impact" if triggered . Cash-flow quality is genuinely good (FY25 OCF $2,057.1M > net income), but a chunk of that cash is credit-sale proceeds, not pure module operating cash.
3. Revenue-recognition / backlog quality — the BP/Lightsource tell. In 2025 FSLR terminated master supply agreements with BP Solar Holding and Lightsource Renewable Energy Trading for breach, triggering $384.6M of contractual termination payments — of which only $61.0M was recognized as revenue — and is now litigating in the NY Supreme Court to collect the rest . This proves the $15B backlog is *not* iron-clad: large counterparties can and do default, and collection is uncertain. Combined with **Pine Gate Renewables' Chapter 11** (Nov 2025) , developer-credit risk inside the backlog is a live, quantified concern.
4. Working-capital build outrunning sales. AR $1,374.0M and inventory $893.9M (current) are rising as the US fleet ramps ``. With Q1 operating cash flow at −$214.9M (seasonal), watch whether receivables/inventory growth continues to outrun revenue — early warning of channel/credit stress.
5. Variable consideration in revenue. A meaningful slice of backlog revenue carries price adjustments (technology, freight, commodity, wattage, domestic-content, tariff) `` — recognized revenue can be revised up or down, adding estimation risk to reported sales.
6. SBC / non-GAAP: modest relative to the subsidy distortion; not the main story.

Regulatory findings (required sub-section). Read from regulatory/regulatory-findings.md (fetched 2026-06-20):

• SEC Litigation Releases: None found naming First Solar in 2021-06-20 → 2026-06-20 (EDGAR EFTS LR search) ``.
• AAERs (accounting/auditing enforcement): None found in the same window ``.
• Non-SEC enforcement (web): No material FTC/DOJ/FDA/CFPB enforcement action, consent decree, or penalty against First Solar surfaced ``. (Note: FSLR is a petitioner/plaintiff in AD/CVD trade actions and the plaintiff in the BP/Lightsource collection suit — i.e., it is litigating offensively, not a defendant in an enforcement matter.)
• 10-K Item 3 (Legal Proceedings): FSLR's own disclosure centers on ordinary-course matters plus the BP/Lightsource collection complaint it filed in NY Supreme Court ``; no disclosed material securities or accounting litigation against the company.
• Verdict: No material regulatory or accounting enforcement findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-20. The forensic concern here is earnings composition (subsidy dependence), not integrity — the books appear honestly kept; they just describe a policy-dependent business.

---

Phase D — Project & stress-test

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

Built bottom-up from FY2025 actuals + reaffirmed FY2026 guidance. All outputs ``; inputs labeled.

Anchors: FY2025 EPS $14.21 ; FY2026 guidance: vol 17.0–18.2 GW, sales $4.9–5.2B, adj EBITDA $2.6–2.8B ; consensus FY2026 EPS ~$17.40 ; ~107.5M diluted shares .

The number that matters more than EPS: does the policy window hold? 45X runs 2023–2032 with phase-down from 2030 ``. The 50.1 GW / $15B backlog largely delivers inside that window — so the base case is well-supported for 2026–2028; the cliff risk is 2029–2032, which is where any DCF should apply a steep terminal haircut.

Tracked forecast (for calibration): base case → "FSLR FY2026 non-GAAP EPS ≥ $17.00, p≈0.62, resolves 2027-02-28." (Per --watchlist rules, NOT logging via forecast.ts create in the unattended loop — recorded here for the human-gated pass.)

Lens 12 · Bull vs Bear

Bull case. First Solar is the single cleanest way to own US energy-policy tailwinds with industrial-grade execution. It has (a) a sold-out $15B/50.1 GW backlog through 2030, (b) a net-cash fortress balance sheet funding ROIC-accretive 45X-capturing capacity, (c) a widening regulatory moat as OBBBA's FEOC rules lock Chinese-linked rivals out of the credit and AD/CVD tariffs wall off SE Asian c-Si, (d) structural supply-chain immunity (no polysilicon, no Xinjiang/UFLPA exposure, no wafer cycle), and (e) a secular demand super-cycle — US power demand growing up to 3.5%/yr through 2040, much of it data-center load ``, against a domestic module-supply shortage that FSLR is positioned to fill. The earnings surprise lever is ASP: if domestic-content scarcity persists, FSLR raises prices into a captive book.

Bear case (permanent-impairment risks).

1. The earnings are a subsidy with a clock on it. ~$1.6B of 45X credit > FY25 net income; the phase-down starts 2030 and a future Congress could accelerate it. Strip the credit and FSLR is a low-margin commodity manufacturer ~15% behind on efficiency. The quality of the earnings is the bear thesis.
2. Technology obsolescence. CdTe at ~19.7% module efficiency is being lapped by TOPCon (23.2%) and, longer-term, perovskite-silicon tandems (lab 34.85%) ``. If tandems commercialize before FSLR's perovskite/CuRe roadmap (1 GW pilot only in 2027), FSLR's product becomes the low-efficiency option without the policy crutch.
3. US oversupply collapses pricing power. US module capacity hit 50 GW in 2025 and is rising ``; the very subsidy that built FSLR's moat is funding the domestic competitors who will erode its ASPs.
4. Backlog is softer than it looks (BP/Lightsource termination, Pine Gate Chapter 11, India volumes excluded from the $15B, variable-consideration repricing).

Pre-mortem (18 months out, thesis broke): A 2027 Congress trims or accelerates the 45X phase-down OR a rate spike + developer-bankruptcy wave (Pine Gate was the first domino) stalls utility-scale buildout → backlog cancellations spike, ASPs roll over as 50+ GW of US c-Si floods in, the subsidy-inflated multiple re-rates to a commodity-manufacturer multiple on a pre-credit earnings base. The stock halves not because the company failed operationally but because the market re-prices what the earnings are.

Are multiples too high? At ~14.8x FY26E `` the headline multiple is reasonable; the honest answer is the multiple is too high relative to pre-subsidy earnings power and about right relative to the policy-supported base case. The market is refusing to fully discount the 2030 cliff — that's the contrarian view: consensus treats 45X-fattened margins as a run-rate, when they are a dated fiscal transfer.

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

Dismantling the bull case.

• What structurally breaks the model: repeal/acceleration of the 45X phase-down, or its FEOC/eligibility tightening. Remove ~$1.6B of credit and ~46% gross margins compress toward low-20s; the entire bull thesis (cheap on forward EPS) inverts because the "E" was mostly subsidy. One Act of Congress is the kill switch.
• Revenue concentration: 96% US, single end-market (utility-scale), exposed to a handful of large developers. The BP/Lightsource breach ($384.6M triggered, $61M recognized, now in court) and Pine Gate's Chapter 11 are not hypotheticals — they are proof that concentrated backlog counterparties default. If 2–3 large developers fail in a rate-stress scenario, the $15B backlog shrinks fast.
• Why the moat is weaker than bulls think: the moat is policy, not product. FSLR is ~15% behind on efficiency and 5c/W behind on cost vs LONGi ($0.22 vs ~$0.27/W ``). The instant the policy wall lowers (2030 phase-down, or a domestic-supply glut), it competes on the losing axis.
• Most dangerous competitor bulls underestimate: not the Chinese giants (walled off by tariffs/FEOC) — it's the fleet of US c-Si manufacturers being subsidized into existence by the same 45X credit (50 GW and climbing), plus perovskite-tandem startups that could leapfrog CdTe on efficiency before FSLR's 2027 pilot scales.
• Capital allocation gripe: hoards cash, no dividend, no meaningful buyback, while the CEO is a net seller — shareholders are asked to fund capacity that only pencils while the credit lasts, with no cash return as compensation for the 2030 policy cliff.
• Assumptions that must hold for today's price: (1) 45X intact and fully monetizable through 2029+; (2) ASPs hold despite domestic oversupply; (3) backlog converts with few cancellations; (4) no perovskite leapfrog before ~2028; (5) rates don't choke utility-scale demand.
• If growth disappoints 20–30%: apply a haircut to a pre-credit margin (~low-20s GM) and FSLR's normalized EPS is closer to $8–10, not $17 — at even a 12–14x "commodity manufacturer" multiple that is a $100–140 stock, roughly half the current ~$258. The low analyst target of $150 `` is the market's version of this scenario.
• Single scenario that permanently impairs: perovskite-tandem commercialization plus a 45X acceleration — product and policy moat gone simultaneously. Plausibility: low-to-moderate this decade, but non-trivial by 2029–2031, exactly when the credit phases down.

Lens 14 · Management Questions (15, ordered by information value)

1. On a pre-45X basis, what was FSLR's 2025 module gross margin and operating margin — and what is the realistic normalized margin once the credit phases down from 2030?
2. What is your base-case assumption for the 45X phase-down trajectory 2030→2032, and how much of the current backlog delivers after 2029?
3. Of the 50.1 GW / $15B backlog, what share is with counterparties you would rate investment-grade, and what is your current estimate of cancellation/default risk given BP/Lightsource and Pine Gate?
4. How much of the BP/Lightsource $384.6M termination claim do you expect to actually collect, and by when?
5. What ASP trajectory are you underwriting for US modules through 2028 as domestic c-Si capacity scales past 50 GW — and where do you lose pricing power?
6. What is the commercial-readiness timeline and target efficiency for CuRe and perovskite-tandem, and what happens to your competitive position if c-Si tandems hit mass production before your 2027 pilot scales?
7. Given the net-cash balance sheet and a dated subsidy window, why not return capital via buyback/dividend rather than build more capacity that only pencils while 45X lasts?
8. What is the tellurium supply runway at your 2027–2030 volume plan, and at what production scale does tellurium become the binding constraint?
9. How exposed is your Malaysia/Vietnam/India fleet to US AD/CVD and FEOC rules — could your own overseas modules face duties selling into the US?
10. What are the FEOC compliance requirements you must certify from 2026, and what is the cost/operational impact?
11. How sensitive is your customers' project economics — and therefore your order flow — to a sustained higher-for-longer interest-rate environment?
12. What is the IRS indemnification exposure on the ~$1.5B+ of credits you've sold, and how do you reserve for disallowance risk?
13. How do you think about FSLR's value if the US reverts to a less subsidy-rich policy regime after 2028 — what's the standalone, unsubsidized equity story?
14. What is the realistic ceiling on US utility-scale solar demand from data-center load, and how much of that can CdTe physically serve given efficiency vs land constraints?
15. What would have to be true for you to not exercise/sell future 45X credits — i.e., under what scenario does the credit stop being accretive?

---