← All thesesFSLRNeutralMedium convictionTuesday 17 November 2026עברית

First Solar

This is the archived full research. Read the story

Makes solar panels a different way, as a thin coating on glass instead of silicon wafers, mostly in American factories, for the companies building large solar farms.

Our viewNeutral Medium conviction
The scores
The profile · 1-4
Growth2/4 ↓

Revenue guided at or below last year and backlog fell 5.0 GW in six months. The profit that did grow is a credit.

Business quality3/4 ↑

The only large US maker whose domestic line imports no cell or polysilicon. But Qcells has just built the same position in silicon.

Financial position4/4 →

Cash of $1.688B against $37.6M of debt, capital spending funded from operations, no maturity inside the horizon and no financing need.

Execution record2/4 ↓

A pattern of slippage: revenue guided below last year, the second South Carolina phase pushed to mid-2027, the Asian capacity decision deferred again.

Per share capture4/4 →

Share count up 0.20% in a year, no debt, no convertible, no share-funded deals. Nothing stands between operating value and the holder.

Potential3/4 ↑

Bounded and largely committed: 45.1 GW contracted to 2030, roughly a year of the whole US market. The price upside is expected, not signed.

Risk profile2/4 →

Two policies decide the profit and both are externally controlled: a credit Congress can cut, a floor Commerce can waive. Two customers, a tenth each.

Valuation3/4 ↑

About 6.0 times guided cash profit, 117% of a value near $139 with the credit counted, between the base case and the $215 bull case.

Evidence quality3/4 →

Central claims now rest on the proclamation's own annex and the filings. The first pass read the scope from industry commentary and got it wrong.

The Verdict

We rate First Solar Neutral. It makes solar panels as a thin coating on glass, mostly in American factories, and the business is solid, but its whole profit is a government manufacturing credit that shrinks from 2030. A new minimum price on imported panels and cells may lift rivals' costs, though by how much is not established. Valued as the credit still to come plus the factories' own future cash, the shares are worth about $139 against $162, a fair price that already pays a little for the floor working. New order prices would decide which way it goes.

The long caseAmerica needs more electricity again and solar is the fastest to build; First Solar makes its panels at home, so new import rules raise its rivals' costs, not its own.

How we weighed it

The price against our value

Cuts both ways

Price at 117% of our value

The price sits a little above our value, inside the margin either side, and well short of the bull case.

The business

Cuts both ways

Solid

A solid maker, but its whole profit is a government credit with an end date.

Growth against what the price assumes

Against it

Sales fall this year and orders are shrinking.

What could go wrong

Against it

The import floor may not bite, the credit can be cut and rivals are building cells at home.

Against competitors

Cuts both ways

No filed rival compares directly; its profit is mostly the credit.

How its group is behaving

Cuts both ways

It sits in none of our ranked groups; trade and energy policy move it.

The latest news and report

Cuts both ways

The shares fell a fifth in a week on election talk, with no company news; the last report left our question open.

What moved it from Unattractive

The shares fell about a fifth with no company news, back inside a fifth of our unchanged value; the floor's test is three weeks away and the value already prices the credit's end.

The evidence under this - 8 claims
C01facthigh confidence
First Solar makes a solar panel from cadmium telluride on glass, containing no polysilicon.
current state · 2026-06-30written 2026-09-15
C05estimatemedium confidence
The 2025 credit was slightly larger than all of operating income: before it the company lost $3.1M.
derivedGross profit of $2.120B (40.62% of $5.219B revenue) less the $1.6B credit in C03 leaves $520.3M; operating costs were $523.5M, being gross profit less the $1.597B operating income in C04.
against this
  • The credit is disclosed as $1.6B to one decimal place, so the residual is inside its own rounding: anything from $1.55B to $1.65B leaves operating income between +$47M and -$53M. The sign is uncertain; the order of magnitude is not.
  • Operating costs in 2025 included production start-up expense for Louisiana, which a steady-state business would not carry.
historical · FY2025written 2026-09-15
C07estimatemedium confidence
At every point of 2026 guidance, operating income before the credit is negative, between -$80M and -$425M.
derivedEach end of the guided range in C06 taken with its own credit assumption: $2.4B gross profit less $2.19B of credits less $635M of costs is -$425M; $2.6B less $2.10B less $610M is -$110M.
against this
  • Guided operating costs include $90M to $100M of production start-up expense for South Carolina, and guided gross profit is struck after $115M to $135M of costs for factories deliberately held idle. Adding both back leaves roughly break-even rather than a loss, so this is a company investing through the line rather than one that cannot make money.
  • The pairing of the worst gross profit with the largest credit is arithmetic on the stated ranges, not a disclosed combination; management does not say the two ends move together.
  1. 2026-10-29editedthe credit assumption narrowed to $2.07B to $2.19B from $2.10B to $2.19B in the October 29 release, lifting the top of the range
forwardwritten 2026-09-15
C08facthigh confidence
From 4 December 2026 imported solar modules face a minimum price of $0.38 per watt and cells $0.22, plus a 15% tariff.
forwardwritten 2026-09-15
C22facthigh confidence
The Section 45X credit runs to 2032 and begins phasing down in 2030.
forwardwritten 2026-09-15
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C50inferencemedium confidence
On a three-stage fair value at an 8% required return First Solar is worth about $139 a share against $162.45 at the 16 November close, where enterprise value is about 6.0 times guided 2026 adjusted EBITDA. The module business is valued before the Section 45X credit, and the roughly $8B of credit still to come is added beside it. Base case: 2025 sales of $5.22B fall 9% in 2026, then grow 5% a year, slowing toward 4% by year ten, each year closing 30% of the gap; the operating margin before the credit moves from about minus 3% to 10% over five years; each added dollar of sales needs about $1.30 of capital, the average the filings show; a 15% tax rate; then 4% growth for ever at a 12% return on new capital. Bear (a 5% fall then flat, a 3% margin, new capital earning 8%) gives about $95, bull (12% and 10% growth, an 18% margin) about $215. As a risk check, at the 10% return asked of a cyclical or indebted business the base case gives about $118.
against this

The credit's value rests on the thesis's rough $8B (C28); every $1B more adds about $9 a share. If the import floor lifts bookings well above $0.36 a watt the value nears the bull case's $215, about a third above the price.

  1. 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
  2. 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
  3. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  4. 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten
  5. 2026-11-17editedRe-priced at the 16 November close of $162.45 after a 19% fall with no company news; the value itself is unchanged at about $139, and the price is now about 117% of it.
current state · 2026-11-16written 2026-11-09
C10management claimmedium confidence
Since the July call the company booked about 1.4 GW in the United States at about $0.36 per watt, inclusive of technology adjusters, and about 4.9 GW in the US so far in 2026 at about $0.35 to $0.36.
against this
  • The figure is inclusive of technology adjusters that have not yet been earned, so it is a contracted maximum rather than cash received.
  • Management said the volume was contracted or in late-stage negotiation, with price and terms largely set, before the Section 232 floor was announced, so it does not yet show what the floor does to price.
  • Volume fell from 1.9 GW the quarter before, and about 4 GW of optional volume in existing contracts still sits at older prices.
  1. 2026-10-29editedrefresh date moved: the October 29 release states no new bookings or price; the call, not yet on the vendor, is owed
  2. 2026-11-11editedre-sourced from the Q3 2026 call: 1.4 GW of US bookings at about $0.36 a watt since July, replacing the Q2 figure
historical · Q3 2026written 2026-09-15
01

The company

Almost every solar panel starts as silicon, melted, sawn into wafers, made into cells in one plant and assembled in another. First Solar skips all of that: it coats a sheet of glass with a thin film of cadmium telluride, cuts it into cells with a laser and seals it under a second sheet.

Its buyers build large solar farms, and many choose it for the tax credits an American-made panel unlocks: about 41 GW of its 45 GW of contracted orders ask for home-grown content.

The evidence under this - 3 claims
C01facthigh confidence
First Solar makes a solar panel from cadmium telluride on glass, containing no polysilicon.
current state · 2026-06-30written 2026-09-15
C37inferencemedium confidence
First Solar's process has no cell step and no polysilicon, so its US production imports nothing the measure covers.
derivedThe 10-K describes one continuous line taking a sheet of glass to a finished module, with cadmium telluride deposited as a film rather than cells being made and assembled (C01). The measure covers polysilicon, ingots and wafers, cells and modules (C08, C09). None of those is an input to a US-made First Solar module.
against this
  • It does import semi-finished product from Malaysia and Vietnam to feed the South Carolina finishing line, and how that is classified on entry is not established here - if it enters as a module it is covered.
  • Glass, steel, copper and the tellurium itself are imported and face other tariff regimes; the claim is only that the polysilicon measure misses them.
  • The 15% tariff and the floor apply at entry, so a rival assembling in the US from imported cells pays on the cell, not the module - a smaller base than the module floor implies.
forwardwritten 2026-09-15
C14management claimmedium confidence
About 41 GW of the 45 GW backlog carries some form of domestic content requirement.
current state · 2026-06-30written 2026-09-15

How it makes money

It sells panels by the watt under contracts signed years ahead, and collects a manufacturing credit of roughly 17 cents on each American-made watt it sells. The credit is booked as a lower cost, so it shows up as profit: without it, 2025's $1.597B of operating profit would have been a $3.1M loss.

The evidence under this - 5 claims
C11facthigh confidence
The contracted backlog was 44.0 GW at September 30, 2026 for an aggregate transaction price of $13.5B, about $0.31 a watt, recognised through 2030, down from 45.1 GW for $13.6B at June 30.
  1. 2026-10-29editedthe Q3 release reports 44.0 GW; the dollar value and price per watt wait for the 10-Q
  2. 2026-11-11editedthe Q3 10-Q gives the backlog's value, $13.5B for 44.0 GW, which the October release did not
current state · 2026-09-30written 2026-09-15
C02management claimhigh confidence
The company expects a Section 45X credit of roughly 17 cents per watt on each US-made module sold to a third party.
against this

The rate is the company's own reading of its eligibility as a vertically integrated producer, and the 10-K says the credit phases down from 2030 and that modifications to the law or its interpretation could change it.

forwardwritten 2026-09-15
C03facthigh confidence
First Solar recognised $1.6B of Section 45X credits as a reduction to cost of sales in 2025.
historical · FY2025written 2026-09-15
C04facthigh confidence
Operating income was $1.597B in 2025 on revenue of $5.219B.
historical · FY2025written 2026-09-15
C05estimatemedium confidence
The 2025 credit was slightly larger than all of operating income: before it the company lost $3.1M.
derivedGross profit of $2.120B (40.62% of $5.219B revenue) less the $1.6B credit in C03 leaves $520.3M; operating costs were $523.5M, being gross profit less the $1.597B operating income in C04.
against this
  • The credit is disclosed as $1.6B to one decimal place, so the residual is inside its own rounding: anything from $1.55B to $1.65B leaves operating income between +$47M and -$53M. The sign is uncertain; the order of magnitude is not.
  • Operating costs in 2025 included production start-up expense for Louisiana, which a steady-state business would not carry.
historical · FY2025written 2026-09-15
Where the money comes from, stream by stream
StreamShare of sales (FY2025)How it earns it
Solar modules, United States95.7%Thin-film panels sold by the watt to utility-scale developers and power producers, mostly under multi-year contracts and mostly made in US plants.
Solar modules, India3.7%Panels from the Indian plant sold in rupees to Indian developers.
Solar modules, other countries0.6%Panels invoiced in France and elsewhere, a remnant of the European business.

Who runs it

Mark Widmar has run First Solar since 2016, after five years as its finance chief, and Alexander Bradley has run finance since then. Insiders own about 0.4% of the shares, so the people running it are paid to run it rather than owners of it. Their bonus rides on profit before start-up costs, which the government credit alone clears, then on cost per panel, research, American sales and cash; long-term shares ride on future orders and a new technology, none on the share price.

The evidence under this - 4 claims
C43facthigh confidence
Mark Widmar has been chief executive since July 2016, having joined as chief financial officer in April 2011; Alexander Bradley has been chief financial officer since October 2016, having joined in May 2008; Michael Ahearn, chief executive from 2000 to 2009, chairs the board.
current state · 2026-04-02written 2026-10-04
C44facthigh confidence
As of March 19, 2026 the 19 directors and executive officers as a group owned 423,164 shares, under 1% of the 107,450,760 outstanding (about 0.4%); Michael Ahearn's 113,227 and Mark Widmar's 102,798 were the largest stakes. Vanguard (12.8%), BlackRock (10.4%), SIG (7.9%), FMR (6.1%) and Citadel (5.3%) held more than 5% each.
against this

The stakes are being sold down on schedule: Widmar adopted a Rule 10b5-1 plan in November 2025 to sell 39,951 shares by 10 November 2026 and Bradley one for 19,646 by 30 September 2026 (S01, Item 9B; and C16).

current state · 2026-03-19written 2026-10-04
C45facthigh confidence
The 2025 annual bonus pool was funded only if adjusted net operating income, operating income before production start-up and underutilisation, exceeded $1.0B; it came in at $1.8B. Pay then rode on cost per watt produced (30%), copper-replacement R&D progress (30%), US-made volume sold (20%), year-end net cash (10%) and safety (10%), paying 1.15 times target, with 40% of each award on individual performance.
against this

The threshold is struck after the Section 45X credit, which reduces cost of sales (C03), and 2025 operating income before the credit was about nil (C05), so the gate that funds the pool was met by the credit rather than by the manufacturing margin.

historical · FY2025written 2026-10-04
C46facthigh confidence
The 2025 performance units, earned over a period ending December 2027, pay half on module sales backlog for delivery after 2027 and half on perovskite technology milestones; that year's equity mix leaned toward time-vested restricted units because, the committee said, regulatory change made goals hard to set. Neither plan pays on shareholder return or return on capital.
against this

The committee says 2026 awards are weighted more heavily to performance units again, and restricted units do tie realised pay to the share price; shareholder return appears in the proxy only in the required pay-versus-performance table, not as an award metric.

historical · 2025 grantswritten 2026-10-04
02

Why it wins, and who could beat it

Why customers come back

Its protection is two rules, not the panel. One is the credit. The other arrives on 4 December: imported modules must sell for at least $0.38 a watt and cells $0.22, plus 15%. That covers First Solar's panel too, but binds only imports, and about 70% of its volume is made in America with no cell step. American silicon assemblers have 69.9 GW of module capacity on about 3 GW of cells, so nearly all pay.

Its two US-listed silicon rivals are each worth under $1B, and both lose money.

The evidence under this - 7 claims
C02management claimhigh confidence
The company expects a Section 45X credit of roughly 17 cents per watt on each US-made module sold to a third party.
against this

The rate is the company's own reading of its eligibility as a vertically integrated producer, and the 10-K says the credit phases down from 2030 and that modifications to the law or its interpretation could change it.

forwardwritten 2026-09-15
C08facthigh confidence
From 4 December 2026 imported solar modules face a minimum price of $0.38 per watt and cells $0.22, plus a 15% tariff.
forwardwritten 2026-09-15
C09facthigh confidence
The floor covers thin-film modules too: the annex lists the 'other than crystalline silicon' line at the same $0.38.
forwardwritten 2026-09-15
C35estimatemedium confidence
A minimum import price binds imports only. First Solar makes roughly 70% of its 2026 volume inside the United States.
derivedThe 2026 guidance assumes $2.10B-$2.19B of Section 45X credits at the company's stated rate of about 17 cents per watt, which implies 12.35-12.88 GW of US-made modules sold to third parties against guided total volume of 17.0-18.2 GW. The Q3 cadence given on the Q2 call is 3.2-3.7 GW from US operations out of 3.9-4.5 GW.
against this
  • The credit is earned on modules produced in the US and sold to a third party, which is not identical to modules sold in the US, so the ratio is an approximation of domestic production rather than of domestic sale.
  • The balance is imported and now faces the same floor as anyone else's imports. Its contracted book is struck near $0.30 a watt, below the $0.38 floor, so that volume depends on the exemption for contracts fixed before 6 August 2026.
forwardwritten 2026-09-15
C37inferencemedium confidence
First Solar's process has no cell step and no polysilicon, so its US production imports nothing the measure covers.
derivedThe 10-K describes one continuous line taking a sheet of glass to a finished module, with cadmium telluride deposited as a film rather than cells being made and assembled (C01). The measure covers polysilicon, ingots and wafers, cells and modules (C08, C09). None of those is an input to a US-made First Solar module.
against this
  • It does import semi-finished product from Malaysia and Vietnam to feed the South Carolina finishing line, and how that is classified on entry is not established here - if it enters as a module it is covered.
  • Glass, steel, copper and the tellurium itself are imported and face other tariff regimes; the claim is only that the polysilicon measure misses them.
  • The 15% tariff and the floor apply at entry, so a rival assembling in the US from imported cells pays on the cell, not the module - a smaller base than the module floor implies.
forwardwritten 2026-09-15
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C32factmedium confidence
Its two US-listed silicon rivals are worth less than $1B each and both lose money.
current state · 2026-08-25written 2026-09-15

The threat

Qcells has built the first American factory that makes every step of a silicon panel on one site, adding 3.3 GW of cells, so it collects the same credits and pays no cell floor.

Cheaper imported silicon is the other threat; First Solar is pursuing patent suits against those rivals in court.

The evidence under this - 3 claims
C20factmedium confidence
Qcells has built the first US factory making ingots, wafers, cells and modules on one site, adding 3.3 GW of cell capacity.
against this
  • Qcells' total US output is put at 8.6 GW against the roughly 17 GW of US module capacity First Solar projects for 2027, so the rival position is real and roughly half the size.
  • The claim rests on the rival's own announcement, which is the party with most reason to overstate readiness; nameplate capacity is not shipped volume.
current state · 2026-08-25written 2026-09-15
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C26facthigh confidence
First Solar asked the trade commission in February 2026 to bar imports of competing panels made by ten named rivals, and on 15 September 2026 said it would withdraw that complaint without prejudice and pursue its district-court patent suits instead.
  1. 2026-09-24editedThe company said on 15 September 2026 that it will withdraw its Section 337 complaint without prejudice and move to terminate the investigation, keeping its district-court TOPCon suits. The claim described a petition still being pursued; it now records both the petition and the withdrawal. The Federal Register shows no termination yet, so the next 10-Q is where it is confirmed.
current state · 2026-09-15written 2026-09-15

The rivals: JKS, CSIQ, and 1 more that is private, in-house or listed abroad.

Who else does this, rival by rival
TickerRivalHow far alongPriced against it
Qcells (Hanwha Solutions), Cartersville, GeorgiaPriced against it: Listed abroadOn saleListed abroad
JKS · CSIQJinkoSolar and Canadian SolarPriced against it: Same productOn saleSame product
03

How big it could get

How big this could get

Developers plan 43.4 GW of large American solar farms in 2026, up 60%. At the $0.38 floor that is about $16.5B of panels a year. First Solar already has 45.1 GW contracted through 2030, roughly one year of the whole market.

The evidence under this - 3 claims
C21facthigh confidence
Developers plan to add 43.4 GW of utility-scale solar in the United States in 2026, a 60% increase.
forwardwritten 2026-09-15
C08facthigh confidence
From 4 December 2026 imported solar modules face a minimum price of $0.38 per watt and cells $0.22, plus a 15% tariff.
forwardwritten 2026-09-15
C11facthigh confidence
The contracted backlog was 44.0 GW at September 30, 2026 for an aggregate transaction price of $13.5B, about $0.31 a watt, recognised through 2030, down from 45.1 GW for $13.6B at June 30.
  1. 2026-10-29editedthe Q3 release reports 44.0 GW; the dollar value and price per watt wait for the 10-Q
  2. 2026-11-11editedthe Q3 10-Q gives the backlog's value, $13.5B for 44.0 GW, which the October release did not
current state · 2026-09-30written 2026-09-15

Who has signed

Contracted orders total 44.0 GW worth $13.5B, delivering through 2030 at about $0.31 a watt; Silicon Ranch and NextEra Energy each took a tenth of 2025 sales. None of it captures the new floor: the contracts adjust for tariffs only within a limit the company does not state.

The evidence under this - 3 claims
C11facthigh confidence
The contracted backlog was 44.0 GW at September 30, 2026 for an aggregate transaction price of $13.5B, about $0.31 a watt, recognised through 2030, down from 45.1 GW for $13.6B at June 30.
  1. 2026-10-29editedthe Q3 release reports 44.0 GW; the dollar value and price per watt wait for the 10-Q
  2. 2026-11-11editedthe Q3 10-Q gives the backlog's value, $13.5B for 44.0 GW, which the October release did not
current state · 2026-09-30written 2026-09-15
C19facthigh confidence
Silicon Ranch and NextEra Energy each took 10% or more of 2025 sales.
current state · 2025-12-31written 2026-09-15
C12facthigh confidence
That price excludes adjusters, among them changes to certain tariff structures within a defined threshold.
current state · 2026-06-30written 2026-09-15
04

The results

The numbers

Is it growing?Shrinking

Sales fell 0.7% over the twelve months to Sep 2026, slower than the 24% in the year to Dec 2025.

Dec 2021: $2.92BDec 2022: $2.62BDec 2023: $3.32BDec 2024: $4.21BDec 2025: $5.22B12 mo to Sep 2026: $5.01B$2.92B$5.01B
Is it profitable?Strong

Operating margin is 38%, up from 26% in the year to Dec 2023.

Dec 2021: 20%Dec 2022: −1.0%Dec 2023: 26%Dec 2024: 33%Dec 2025: 31%12 mo to Sep 2026: 38%20%38%
Does growth pay off?Pays off

Running costs take 12¢ of each sales dollar, up from 5¢ in the year to Dec 2021.

Dec 2021: 5¢Dec 2022: 4¢Dec 2023: 13¢Dec 2024: 11¢Dec 2025: 10¢12 mo to Sep 2026: 12¢5¢12¢
Does it earn good returns?Decent

Each dollar put into the business earns 18% a year after tax, up from 8.0% in the year to Dec 2021.

Dec 2021: 8.0%Dec 2023: 12%Dec 2024: 16%Dec 2025: 17%12 mo to Sep 2026: 18%8.0%18%
Is it making cash?Thin

It turned 4.6% of sales into free cash, $233M over the last twelve months.

Dec 2021: −$303MDec 2022: −$30.2MDec 2023: −$785MDec 2024: −$308MDec 2025: $1.19B12 mo to Sep 2026: $233M−$303M$233M
Is the balance sheet safe?Net cash

It holds $1.62B of cash against $108M of debt, $1.52B more cash than it owes.

Dec 2021: $1.59BDec 2022: $2.39BDec 2023: $1.54BDec 2024: $1.18BDec 2025: $2.36B12 mo to Sep 2026: $1.52B$1.59B$1.52B
Are shares being diluted?Stable

The share count rose 0.7% in 4.8 years, and pay in shares costs 0.5% of sales.

Dec 2021: 107MDec 2022: 107MDec 2023: 107MDec 2024: 108MDec 2025: 108M12 mo to Sep 2026: 108M107M108M
All the numbers

Filed figures from the company's 10-K and 10-Q, read 2026-11-10. Fiscal years end in December.

Dec 2021Dec 2022Dec 2023Dec 2024Dec 202512 mo to Sep 2026
Size and growth
Sales$2.92B$2.62B$3.32B$4.21B$5.22B$5.01B
Sales growth+7.8%−10%+27%+27%+24%−0.7%
Profit
Gross margin25%2.7%39%44%41%49%
Operating margin20%−1.0%26%33%31%38%
Net margin16%−1.7%25%31%29%37%
What growth costs, per sales dollar
Research and development3.4%4.3%4.6%4.5%4.5%5.5%
Selling, general and admin5.8%6.3%6.0%4.5%3.9%4.7%
Pay in shares0.7%1.1%1.0%0.7%0.4%0.5%
Capital spending18%34%42%36%17%12%
Cash
Cash from operations$238M$873M$602M$1.22B$2.06B$839M
Free cash flow−$303M−$30.2M−$785M−$308M$1.19B$233M
Balance sheet
Cash and short-term investments$1.83B$2.58B$2.10B$1.79B$2.86B$1.62B
Debt$240M$184M$560M$610M$499M$108M
Net cash$1.59B$2.39B$1.54B$1.18B$2.36B$1.52B
Returns and shares
Return on capital8.0%-12%16%17%18%
Shares (diluted)107M107M107M108M108M108M

What growth costs

Growing has made First Solar's running costs a smaller slice of each sale, since office and sales costs barely rose as revenue nearly doubled. The real cost of growth is factories: for three years it spent more than a third of its sales building them. And its reported profit comes from the credit, not from this.

The evidence under this - 3 claims
C41facthigh confidence
Per sales dollar, selling, general and administrative cost fell from 8.2 cents in 2020 to 3.9 in 2025 while research and development rose from 3.5 to 4.5 cents, so the two together fell from 11.7 to 8.4; stock pay fell from 1.1 cents to 0.4. Sales rose 1.9 times over the five years.
against this
  • Over the four quarters to June 2026 the two lines rose back to 9.0 cents, research and development reaching 5.0, so the fall is not one way.
  • Production start-up is a separate income-statement line of 1.5 to 2.8 cents a year that the numbers card does not carry; it is what each new factory costs before it ships, and it never fell below 1.5 cents.
historical · FY2020 to FY2025, and the four quarters to June 2026written 2026-10-04
C42facthigh confidence
Capital spending ran at 34.5, 41.8 and 36.3 cents per sales dollar in 2022, 2023 and 2024, then 16.7 in 2025 and 12.2 over the four quarters to June 2026; it totalled $5.23B from 2021 to 2025 against $4.99B of operating cash flow, which includes the proceeds of selling Section 45X credits.
against this

The 2025 fall reflects factories finished rather than a cheaper growth model: the second South Carolina phase is still to be built, so the ratio rises again with the next capacity step.

historical · FY2021 to the four quarters to June 2026written 2026-10-04
C05estimatemedium confidence
The 2025 credit was slightly larger than all of operating income: before it the company lost $3.1M.
derivedGross profit of $2.120B (40.62% of $5.219B revenue) less the $1.6B credit in C03 leaves $520.3M; operating costs were $523.5M, being gross profit less the $1.597B operating income in C04.
against this
  • The credit is disclosed as $1.6B to one decimal place, so the residual is inside its own rounding: anything from $1.55B to $1.65B leaves operating income between +$47M and -$53M. The sign is uncertain; the order of magnitude is not.
  • Operating costs in 2025 included production start-up expense for Louisiana, which a steady-state business would not carry.
historical · FY2025written 2026-09-15

The last report (29 Oct)

Quarter ended September 2026, filed figures.

-23%
Sales growth
against the same quarter a year earlier
16.5%
Growth from the quarter before
against 1.1% the quarter before that
44¢
Operating profit
per dollar of sales

The third quarter brought sales of $1.23B, down 23% on a year, and profit of $571M, or $5.30 a share, up 25%. Profit came from credits that cover about 17 cents of every watt it makes in America, which is why selling less still paid more.

Contracted orders fell for a fourth quarter, to 44.0 GW from 45.1 GW. The company cut its 2026 sales forecast to $4.6B to $4.9B and held its profit forecast. It also announced its first share buyback, up to $500M. The price of new orders, our test, was not stated.

What it did to the thesisIt left the question we asked unanswered: the release does not say what new orders paid, so nothing in the case moved. Shrinking orders and cut volume weigh on the near term, but the forecast for profit held, and so does the argument.

The quarter in detail

Sales growth from the previous quarter, filed figures.

Earnings per share $5.30 against $4.67 expected.

The evidence under this - 5 claims
C06management claimhigh confidence
2026 guidance is now gross profit of $2.4B to $2.6B (unchanged) assuming $2.07B to $2.19B of Section 45X credits, operating expenses of $610M to $635M (unchanged), net sales cut to $4.6B to $4.9B from $4.9B to $5.2B, volume sold cut to 16.0 GW to 17.0 GW from 17.0 GW to 18.2 GW, and adjusted EBITDA of $2.55B to $2.75B from $2.6B to $2.8B.
against this
  • The guidance was set on 2026-02-24 and reaffirmed on 2026-07-30, both before the August proclamation, so it prices no part of the measure.
  • A securities class action covering the period to 2026-02-24 alleges the company understated how far its tariff response would hurt 2026.
  1. 2026-10-29editedthe September-quarter release cut volume, sales and adjusted EBITDA guidance while holding gross profit and operating expenses
forwardwritten 2026-09-15
C11facthigh confidence
The contracted backlog was 44.0 GW at September 30, 2026 for an aggregate transaction price of $13.5B, about $0.31 a watt, recognised through 2030, down from 45.1 GW for $13.6B at June 30.
  1. 2026-10-29editedthe Q3 release reports 44.0 GW; the dollar value and price per watt wait for the 10-Q
  2. 2026-11-11editedthe Q3 10-Q gives the backlog's value, $13.5B for 44.0 GW, which the October release did not
current state · 2026-09-30written 2026-09-15
C13facthigh confidence
Backlog fell four quarters running, from 50.1 GW to 47.9 GW to 45.1 GW to 44.0 GW.
  1. 2026-10-29editedthe September 30 figure of 44.0 GW is a fourth straight fall
historical · December 2025 to June 2026written 2026-09-15
C49facthigh confidence
First Solar's board authorised a $500M share repurchase program on October 29, 2026, with no expiry; the company had never bought back stock (C23). Net cash fell to $1.5B at September 30 from $2.4B at the year end, on working capital and US factory spending, and long-term government grants receivable (Section 45X credits not yet sold or received) rose to $1.59B from $0.13B.
current state · 2026-10-29written 2026-10-29
C10management claimmedium confidence
Since the July call the company booked about 1.4 GW in the United States at about $0.36 per watt, inclusive of technology adjusters, and about 4.9 GW in the US so far in 2026 at about $0.35 to $0.36.
against this
  • The figure is inclusive of technology adjusters that have not yet been earned, so it is a contracted maximum rather than cash received.
  • Management said the volume was contracted or in late-stage negotiation, with price and terms largely set, before the Section 232 floor was announced, so it does not yet show what the floor does to price.
  • Volume fell from 1.9 GW the quarter before, and about 4 GW of optional volume in existing contracts still sits at older prices.
  1. 2026-10-29editedrefresh date moved: the October 29 release states no new bookings or price; the call, not yet on the vendor, is owed
  2. 2026-11-11editedre-sourced from the Q3 2026 call: 1.4 GW of US bookings at about $0.36 a watt since July, replacing the Q2 figure
historical · Q3 2026written 2026-09-15
05

The price, and where we disagree

In short. Fair. Counting the tax credit still to come, the price pays a little more than our central case and well short of the import floor's best case.

Share price
$206.40
close,
Price for each $1 of yearly profit
12.7
its 6-year normal 19.0
Company value for each $1 of cash profit
8.5
its 6-year normal 13.0

The shares hit $269.70 the day after the floor was announced in August, then gave the whole move back within three weeks. The company's forecast predates the rule, and the October report did not say what new orders were fetching. The first sign comes with the call and the quarterly filing: the price of new orders.

The evidence under this - 4 claims
C15facthigh confidence
The shares reached $269.70 during 7 August, then gave the whole gain back to close at $206.32 on 25 August.
historical · July to August 2026written 2026-09-15
C06management claimhigh confidence
2026 guidance is now gross profit of $2.4B to $2.6B (unchanged) assuming $2.07B to $2.19B of Section 45X credits, operating expenses of $610M to $635M (unchanged), net sales cut to $4.6B to $4.9B from $4.9B to $5.2B, volume sold cut to 16.0 GW to 17.0 GW from 17.0 GW to 18.2 GW, and adjusted EBITDA of $2.55B to $2.75B from $2.6B to $2.8B.
against this
  • The guidance was set on 2026-02-24 and reaffirmed on 2026-07-30, both before the August proclamation, so it prices no part of the measure.
  • A securities class action covering the period to 2026-02-24 alleges the company understated how far its tariff response would hurt 2026.
  1. 2026-10-29editedthe September-quarter release cut volume, sales and adjusted EBITDA guidance while holding gross profit and operating expenses
forwardwritten 2026-09-15
C08facthigh confidence
From 4 December 2026 imported solar modules face a minimum price of $0.38 per watt and cells $0.22, plus a 15% tariff.
forwardwritten 2026-09-15
C10management claimmedium confidence
Since the July call the company booked about 1.4 GW in the United States at about $0.36 per watt, inclusive of technology adjusters, and about 4.9 GW in the US so far in 2026 at about $0.35 to $0.36.
against this
  • The figure is inclusive of technology adjusters that have not yet been earned, so it is a contracted maximum rather than cash received.
  • Management said the volume was contracted or in late-stage negotiation, with price and terms largely set, before the Section 232 floor was announced, so it does not yet show what the floor does to price.
  • Volume fell from 1.9 GW the quarter before, and about 4 GW of optional volume in existing contracts still sits at older prices.
  1. 2026-10-29editedrefresh date moved: the October 29 release states no new bookings or price; the call, not yet on the vendor, is owed
  2. 2026-11-11editedre-sourced from the Q3 2026 call: 1.4 GW of US bookings at about $0.36 a watt since July, replacing the Q2 figure
historical · Q3 2026written 2026-09-15
How we value it
Method
Cash the business will produce, in three stages
Required return
8%
Long-run growth
4%
Moat
An average moat
BearBaseBull
Sales growth, years 1 to 3-9%, -5%, 0%-9%, 5%, 5%-9%, 12%, 10%
Normal operating margin3%10%18%
Return on new capital after year ten8%12%15%
Value a share$95.95$139$215
Price
$162
Priced on
16 Nov
Price as a share of the base value
117%
Yearly return the price offers on the base case
6.9%
Risk check: valued at 10%, the rate for a cyclical or indebted business
$118

Where each figure comes from

FigureSourceWhere exactly
Sales, the last full yearFiled reportAnnual report (10-K) for the period to 31 Dec 2025, filed 24 Feb 2026; the full year.
Operating margin todayFiled reportRead by hand from the filed statements; the period it was read from was not recorded.
Tax rateFiled reportQuarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the four quarters to that date.
Sales per dollar of new capitalFiled reportAnnual report (10-K) for the period to 31 Dec 2025, filed 24 Feb 2026: the year's sales over its operating capital.
SharesFiled reportQuarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the diluted share count.
Cash and investmentsFiled reportQuarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the balance sheet on that date.
DebtFiled reportQuarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the balance sheet on that date.
Sales growth, years 1 to 3Our judgementOur reading of the evidence, set against the company's guidance; the base case below says where it follows it and where it departs. FMP's consensus of 15 to 30 analysts read 9.1%, 14.5%, 11.6%.
Normal operating marginOur judgementOur reading of the evidence; each case below names it.
Return on new capital after year tenOur method's ruleSet by our reading of the moat: the 8% rate with none, 12% with an average one, 15% with a strong one; the bear case earns the rate.
The bear and bull casesOur judgementOur reading of the evidence; each case below names it.
PriceMarket priceFMP's closing price on 16 Nov 2026.
Other investmentsFiled reportRead by hand from the filed statements; the period it was read from was not recorded.

What each case rests on

Bear caseThe import floor does not bind and new bookings settle near today's backlog price: sales fall 5% then hold flat, the margin before the credit reaches only 3%, about 2025's break-even before it (C05), and new capital earns only its 9% cost.

Base caseThe module business is valued without the Section 45X credit, which is carried beside it at the roughly $8B still to come (C28): 2026 sales fall about 9% on the guide, then grow 5% a year as the 44.0 GW backlog at about $0.30 a watt runs through 2030 (C11); the operating margin before the credit rises from about minus 3% (C07) to 10% as new bookings near $0.36 a watt (C10) replace old contracts.

Bull caseThe floor lands on rivals' costs (C30) and bookings rise well above $0.36 a watt: sales grow 12% and 10% as factories fill, and the margin before the credit reaches 18%.

What moves the base value

LowerHigher
Growth in years 1 to 3, a quarter lower or higher$140$138
The normal margin, 15% lower or higher$129$149
Capital per dollar of growth, a quarter more or less$135$141
Return on new capital after year ten, at the rate or one step up$126$144
The rate, a point higher or lower$126$160
Long-run growth a point lower$134-

Value a share as growth and margin change

Normal margin ↓ · growth →50%75%100%125%150%
8%$128$127$126$125$124
9%$134$133$132$131$130
10%$141$140$139$138$137
11%$147$146$145$145$144
12%$153$153$152$151$151

Growth is shown as a share of the base case's. Green: the price is at or under 80% of that value. Red: at or over 120%.

How we read the multiple

Buyers pay about 13 times a year's profit, against about 19 usually. But most of that profit is a tax credit that shrinks from 2030 and ends in 2032. Counting cash and debt, the company costs about $16B:

  1. About $8B of that pays for the tax credit still to come.
  2. That leaves about $8B for the factories, forecast to lose money this year without the credit.
  3. Valued with margins before the credit rising to 10 cents a dollar, the shares are worth about $139, and $215 at best.

Against its own history. Cheaper than its own usual level for each dollar of profit, and that profit is mostly the credit the market expects to end.

The evidence under this - 6 claims
C28estimatelow confidence
Roughly $8B of the $20.68B enterprise value is the credit stream still to come.
derivedThe guided annual credit of about $2.15B (C06) held flat to 2029, then phased at 75%, 50% and 25% through 2032 as the statute provides (C22), discounted at 10%. Half of 2026 is already earned. The chain totals about $10.7B before discounting and $8.4B after.
against this
  • The discount rate and the flat-to-2029 profile are both mine. At 8% the figure is nearer $9B and at 14% nearer $7.3B, so the estimate carries a spread of about a fifth of itself either way.
  • The credits are sold to third parties at a discount the company does not disclose, so the cash is less than the credit.
  • Congress cut other energy credits in July 2025 and could reach this one before 2032, which would make the estimate an upper bound.
forwardwritten 2026-09-15
C34estimatelow confidence
The price pays roughly $8B for the credit still to come and about $8B for a manufacturer guided to lose money before it.
derivedEnterprise value of about $16.0B at the 16 November close (C27), less the roughly $8B remaining credit stream (C28), leaves about $8B set against operating income before the credit of -$110M to -$425M on 2026 guidance (C07).
against this
  • The $8B rests on my own discount rate and a flat-to-2029 credit profile; at 8% it is nearer $9B and at 14% nearer $7.3B, which moves the residual by about a fifth either way.
  • The residual is not paying for nothing. It buys 45.1 GW contracted through 2030, $1.688B of net cash, six US plants and a technology no rival runs - the loss before the credit is struck after start-up and idle costs a steady-state business would not carry.
  1. 2026-11-17editedRe-struck at the 16 November close: enterprise value fell from $20.68B to about $16.0B, so the residual after the credit is about $8B rather than $12B.
current state · 2026-11-16written 2026-09-15
C07estimatemedium confidence
At every point of 2026 guidance, operating income before the credit is negative, between -$80M and -$425M.
derivedEach end of the guided range in C06 taken with its own credit assumption: $2.4B gross profit less $2.19B of credits less $635M of costs is -$425M; $2.6B less $2.10B less $610M is -$110M.
against this
  • Guided operating costs include $90M to $100M of production start-up expense for South Carolina, and guided gross profit is struck after $115M to $135M of costs for factories deliberately held idle. Adding both back leaves roughly break-even rather than a loss, so this is a company investing through the line rather than one that cannot make money.
  • The pairing of the worst gross profit with the largest credit is arithmetic on the stated ranges, not a disclosed combination; management does not say the two ends move together.
  1. 2026-10-29editedthe credit assumption narrowed to $2.07B to $2.19B from $2.10B to $2.19B in the October 29 release, lifting the top of the range
forwardwritten 2026-09-15
C22facthigh confidence
The Section 45X credit runs to 2032 and begins phasing down in 2030.
forwardwritten 2026-09-15
C27facthigh confidence
Enterprise value is about 6.0 times the midpoint of guided 2026 adjusted earnings before interest, tax and depreciation at the 16 November close of $162.45, down from 7.4 to 8.0 times in late August, and the shares cost $12.69 for each $1 of trailing yearly profit at the August pack against a six-year usual level of $19.05.
  1. 2026-11-17editedRe-struck at the 16 November close: enterprise value of about $16.0B (107.7M shares at $162.45, $108M of debt, $1.62B of cash and securities) over the $2.65B guided midpoint.
current state · 2026-11-16written 2026-09-15
C50inferencemedium confidence
On a three-stage fair value at an 8% required return First Solar is worth about $139 a share against $162.45 at the 16 November close, where enterprise value is about 6.0 times guided 2026 adjusted EBITDA. The module business is valued before the Section 45X credit, and the roughly $8B of credit still to come is added beside it. Base case: 2025 sales of $5.22B fall 9% in 2026, then grow 5% a year, slowing toward 4% by year ten, each year closing 30% of the gap; the operating margin before the credit moves from about minus 3% to 10% over five years; each added dollar of sales needs about $1.30 of capital, the average the filings show; a 15% tax rate; then 4% growth for ever at a 12% return on new capital. Bear (a 5% fall then flat, a 3% margin, new capital earning 8%) gives about $95, bull (12% and 10% growth, an 18% margin) about $215. As a risk check, at the 10% return asked of a cyclical or indebted business the base case gives about $118.
against this

The credit's value rests on the thesis's rough $8B (C28); every $1B more adds about $9 a share. If the import floor lifts bookings well above $0.36 a watt the value nears the bull case's $215, about a third above the price.

  1. 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
  2. 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
  3. 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
  4. 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten
  5. 2026-11-17editedRe-priced at the 16 November close of $162.45 after a 19% fall with no company news; the value itself is unchanged at about $139, and the price is now about 117% of it.
current state · 2026-11-16written 2026-11-09
Priced against its rivals

There is no group to price it against. JinkoSolar and Canadian Solar sell the same panel to the same buyers, but the vendor cannot reconcile JinkoSolar's income statement, and one rival alone is not a group. The price is judged against its own history instead.

Where we disagree with the market

The market believes

A subsidised commodity maker whose earnings expire with the credit, so the tariff win is a trade; 30 of 74 analysts say hold or sell.

We believe

The floor lifts what rivals must charge. First Solar makes most of its panels here, so the rule lands on their costs, not its own.

The easy reading was that thin-film panels are exempt. They are not, and once that proved false the real mechanism was easy to miss: a floor that binds imports, in a country whose silicon assemblers import almost every cell.

The rule is also softer than it looks. An importer that documents its price pays only the gap to the floor, so a rival importing at $0.32 pays $0.06, not $0.38, and contracts signed before 6 August are exempt.

What analysts say, and the evidence
The evidence under this - 6 claims
C27facthigh confidence
Enterprise value is about 6.0 times the midpoint of guided 2026 adjusted earnings before interest, tax and depreciation at the 16 November close of $162.45, down from 7.4 to 8.0 times in late August, and the shares cost $12.69 for each $1 of trailing yearly profit at the August pack against a six-year usual level of $19.05.
  1. 2026-11-17editedRe-struck at the 16 November close: enterprise value of about $16.0B (107.7M shares at $162.45, $108M of debt, $1.62B of cash and securities) over the $2.65B guided midpoint.
current state · 2026-11-16written 2026-09-15
C09facthigh confidence
The floor covers thin-film modules too: the annex lists the 'other than crystalline silicon' line at the same $0.38.
forwardwritten 2026-09-15
C35estimatemedium confidence
A minimum import price binds imports only. First Solar makes roughly 70% of its 2026 volume inside the United States.
derivedThe 2026 guidance assumes $2.10B-$2.19B of Section 45X credits at the company's stated rate of about 17 cents per watt, which implies 12.35-12.88 GW of US-made modules sold to third parties against guided total volume of 17.0-18.2 GW. The Q3 cadence given on the Q2 call is 3.2-3.7 GW from US operations out of 3.9-4.5 GW.
against this
  • The credit is earned on modules produced in the US and sold to a third party, which is not identical to modules sold in the US, so the ratio is an approximation of domestic production rather than of domestic sale.
  • The balance is imported and now faces the same floor as anyone else's imports. Its contracted book is struck near $0.30 a watt, below the $0.38 floor, so that volume depends on the exemption for contracts fixed before 6 August 2026.
forwardwritten 2026-09-15
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C39facthigh confidence
The minimum import price is a top-up rather than a floor price, and the proclamation says so. An importer that documents at entry pays 'a specific tariff equal to the difference between the entered value on the entry summary and the MIP'; an importer that submits no documentation pays 'a specific tariff equal to the applicable MIP' in full. The certification is of one of two things: that the first arm's-length US sale will be at or above the minimum price, or that the sale is on fixed terms in a contract entered into before the proclamation was signed. A certification CBP finds materially inaccurate permanently prohibits that importer and its affiliates from importing polysilicon or its derivatives.
against this
  • The top-up structure makes the measure weaker per module than a flat $0.38 would be: a rival importing at $0.32 a watt pays $0.06, not $0.38. It also makes the burden proportional to how cheap the import is, which is the design - the cheapest importer pays the most - so the competitive effect is preserved even though the headline overstates the tariff on any given entry.
  • The permanent-prohibition penalty cuts the other way and is unusually severe for a customs certification. Its practical effect is to make the pre-August contract exemption expensive to use aggressively, since the downside of getting it wrong is exclusion from the US market rather than a duty bill - which argues that fewer rivals will lean on the exemption than its existence implies.
current state · 2026-08-11written 2026-09-15
C38factlow confidence
Importers may use the price fixed in a contract entered into before 6 August 2026 rather than the floor.
against this
  • This rests on law-firm readings of the proclamation rather than on the customs guidance that will actually administer it, and the documentation an importer must produce is not established here.
  • It cuts both ways: it protects First Solar's own imported volume against a floor above its contracted price, and it equally protects every rival's pre-existing contracts, so it delays the benefit as well as the cost.
forwardwritten 2026-09-15
06

What could go wrong, and how we would know

The risks

  • The floor may not bite. Importers that document their price pay only the gap to the floor, and contracts signed before 6 August are exempt.
  • The credit can be cut. Profit rests on a credit that begins phasing down in 2030 and ends in 2032; Congress could cut it sooner.
  • Cells made at home. Qcells added 3.3 GW of American cells. Enough more, and rivals stop paying the cell floor.
  • Shrinking contracted orders. They have fallen three quarters running, from 50.1 GW to 45.1 GW.
  • A shareholder lawsuit. It alleges the company understated what tariffs would cost it in 2026.
The evidence under this - 8 claims
C39facthigh confidence
The minimum import price is a top-up rather than a floor price, and the proclamation says so. An importer that documents at entry pays 'a specific tariff equal to the difference between the entered value on the entry summary and the MIP'; an importer that submits no documentation pays 'a specific tariff equal to the applicable MIP' in full. The certification is of one of two things: that the first arm's-length US sale will be at or above the minimum price, or that the sale is on fixed terms in a contract entered into before the proclamation was signed. A certification CBP finds materially inaccurate permanently prohibits that importer and its affiliates from importing polysilicon or its derivatives.
against this
  • The top-up structure makes the measure weaker per module than a flat $0.38 would be: a rival importing at $0.32 a watt pays $0.06, not $0.38. It also makes the burden proportional to how cheap the import is, which is the design - the cheapest importer pays the most - so the competitive effect is preserved even though the headline overstates the tariff on any given entry.
  • The permanent-prohibition penalty cuts the other way and is unusually severe for a customs certification. Its practical effect is to make the pre-August contract exemption expensive to use aggressively, since the downside of getting it wrong is exclusion from the US market rather than a duty bill - which argues that fewer rivals will lean on the exemption than its existence implies.
current state · 2026-08-11written 2026-09-15
C38factlow confidence
Importers may use the price fixed in a contract entered into before 6 August 2026 rather than the floor.
against this
  • This rests on law-firm readings of the proclamation rather than on the customs guidance that will actually administer it, and the documentation an importer must produce is not established here.
  • It cuts both ways: it protects First Solar's own imported volume against a floor above its contracted price, and it equally protects every rival's pre-existing contracts, so it delays the benefit as well as the cost.
forwardwritten 2026-09-15
C22facthigh confidence
The Section 45X credit runs to 2032 and begins phasing down in 2030.
forwardwritten 2026-09-15
C05estimatemedium confidence
The 2025 credit was slightly larger than all of operating income: before it the company lost $3.1M.
derivedGross profit of $2.120B (40.62% of $5.219B revenue) less the $1.6B credit in C03 leaves $520.3M; operating costs were $523.5M, being gross profit less the $1.597B operating income in C04.
against this
  • The credit is disclosed as $1.6B to one decimal place, so the residual is inside its own rounding: anything from $1.55B to $1.65B leaves operating income between +$47M and -$53M. The sign is uncertain; the order of magnitude is not.
  • Operating costs in 2025 included production start-up expense for Louisiana, which a steady-state business would not carry.
historical · FY2025written 2026-09-15
C20factmedium confidence
Qcells has built the first US factory making ingots, wafers, cells and modules on one site, adding 3.3 GW of cell capacity.
against this
  • Qcells' total US output is put at 8.6 GW against the roughly 17 GW of US module capacity First Solar projects for 2027, so the rival position is real and roughly half the size.
  • The claim rests on the rival's own announcement, which is the party with most reason to overstate readiness; nameplate capacity is not shipped volume.
current state · 2026-08-25written 2026-09-15
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C13facthigh confidence
Backlog fell four quarters running, from 50.1 GW to 47.9 GW to 45.1 GW to 44.0 GW.
  1. 2026-10-29editedthe September 30 figure of 44.0 GW is a fourth straight fall
historical · December 2025 to June 2026written 2026-09-15
C17facthigh confidence
A securities class action covers buyers between 26 February 2025 and 24 February 2026 and alleges the tariff impact on 2026 was understated.
current state · 2026-06-30written 2026-09-15
What we still do not know
  • How much the cell floor actually adds to a rival's cost. Nobody has published it, though customs trade data could answer it.
  • What the tariff clauses in existing contracts are worth; the company has not said.
The evidence under this - 3 claims
C36factmedium confidence
American module capacity is 69.9 GW against about 3 GW of cell capacity, so nearly every US-assembled silicon module uses an imported cell.
against this
  • Cell capacity is being built: Qcells alone is adding 3.3 GW at Cartersville, which roughly doubles the national figure, and the trade body's own dashboard would not open to an automated fetch so this rests on its figures as reported.
  • Capacity is not utilisation. A module maker holding cell inventory or sourcing from an allied country pays the floor later rather than never.
current state · 2026-06-30written 2026-09-15
C39facthigh confidence
The minimum import price is a top-up rather than a floor price, and the proclamation says so. An importer that documents at entry pays 'a specific tariff equal to the difference between the entered value on the entry summary and the MIP'; an importer that submits no documentation pays 'a specific tariff equal to the applicable MIP' in full. The certification is of one of two things: that the first arm's-length US sale will be at or above the minimum price, or that the sale is on fixed terms in a contract entered into before the proclamation was signed. A certification CBP finds materially inaccurate permanently prohibits that importer and its affiliates from importing polysilicon or its derivatives.
against this
  • The top-up structure makes the measure weaker per module than a flat $0.38 would be: a rival importing at $0.32 a watt pays $0.06, not $0.38. It also makes the burden proportional to how cheap the import is, which is the design - the cheapest importer pays the most - so the competitive effect is preserved even though the headline overstates the tariff on any given entry.
  • The permanent-prohibition penalty cuts the other way and is unusually severe for a customs certification. Its practical effect is to make the pre-August contract exemption expensive to use aggressively, since the downside of getting it wrong is exclusion from the US market rather than a duty bill - which argues that fewer rivals will lean on the exemption than its existence implies.
current state · 2026-08-11written 2026-09-15
C12facthigh confidence
That price excludes adjusters, among them changes to certain tariff structures within a defined threshold.
current state · 2026-06-30written 2026-09-15

What would change our view

  1. Neutral→UnattractiveThe price rises past 120% of fair value

    It needs:

    • ○The share price rises to about $167, 20% above what we think the business is worth.about ≈$167.86 today, -19% above the last close
  2. Neutral→AttractiveThe price falls 20% below fair value

    It needs:

    • ○The share price falls to about $111, 20% below what we think the business is worth.about ≈$105.59 today, 49% below the last close
We would also be wrong if
  • New US bookings settle above $0.40 a watt once the floor is in force.
  • Congress extends the Section 45X credit beyond 2032.

What is coming

  • 4 December 2026 Minimum import price and 15% tariff take effect, with customs guidance due before it: How the floor is administered: what documentation importers owe and how widely the pre-August contract exemption is used, which decides how much of rivals' cost it reaches.
Every dated event ahead
  • 4 December 2026 Minimum import price and 15% tariff take effect, with customs guidance due before it: How the floor is administered: what documentation importers owe and how widely the pre-August contract exemption is used, which decides how much of rivals' cost it reaches.
  • February 2027 Q4 2026 results and the FY2026 10-K: 2026 delivered against the cut guide of $4.6B to $4.9B of revenue and $2.07B to $2.19B of credits, the first 2027 guidance under the floor, and the year-end backlog.
Since we wrote this
Written
25 Aug
Price the thesis read
$206.40
25 Aug
  • 17 NovPrice moveShares fell a fifth; view moved to NeutralA 19% fall since 6 November with no company news put First Solar at about 117% of our $139 value, back inside the margin, so the view moved from Unattractive to Neutral.
  • 11 NovPrice moveShares fell 10%; still above our value, view holds UnattractiveThe price is about 130% of our value, short of the 120% line that would make it Neutral. US orders still fetch about $0.36 a watt, priced before the import floor.
  • 9 NovThesis reviewUnattractive: priced above a value of about $137Re-valued in three stages at an 8% return, the method now drawn on every Attractive thesis: the price sits more than 20% above its base-case value and above the bull case.
  • 30 OctEarningsFirst Solar call: new US orders at $0.36 a watt, buyers waiting. View heldQ3 2026 results.
  • 29 OctEarningsBacklog fell again to 44.0 GW; volume guide cut, profit guide heldThe release names no booking price, the question we set; the view holds until the call or 10-Q shows it.
  • 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
  • 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
  • 2 OctNewsFirst Solar files a new TOPCon patent suit; the case rests elsewhere. View heldFirst Solar sued JA Solar and American Panel Solutions in Delaware for infringing its TOPCon cell patent.
  • 24 SepNewsTrade-commission patent case dropped; court suits continue. View heldThe import ban it sought was upside the thesis never counted, and the patent suits against its silicon rivals carry on in court.
  • 15 SepThesis reviewThe floor is a top-up, but harder to game. View heldRead in full, the rule charges a documented importer only the gap to the floor, but bars for good any importer whose paperwork proves wrong. The two roughly offset.
  • 25 AugThesis writtenAttractive, medium convictionWritten after the shares gave back the whole jump that followed August's import floor, on the reading that its thin-film panel fell outside the rule.
  • 25 AugThesis reviewThin-film exemption proved false; the case survives. View heldThe rule covers its panel too. The advantage is that the floor binds imports, and First Solar makes most panels at home with no cells, while American rivals import theirs.

Also checked 5 times with nothing new to act on.

Sources 22

Where this came from

Sources - 22

Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.

  1. S01primarycompany filing2026-02-24
    the product and process description, the Section 45X credit rate and the $1.6B recognised in cost of sales, contracted backlog at year end, customer concentration, the USITC petition and the BP Solar dispute
    backsC01C02C03C04C13C19C22C26C33C40C41C42C43C47C48
  2. S02primarycompany filing2026-02-24
    the original 2026 guidance and the footnote naming the Section 45X credits assumed inside guided gross profit
    backsC06C31
  3. S03primarycompany filing2026-07-30
    backlog and its variable-consideration carve-outs, the IEEPA refund benefit, the securities class action and derivative suit, cash and debt, share counts
    backsC01C12C13C17C18C23C24C25C26C41C42C48
  4. S04primarycompany filing2026-07-30
    reaffirmed 2026 guidance with the Section 45X assumption restated, and the Q3 cadence
    backsC06C31
  5. S05primarytranscript2026-07-30
    Q2 2026 earnings call transcript
    new bookings volume and price per watt, India pricing, the domestic-content share of backlog, idle international capacity; and eight analyst questions on Section 232 timing and waivers
    Since our last earnings call, we recorded approximately 1.9 gigawatts of additional U.S. gross bookings at an average selling price of approximately $0.36 per watt, inclusive of applicable technology adjusters. ... approximately 41 gigawatts of our 45 gigawatt backlog includes some form of domestic content requirement.
    backsC14C29
  6. S06primarygovernment data2026-08-06
    the tariff rate, the minimum import prices of $0.22 per watt for cells and $0.38 per watt for modules, and the December 4, 2026 effective date
    Imports of polysilicon ingots and polysilicon derivatives specified in Annexes I and II of this proclamation shall be subject to an additional 15 percent ad valorem rate of duty ... $0.22 per watt for solar cells; and $0.38 per watt for solar modules ... Effective with respect to goods entered for consumption, or withdrawn from warehouse for consumption, on or after 12:01 a.m. eastern time on December 4, 2026
    backsC08
  7. S07secondaryindustry data2026-08-13
    a buyer-side reading of which tariff codes the annexes cover, and median platform module pricing of 27.1 cents per watt on August 5, 2026
    First Solar is particularly well positioned, as its CdTe thin-film modules fall under none of the covered HTS codes.
    backsC30C38
  8. S08primarygovernment data2026-02-20
    the size of the US utility-scale solar market this company sells into, measured by the government rather than the industry
    Developers plan to add 43.4 GW of new utility-scale solar capacity in 2026, a 60% increase in capacity additions from last year if realized.
    backsC21
  9. S09primarycompany release2026-08-06
    management's own reading of the measure, and its projection of roughly 17 GW of US module capacity by 2027
    This action closes that loophole, and it is built to be enforced, with a minimum import price, an ad valorem tariff behind it, and real consequences for violators.
    read, but no claim rests on it
  10. S10primarycompany release2026-06-17
    a named rival's own account of building the fully domestic ingot-to-module position First Solar's advantage rests on
    by Q3 2026, Cartersville will make 3.3 GW each of ingots, wafers, cells, and 3.5 GW of modules a year.
    backsC20
  11. S11secondaryvendor data2026-08-25
    Research pack, frozen beside this thesis
    content/FSLR/research-pack.json
    price and the 20-day range, the valuation series against its own six-year medians, daily bars around the proclamation, Form 4 insider rows, the analyst split, and the two rivals' trailing multiples
    backsC04C15C16C18C25C27C32
  12. S12primarycompany filing2026-04-30
    the middle point of the backlog decline at 47.9 GW, and the second of three reaffirmations of 2026 guidance carrying the same Section 45X assumption
    backsC06C13
  13. S13primarygovernment data2026-08-06
    the actual scope of the measure: both the crystalline silicon module line and the 'other' line that carries thin film, each at the $0.38 per watt floor
    (iv) $0.38 per watt for solar modules. HTSUS Description 8541.43.0010 ... Photovoltaic cells assembled in modules or made up into panels: crystalline silicon photovoltaic cells of a kind described in statistical note 12 to chapter 85 of the HTSUS 8541.43.0080 ... Photovoltaic cells assembled in modules or made up into panels: other than provided for in statistical reporting number 8541.43.0010
    backsC09
  14. S14primarycompany filing2026-08-18
    the footnotes: three of the sales were pre-scheduled Rule 10b5-1 plans adopted in early May 2026, and the fourth disposal was tax withholding on vesting
    backsC16
  15. S15secondaryindustry data2026-06-30
    the gap that makes the cell floor bind on domestic assembly: US module capacity is more than twenty times US cell capacity, so nearly every American-assembled silicon module uses an imported cell
    there is currently 69.9 gigawatts (GW) of domestic module manufacturing capacity online ... the United States currently has just 3 gigawatts of solar cell manufacturing capacity
    backsC36
  16. S16primarygovernment data2026-08-11
    how the floor actually bites - as a top-up to entered value for importers that document, and as the whole minimum price for importers that do not - plus the penalty for a wrong certification, none of which is in the White House summary this thesis read
    (b) If an importer fails to submit the documentation referenced in subclause (a) of this clause, the imported merchandise shall be subject to a specific tariff equal to the applicable MIP. (c) For importers that submit the documentation referenced in subclause (a) of this clause, in the event that the entered value on the entry summary of the imported merchandise is less than the MIP, the imported merchandise shall be subject to a specific tariff equal to the difference between the entered value on the entry summary and the MIP. (3) ... If CBP determines that an importer's documentation was materially inaccurate or that an importer has materially failed to comply with its certification, that importer and its affiliates shall permanently be prohibited from importing polysilicon and polysilicon derivatives into the United States.
    backsC39
  17. S17primarycompany release2026-09-15
    the intent to withdraw the Section 337 complaint and move to terminate Investigation No. 337-TA-1494 without prejudice, and to continue the district-court TOPCon patent suits
    First Solar, Inc. (Nasdaq: FSLR) ("First Solar" or "the Company") today announced a recalibration of its intellectual property (IP) enforcement strategy and its intent to voluntarily withdraw its Section 337 complaint and move to terminate the investigation currently pending before the US International Trade Commission (USITC) without prejudice to refiling at a later date. The Company will continue pursuing its existing TOPCon patent lawsuits in US District Court
    backsC26
  18. S18primarycompany filing2026-04-02
    beneficial ownership as of March 19, 2026, the 2025 Bonus Plan's threshold, metrics, weightings and payout, the 2025 performance-unit metrics, director tenures and the CEO's biography
    backsC43C44C45C46
  19. S19primarycompany filing2023-02-28
    the 2020 to 2022 income statement as a percentage of net sales: selling and admin, research and development, production start-up and gains on sales of businesses
    backsC41C42C47
  20. S20primarycompany filing2026-10-29
    the September-quarter income statement and balance sheet, backlog of 44.0 GW, the cut 2026 guidance, and the $500M buyback authorisation
    backsC06C11C13C31C49C50
  21. S21primarycompany filing2026-10-29
    backlog of 44.0 GW for an aggregate transaction price of $13.5B through 2030, cash and marketable securities of $1.624B, long-term government grants receivable of $1.588B, the class-action schedule
    backsC11
  22. S22primarytranscript2026-10-29
    Q3 2026 earnings call transcript (FMP), held 29 October 2026
    Q3 US bookings of about 1.4 GW at about $0.36 a watt, 4.9 GW booked in the US this year at $0.35 to $0.36, the backlog's $13.5B value, about 4 GW of variable volume
    Since our last earnings call, we recorded approximately 1.4 gigawatts of additional U.S. gross bookings at an average selling price of approximately $0.36 per watt, inclusive of applicable technology adjusters. Importantly, substantially all the volume was contracted or in late-stage negotiation, and pricing and terms are largely set ahead of the Section 232 announced.
    backsC10
Rendered from content/FSLR/thesis.json · thesis reviewed 2026-11-17. This page explains a company; it does not say what to pay, when to buy or how much.