Phase Space AI

Valuation

First Solar [FSLR]

First Solar [FSLR] — Valuation

As of 2026-07-29 · price at memo $202.59 (2026-07-28 close) · EV $19,768m · TTM revenue $5,419.0m

Two outputs, two horizons, neither replacing the other:

---

1. Inputs, and the two corrections to the screen

---

2. The implied-path test

Solved for: the 5-year revenue CAGR the current price requires. Held fixed, explicitly: terminal operating margin, exit EV/EBIT multiple, WACC 10.0%, horizon 5 years, net cash $2,000.8m, base revenue $5,419.0m TTM, share count 107.453m.

Terminal value is 100% of the modelled EV in this construction, so the reverse DCF is mandatory as the primary long-horizon output and the exit multiple is the highest-variance parameter. Sensitivity is run over the exit multiple, never over scenario probabilities.

2.1 Anchoring the exit multiple

A growth-matched external comparator set does not exist for this business and I am declaring that rather than substituting a peer median. The subject is a US thin-film module manufacturer whose reported operating income is ~100% a statutory production credit that terminates in 2032. The candidate comparators — JinkoSolar, Trina, Canadian Solar, Maxeon and the other crystalline-silicon module makers — are loss-making or near-loss-making at the operating line, report under different accounting regimes, and have no equivalent credit; their multiples cannot bracket FSLR's growth and margin structure at the exit year. The screen's exit multiple of 23.3x, described as GROWTH_MATCHED from 353 peers, is a universe-wide artefact: 353 names is not a matched comparator set, and 23.3x sits at the 93rd percentile of FSLR's own post-45X trading history.

The anchor used instead is FSLR's own EV/EBIT trading history, computed daily from its own filed TTM EBIT and its own price, restricted to the post-§45X regime (from 2024-02-27, the first date on which the trailing twelve months contained four full quarters of the credit — a deliberate regime cut, because pre-2023 multiples describe a business with no credit at all):

Base exit multiple: 15.5x — the median of the subject's own post-45X history, i.e. the 50th percentile. It is deliberately set at the midpoint of the stated anchor range rather than below every anchor: the NTRA defect (anchors of 7.0x/7.5x with a base of 5.0x) is not repeated here. If anything 15.5x is generous, because by FY2031 the §45X credit is at 50% of statutory value and two years from zero, so a multiple drawn from the full-credit era over-rewards the exit year.

2.2 The result — it depends entirely on the terminal margin, and that is the finding

Headline result

Holding the terminal operating margin at the best product-level margin First Solar has ever demonstrated (9.05%, FY2024) and the exit multiple at the median of its own post-§45X trading history (15.5x), today's price requires a 5-year revenue CAGR of 33.2% against a demonstrated 25.8%. Margin = −7.4pp. The Valuation Criteria FAIL.

And the demonstrated 25.8% is itself withdrawn: First Solar guides FY2026 net sales to $4.9–5.2bn against $5.219bn delivered in FY2025, a decline of 0.4% to 6.1%.

2.3 Sensitivity over the exit multiple — the required parameter

At terminal margin 9.05% (best-ever product margin):

The price does not become achievable on product economics at any exit multiple inside the observed range. Even at 20.0x — the 90th percentile of its own post-45X history — the required 26.6% CAGR still exceeds the demonstrated 25.8%, and exceeds the guided FY2026 path (a decline) by roughly 30 percentage points. Only at 23.3x, the screen's multiple and the 93rd percentile of its own record, does the sign flip — and it flips by 3.1pp, not by the screen's 17.6pp.

For contrast, the same grid on the reported (subsidy-inclusive) margin:

The entire +17.6pp margin the screen reported is the distance between these two tables. It is not a statement about First Solar's growth; it is a statement about which margin was fed in.

2.4 The required-margin view (the same test, inverted)

Holding the company's own guided revenue path fixed and solving for the terminal operating margin the price requires:

At the company's own guided revenue path and its own median multiple, the price requires a 46.5% terminal operating margin — 13.4pp above the highest margin First Solar has ever reported including the subsidy (33.1%, FY2024), and ~47pp above what it has demonstrated on product.

Valuation Criteria (BINDING): FAIL. Not a narrative FAIL — the required path exceeds the demonstrated path by 7.4pp on the most generous product-margin assumption available, and the argument offered for exceeding it (more §45X) is an argument that the credit will be larger, which is true through 2029 and legislated to zero after 2032.

---

3. The 12-month target

Not a DCF. Built off near-term company guidance and First Solar's own trailing multiple history.

3.1 NTM EPS

No Street consensus was obtainable — Alpha Vantage's shared 25/day quota was exhausted. Consensus Criteria: INDETERMINATE; it blocks nothing. The base is therefore built from the company's own reaffirmed guidance (2026-04-30) plus one explicitly-labelled estimate.

3.2 The multiple — the name's own history, percentile stated

Daily trailing P/E from FSLR's own filed EPS and its own price, restricted to the post-§45X regime (2024-02-27 → 2026-07-29, n = 607 daily observations):

Regime-change note, stated rather than ignored. FSLR's pre-2023 multiple history describes a company with no §45X credit and, in 2022, negative EPS — its P/E is undefined or meaningless there. The post-45X window is 2.4 years, which is short but is a genuine 607-observation sample spanning a 2.6x price range ($174.73 to $318.25 in the last year alone). It is used, with the window stated. The full 2022+ sample is not used and the reason is recorded: mixing regimes here would import a period in which the denominator was near zero.

Applying a trailing-P/E band to a forward EPS is conservative-to-neutral while EPS is rising, which it is ($15.48 TTM → $17.40 NTM); the mismatch is stated rather than hidden.

3.3 The target

Base case: $235, +16.0% to spot — above spot.

Why the base assumes essentially no re-rating. The stock already sits at the 10th percentile of its own post-45X multiple range; the base case simply holds it there while NTM EPS grows 12%. It does not assume mean reversion to the 17.3x median, because there is no dated event inside the 12-month window that resolves the subsidy-dependence discount — and one event inside the window makes it more likely to widen: 2027 is the last year US utility-scale projects can be placed in service without having safe-harboured before ~2026-07-04 (§9 of the Research doc), so 2028 estimates come into the forecast window during the target period.

Why the base is nonetheless above spot. On guidance, FSLR earns ~$17 in 2026 and ~$17.4 NTM. At $202.59 that is 11.6x forward with $18.6/share of net cash — 9.2% of the market cap — and no leverage. The market is already pricing the subsidy discount; the base case does not require it to stop.

Bull-case named events: a USITC general exclusion order in Inv. 337-TA-1494 against imported TOPCon modules (instituted 2026-03-25; target date unknown and not estimated); a bookings re-acceleration disclosed at any quarterly release; the second AD/CVD petition (India/Indonesia/Laos, filed 2025-07-17) reaching final orders.

Bear-case named cause: the FY2026 guidance the base rests on is withdrawn or missed, as FY2025's was — the FY2024 release guided 2025 net sales to $5.3–5.8bn and the outcome was $5.219bn, below the original low end.

3.4 Sanity band vs an external professional target

Not run — recorded as a gap. No external professional target was obtained for FSLR in this run. The valuation.md divergence check therefore could not be performed, and no external number is asserted.

---

4. Reconciliation to the screen — exactly where the +17.6pp came from

The screen's result is fully reproducible and the discrepancy is fully attributable:

Single largest driver: the terminal margin. The screen capped a subsidy-inclusive 30.6% down to 16.4% and treated the cap as conservatism. It is not conservatism — 16.4% is still infinitely above the ~0% the business earns on product, and the cap disguised the fact that the input was the wrong quantity, not merely too large.

Calibration item. The generalisable defect: a margin cap applied to a contaminated input looks like prudence and functions as concealment. Any name whose reported operating income contains a government grant, a milestone, or a settlement needs the input decomposed before it is capped. FSLR discloses the decomposition in a published table; the screen had no field for it.