Alnylam Pharmaceuticals [ALNY] — Financial Model Notes

2026-07-29. Every figure below is traceable to a named primary filing. Where a figure is a house estimate it is labelled [HOUSE]. Where a figure is management guidance it is labelled [GUIDANCE]. Nothing else is asserted.

1. Data lineage — what was read, and from where

Critical lineage warning. ALNY's SEC XBRL companyfacts feed has a maximum filed date of 2026-04-06. It therefore does not contain the Q1-2026 10-Q (filed 2026-04-30) and cannot produce a current TTM. This is the defect that produced the screen's $3,117.8m / 301-day-stale revenue. Every revenue figure in this model comes from the rendered R-files of the actual filings, not from companyfacts. Cash flow and SBC — which are annual and were captured before the cut-off — are the only companyfacts-sourced items, and they are FY2025 annuals, so the staleness does not bite there.

2. Revenue build — the three lines, kept separate on purpose

2.1 Annual, $m

TTM product+royalty = 3,750.7; TTM collaborations = 536.3, of which $300.0m is the single Q3-2025 Roche licence-obligation recognition. Recurring TTM used in the reverse DCF = 4,286.9 − 300.0 = $3,986.9m.**

2.2 Quarterly, $m — the series the model runs on

Q3-2025's $351.7m collaboration line is 6.8x the Q4-2025 figure. That single quarter is why the total-CAGR question had to be asked, and §2.3 answers it.

2.3 Product detail, $m

AMVUTTRA by geography, FY2023 → FY2024 → FY2025: US 411.2 → 630.6 → 1,731.2; Europe 70.9 → 235.4 → 405.9; RoW 75.8 → 104.4 → 176.7.

2.4 Collaboration detail by counterparty, $m — the lumpiness, itemised

Modelling instruction: the licence-obligation sub-line is not forecastable and must not be extrapolated. It ran 310.0 / 65.0 / 300.0. Only the development-services sub-line (24.0 / 45.8 / 89.0) has a trajectory, and it is small.

3. Cost structure

Gross margin on product = (product revenue − COGS) / product revenue: FY2023 78.4%, FY2024 81.4%, FY2025 77.3%, Q1-2026 80.0%.

Modelling flag — COGS is the under-discussed line. COGS as a percentage of product revenue moved from 15.0% (Q1-2025) to 20.0% (Q1-2026), which management attributes to "an increased blended royalty rate payable on net sales of AMVUTTRA." The rate itself is not disclosed. As AMVUTTRA mixes further up, this ratio rises mechanically. The declared offset is the $250m siRELIS enzymatic-ligation facility in Norton, Massachusetts, which management says will "significantly reduce production costs" — with no date and no quantum. A model that holds COGS at 20% is making an assumption, not reading a disclosure.

4. Below the line, and the accounting change

Q1-2026 interest expense of $69.3m decomposes as $40m on the liability related to the sale of future Leqvio royalties and $26m on the vutrisiran/zilebesiran development-funding liabilities. This is the unwind of cash already received; it is a real economic cost of the revenue being capitalised.

The restatement — model this correctly or the EPS series is wrong

The FY2025 10-K restated the first three quarters of 2025 under new guidance on liabilities related to the sale of future royalties:

+$74.9m of net income, +58.8%, from a presentation change. Revenue is unaffected. Use the restated figures; the point of recording it here is that any EPS growth rate spanning 2024→2025 is not like-for-like.

5. Balance sheet and capital structure, 2026-03-31

Two net-cash conventions, both stated: - Base case (+$324.9m) treats the royalty-monetisation and development-funding liabilities as debt. This is the correct choice here because the revenue being capitalised (Leqvio royalties, AMVUTTRA product revenue) is the same cash flow those liabilities are repaid from. Excluding them would double-count. - Alternative (+$2,022.0m) counts only the convertible debt. It lowers the required CAGR from 10.7% to 9.8% — a 0.9pp effect. Reported so the choice is visible, not used.

DSO: 777.6 / 3,713.9 × 365 = 76 days (FY2025), against 405.3 / 2,248.2 × 365 = 66 days (FY2024). +10 days on a year of 65% growth. Watch item.

6. Cash flow

Accruals check (retained on both archetypes): (net income − operating cash flow) / average total assets = (313.7 − 524.1) / ((4,966.3 + 4,240.0) / 2) = −4.6%. Cash exceeds earnings. PASS.

FY2025 operating cash flow includes $118m of payments on the Leqvio royalty-sale liability and $85m on the development-funding liabilities, both recorded in interest expense.

7. FY2026 — [GUIDANCE], and what it implies

FX assumptions in the guidance are disclosed: 1 EUR = 1.17 USD, 1 USD = 157 JPY, as of 2025-12-31. Growth is guided as 64–77% at both actual and constant currency, so FX is not doing the work.

Implied FY2026 GAAP operating income [HOUSE arithmetic on GUIDANCE inputs]: revenue $5,550m − COGS (~20% of $5,100m product = $1,020m) − cost of collaborations (~$15m) − GAAP R&D+SG&A (non-GAAP $2,750m + SBC $350m = $3,100m) ≈ $1,415m, a 25.5% GAAP operating margin.

Note the shape of the guidance. Product revenue up 64–77%; the lumpy collaboration line down 31–45%. Management is guiding the recurring line up and the non-recurring line down, which is the same conclusion this memo reached from the historical decomposition.

8. FY2027 — [HOUSE], isolated so it can be replaced

No guidance and no consensus exists for FY2027 (Alpha Vantage quota exhausted — see the disclosed limitations). The 12-month target depends on it, so it is isolated here:

TTM revenue reportable twelve months from now (through Q2-2027) ≈ (5,550 + 7,000) / 2 = $6,300m. This is the denominator of the 12-month target and it is the single most replaceable input in the model.

9. Model verification and known weaknesses

Verified: - Quarterly revenue splits sum to the 10-K annuals to the dollar: 468.5 + 672.2 + 851.1 + 994.7 = 2,986.5 ✓; 99.2 + 61.5 + 351.7 + 40.9 = 553.4 ✓; 26.5 + 40.0 + 46.2 + 61.4 = 174.0 ✓. - Total revenue reconciles: 2,986.5 + 553.4 + 174.0 = 3,713.9 ✓ against the 10-K income statement. - Share count cross-checked three ways: 10-Q cover 133,512,573 (2026-04-24); balance sheet 133,444k issued and outstanding (2026-03-31); diluted weighted 138,226k (Q1-2026). - Market cap reconciles to the screen: 288.50 × 133.513m = $38.52bn vs screen $38.3bn. - Momentum reproduces the scanner independently: 12-1 −9.1% vs scanner −9.7%; 58.7% of 52-week high in both. - Operating margin arithmetic ties: 501.6 / 3,713.9 = 13.51% ✓.

Known weaknesses, stated: 1. No consensus. FY2027 is a house estimate. If the Street is materially different, the 12-month target moves with it and this memo would not know. 2. The EV/TTM-sales history uses annual capital-structure snapshots, not daily. Share counts and net debt are stepped by fiscal year. This is materially better than a constant-share approximation but it is not exact; the current-day figure (9.22x) is exact. 3. Terminal margin of 35% is an extrapolation from a company that has been profitable for one year. It is sensitivity-tested (25% → required CAGR 18.5%; 40% → 7.8%) rather than defended. 4. The blended royalty rate on AMVUTTRA is undisclosed. COGS at 20% of product revenue is an observation, not a forecast. 5. No regressed beta. WACC of 10% is the framework default and is named as such, not derived. 6. No Q4-2025 stand-alone XBRL quarterly tag exists (10-K filers do not tag Q4 separately). Q4-2025 revenue of $1,097.0m is derived as FY2025 less the three reported quarters, and cross-checked against the 8-K Ex-99.1 Q4 revenue summary, which states the same figure. ✓