Alnylam Pharmaceuticals [ALNY] — Company Research

Tier-2 initiation · 2026-07-29 · framework: Criteria, 2026-07-29 (v1.5.1 lineage) This document issues no position verdict. It scores Criteria and outputs an analysis. Whether that analysis justifies a position is a question about a particular book.

0. Screen-input validation — read this first

Five agents on 2026-07-29 found the scanner producing confident wrong numbers. ALNY is another instance, and the error is large.

The file the brief names — reports/scan_all_v2/ALNY_analysis.json — does not exist in this worktree. The nearest artefact is .cache/universe_scan/ALNY_analysis.json, whose contents are quoted above.

Root cause of the stale TTM, verified. SEC XBRL companyfacts for CIK 0001178670 has a maximum filed date of 2026-04-06, which predates the Q1-2026 10-Q (filed 2026-04-30) and post-dates but does not contain a usable quarterly decomposition of the FY2025 10-K (filed 2026-02-12). Any pipeline that reads companyfacts alone gets a revenue base 301 days old with no error raised. Recovering the current quarter required going to the filing's rendered R-files (Archives/edgar/data/1178670/000162828026028605/R4.htm) directly. HALO's companyfacts, by contrast, is current (max filed 2026-05-11) — so this is an issuer-level staleness, not a systematic one, which makes it harder to catch. The valuation.md rule "assert recency" is the correct control and it fired here.

1. The trap: is the 53% CAGR product revenue or milestones?

This was the single most important thing to check, and the answer is: it is product revenue. The trap does not bite.

Alnylam reports three revenue lines. All figures below are from the FY2025 10-K income statement (filed 2026-02-12) and the quarterly 8-K Ex-99.1 revenue summaries.

Revenue by line, $m

The decomposed CAGR — this is the answer to the question

Only 1.1 percentage points of the 53.0% demonstrated CAGR is attributable to the lumpy collaboration line. Recurring revenue compounded at 51.9%. The implied-path test is therefore meaningful, and I run it on the recurring base anyway.

Over the shorter, more relevant 2-year window the recurring line grows faster than the total, because collaboration revenue was flat: total 2023→2025 CAGR 42.5%, recurring 57.0%, product 55.1%.

The lumpy items, named and dated

These are the specific non-recurring recognitions inside the collaboration line. Each is disclosed in the quarter's 8-K Ex-99.1 or the 10-K collaboration note:

The Roche line is the clearest demonstration that this is not a run-rate: $337.8m (2023) → $119.5m (2024) → $394.9m (2025), driven by discrete licence-obligation recognitions of $310m / $65m / $300m. Regeneron moved $100.5m → $302.8m → $114.0m. Novartis went to zero.

Management agrees. FY2026 guidance puts combined collaborations and royalties at $400–500m, against $727.4m actual in FY2025 — a guided 31–45% decline in the lumpy line, at the same time as product revenue is guided +64% to +77%. That is management telling you which line is the business.

What I do with it. The implied-path test (ALNY_Valuation.md) is run on a revenue base of $3,986.9m — TTM through Q1-2026 less the identified $300m Roche recognition. The total-TTM run is reported alongside it so nothing is hidden.

2. Mechanism — specific, named, evidenced

AMVUTTRA (vutrisiran) in ATTR amyloidosis with cardiomyopathy (ATTR-CM). Not a theme; a single product with a single label expansion, and the revenue has already moved.

Product-level revenue, $m (10-K disaggregation + quarterly releases)

Geographic detail confirms it is the US label expansion, not price or FX: US AMVUTTRA $411.2m (2023) → $630.6m (2024) → $1,731.2m (2025); Europe $70.9m → $235.4m → $405.9m.

ONPATTRO is being cannibalised by design — management states the decline is "due to patient switches to AMVUTTRA." Total TTR is the franchise number and it tripled.

The evidence chain

1. HELIOS-B Phase 3 in ATTR-CM read out and supports first-line use. New analyses at ACC.26 (April 2026) showed quality-of-life benefit, consistent effect across the disease spectrum including the most advanced patients, and real-world adherence to quarterly HCP-administered dosing. At AHA 2025, cardiac MRI showed amyloid regression in 22% of vutrisiran-treated patients and none on placebo — a structural, not symptomatic, endpoint.
2. US launch dated Q1-2025. Q1-2026 marks "one year since the U.S. launch of AMVUTTRA for ATTR-CM" and the first $1bn product-revenue quarter in the company's history.
3. Diagnosis-funnel investment, named. A partnership with Viz.ai to build an AI-enabled ATTR-CM care pathway for earlier identification, and support for a three-year American Heart Association ten-site learning collaborative. ATTR-CM is a diagnosis-limited market; both are direct attacks on the funnel rather than on share.
4. Next-generation asset already in Phase 3. nucresiran (TRITON-CM), enrolling faster than planned — target enrolment expanded from 1,250 to ~1,750 patients using a pre-specified protocol option, with launch still expected by 2030 if positive. TRITON-PN runs in hATTR-PN.

Why this is the driver of future revenue, not just past

Management's own FY2026 guidance, reiterated at Q1: Total TTR net product revenue $4,400–4,700m, against $2,486.6m in FY2025. That is the mechanism quantified by the people who have the prescription data.

3. Transcript mention-frequency — required core metric

Corpus: EDGAR 8-K Exhibit 99.1 quarterly earnings releases, CIK 0001178670. Fourteen consecutive quarters, 2022Q4 → 2026Q1. One source, never mixed.

Source disclosure, stated plainly. Alpha Vantage EARNINGS_CALL_TRANSCRIPT was the intended corpus. Its free-tier quota (25 requests/day, shared) was already exhausted by other agents when this memo ran — verified, the API returned the rate-limit notice. mention-frequency.md names EDGAR 8-K Ex-99.1 earnings releases as an acceptable first-party substitute ("complete and near-constant length"), which is what is used here. This is prepared corporate text, not a transcript: there is no Q&A section, so the prepared-remarks/Q&A split required by the reference cannot be reported. Counts are normalised per 10,000 words as the ISRG failure mode requires; raw counts are not used anywhere.

Counts per 10,000 words

Emerging / decaying / stable

Emerging. - ATTR-CM — zero for six consecutive quarters (22Q4 → 24Q1), first appearance 24Q2 at 3.9, step change to 29.7 in 25Q1 (the launch quarter), and a new high of 36.7 in 26Q1. This is the cleanest emergence in the series and it corroborates the mechanism rather than being the mechanism's only evidence — the independent corroboration is the revenue line ($970m → $2,314m) and the HELIOS-B readouts, both opened and read, not cited second-hand. - nucresiran — zero until 24Q4, then persistent. The next-generation TTR silencer entering the language before it enters the P&L. - obesity — first-ever appearance 25Q4, sustained into 26Q1. This is the ALN-2232 (ACVR1C, adipose-targeted) Phase 1. It is the one term in the set that is not explicable by anything already in the financials, and it is the open question this method is meant to generate: why is a rare-disease RNAi company putting obesity in its prepared text? Answer not yet established; logged as an open question, not a thesis.

Decaying. - patisiran — 16.9 → 3.7, a 78% decay. ONPATTRO is being retired into AMVUTTRA on purpose, and the language is running ahead of the revenue line. - mivelsiran — spiked 24Q2–24Q3 (15.7, 16.0) then decayed to 3.7. Attention moved on.

Stable. GIVLAARI, OXLUMO, AMVUTTRA (already saturated), pricing (1.7–3.3 throughout — pricing is not a topic management is being pushed on).

Conspicuous absence, and it matters. acoramidis appears 0.0 times in all fourteen quarters. tafamidis appears twice, at 2.0 and 1.6. Alnylam's own prepared text does not name its two direct ATTR-CM competitors. That is not evidence they are unimportant; it is evidence that this corpus will not tell you about them, and the competitive read has to come from elsewhere. It is logged under Downside.

4. Accounting quality — is the reported growth real?

The growth is real. Three things about the reported earnings are not what they look like.

4.1 A new accounting standard added $74.9m to 9M-2025 net income — retrospectively

The FY2025 10-K carries a table (SUMMARY OF SIGNIFICANT ACCOUNTING POLICIES — Schedule of Reported Quarterly Amounts and Adjusted Amounts After Applying New Guidance) restating the first three quarters of 2025 for new guidance on liabilities related to the sale of future royalties:

Revenue is unaffected. But any year-on-year EPS comparison spanning that boundary is not like-for-like, and 58.8% of the reported 9M-2025 earnings improvement is a presentation change. Flagged, not adjusted away — the restated figures are the correct ones to use going forward; the point is that the change in reported earnings is not all operational.

4.2 Customer concentration is rising fast and is barely discussed

10-K, customers above 10% of gross revenues:

A single US specialty distributor moved from 29% to 45% of gross revenue in one year. This is the mechanical consequence of the AMVUTTRA ATTR-CM ramp running through a concentrated specialty channel, and it is not a fraud signal — but it is a real operational dependency that the prepared text does not mention once. Accounts receivable rose from $405.3m to $777.6m; DSO went from 66 days to 76 days, +10 days on a year in which revenue grew 65%. Worth watching; not yet a red flag.

4.3 Non-GAAP excludes 9.4% of revenue in stock compensation

FY2025 stock-based compensation was $348.2m on $3,713.9m of revenue = 9.4%. Every ALNY non-GAAP measure excludes it in full, and FY2026 guidance explicitly excludes $300–400m of SBC from the $2,700–2,800m "non-GAAP R&D and SG&A" figure. This is the Twist precedent in structure (an adjusted metric excluding a large, recurring, cash-equivalent cost) though smaller in relative terms and fully disclosed. All valuation in this memo is on GAAP operating income.

4.4 What is genuinely clean

• Accruals are strongly negative — FY2025 net income $313.7m against operating cash flow $524.1m. Accruals (NI − CFO) / average assets = −4.6%. Cash exceeds earnings; this is the good direction.
• Free cash flow positive: CFO $524.1m less capex $58.7m = $465.4m.
• No revenue-recognition gymnastics on the product line: net product revenue is straightforward gross-to-net on specialty-pharmacy sell-through, and the allowance/reserve rollforward is disclosed.
• No settlements, no legal recoveries, no "other income" in revenue. Unlike the 10x Genomics precedent, there is no non-recurring income buried in the top line. The one large non-recurring item — the $300m Roche licence recognition — is in a separately captioned line, separately disclosed, and separately guided down.

4.5 The item that is not an accounting issue but is an economic one

ALNY carries $1,697.2m of "liabilities related to the sale of future royalties and development funding" (Q1-2026). It has sold forward part of the Leqvio royalty stream and taken development funding against vutrisiran and zilebesiran. The unwind runs through interest expense: Q1-2026 interest expense of $69.3m included $40m on the Leqvio royalty sale and $26m on the development funding. Those are prior claims on the same cash flows this memo capitalises. They are treated as debt in the base-case enterprise value. The alternative treatment (excluding them) raises net cash to +$2,022m and lowers the required CAGR from 10.7% to 9.8% — a small effect, reported for completeness.

5. Product-cycle intelligence

Manufacturing. $250m committed to expand Norton, Massachusetts into "the industry's first fully dedicated, proprietary siRNA enzymatic-ligation manufacturing facility" (siRELIS), expected to expand capacity and "significantly reduce production costs." Relevant because cost of goods sold rose to 20.0% of net product revenue in Q1-2026 from 15.0% — attributed by management to "an increased blended royalty rate payable on net sales of AMVUTTRA." Gross margin is structurally compressing as AMVUTTRA mixes up; siRELIS is the declared offset and it is unproven.

6. Criteria scoring

Types are as defined in criteria.md for the long-only absolute-return strategy. The memo blocks on none of them. Every Criteria returns PASS / FAIL / INDETERMINATE; a missing input is INDETERMINATE.

Quality Criteria — BINDING — archetype INFLECTION → COMPOUNDER — PASS

FY2025 was the first GAAP-profitable year in the company's history (operating income $501.6m, net income $313.7m). The INFLECTION standard is applied, since the transition completed inside the measurement window.

Note on the data. ALNY publishes no GrossProfit XBRL tag. Under the pre-D1 logic this would have evaluated nan > 0.50 and returned FAIL. It is computed here from the disclosed income statement (revenue less cost of goods sold) and returns PASS. This is exactly calibration item D1 and it would have rejected ALNY silently.

Valuation Criteria — BINDING — PASS (full working in ALNY_Valuation.md)

• Required revenue CAGR 10.7% at 27.4x EBIT exit / 35% terminal margin / 10% WACC / 5y, on the $3,986.9m recurring base.
• Demonstrated (recurring, 3y) 51.85%. Margin = +41.1pp.
• Exit multiple 27.4x EBIT, basis GROWTH_MATCHED (n=3). Implied compression from today's trading multiple: 0.3x of sales, −3.4%.
• Sensitivity is run over the exit multiple. At the punitive end (15x) the requirement is 24.9% — still 27.0pp below demonstrated.

Downside Criteria — MEASURED — scored, blocks nothing

Named cause, realistic permanent-loss scenario and probability in §7 below and in ALNY_Trade_Construction.md.

Liquidity Criteria — BINDING — PASS on the common; options vehicle FAILS at size

• Equity: 63-day median $ADV $328.9m; median 1.13m shares/day. A $10m position is 3.0% of one day.
• Options chain pulled, not assumed. January-2027 calls: open interest is real at two strikes (280: 3,578; 370: 1,736) but quoted size is 1–5 contracts and the 280-strike is quoted 45.02 / 49.97, a 10.4% bid-ask on mid. March-2027 total call open interest across $260–400 is 326 contracts, maximum 116. Any defined-risk structure above roughly 50 contracts is not fillable at quoted size. This is the HCA failure mode and it is disclosed rather than assumed away.

Momentum Criteria — MEASURED — entry timing only, never a selection veto

This governs when to enter a position the thesis already justifies. It never governs whether to own one. ALNY sits in the second momentum quintile, 41% below its 52-week high, with revenue guided +49%. That combination is a timing observation, not a rejection.

Catalyst Criteria — MEASURED — see ALNY_Catalyst_Calendar.md

Four dated pipeline readouts in H2-2026 plus quarterly guidance checkpoints. Time works for a long, so a dated event is not required to own this.

Consensus Criteria — MEASURED — INDETERMINATE

Alpha Vantage EARNINGS_ESTIMATES quota was exhausted by parallel agents at the time of this run (verified — the API returned the rate-limit notice). No consensus NTM revenue or EPS was obtained. A quota gap leaves this blank and blocks nothing. House-versus-Street is therefore not computable and is not estimated. The near-term base used for the 12-month target is management's own reiterated FY2026 guidance, which is a filed number, not a fabricated consensus.

Peer Spread Criteria — MEASURED

Named peer: Vertex Pharmaceuticals [VRTX] — large-cap, single-dominant-franchise biopharma, the closest structural analogue.

ALNY trades at the same sales multiple as a peer growing 7x slower, because its margin has not yet converged. Versus its own history, ALNY's EV/TTM-sales of 9.2x sits at the 2nd percentile of its 2022–2026 range and the 3rd percentile of the trailing three years.

Short Mechanism Criteria — MEASURED — FAIL as a short (i.e. no short case)

Requires decelerating growth and exhausted margin runway. ALNY has accelerating product growth (+81.4% FY2025 vs +32.6% FY2024) and a margin runway that is visibly opening, not spent (13.5% GAAP operating margin FY2025 against a guided path to roughly 25% in FY2026). Neither leg holds.

Sub-sector Criteria — MEASURED

Pharma on the reference taxonomy (rare disease / RNAi platform). Correlation-relevant neighbours for concentration purposes: VRTX, REGN, NBIX, BMRN.

7. Downside Criteria in full — the named cause

Volatility is not the risk. Permanent impairment is. 20-day realised volatility is 42%; that is a sizing input, not a bear case.

Named cause 1 — ATTR-CM class economics reset

Two competitors are already in the ATTR-CM market: Pfizer's tafamidis (the incumbent stabiliser) and BridgeBio's acoramidis. Alnylam's own prepared text names neither in a meaningful way — acoramidis appears zero times in fourteen quarters. Alnylam's mechanism (silencing) is differentiated from stabilisation and HELIOS-B showed amyloid regression, so the clinical case for premium share is real. The economic risk is not displacement but net-price compression: three branded agents competing for the same diagnosed pool, into a US payer environment that now includes Medicare Part D negotiation.

Scenario: US AMVUTTRA net price compresses 20–25% from 2028 and TTR share settles below plan, capping the franchise near $6bn rather than the $9bn+ the current trajectory implies. On $6.5bn of total revenue at 8.5x sales — the bottom of ALNY's entire observed 4.5-year multiple range — enterprise value is approximately $55bn, still above today's $39.6bn. On honest arithmetic this is not a permanent-impairment case at this price. It is a "the stock does nothing for three years" case.

Named cause 2 — a HELIOS-B-class safety or label event

The permanent-loss case for ALNY is scientific, not economic. A post-marketing safety signal in vutrisiran, or an ATTR-CM label restriction that narrows the eligible population, removes the mechanism rather than repricing it. At that point AMVUTTRA is worth a fraction of $2.3bn of run-rate revenue and the pipeline is discounted with it.

• Probability: 10%.
• Drawdown in that event: −55% to −65%, to roughly $100–130, which is ~4–5x recurring revenue — the trough multiple the sector applies to a broken single-franchise story.

Named cause 3 — channel and capital-structure claims

• Distributor A is 45% of gross revenue. A disruption, a contract renegotiation, or a channel-inventory correction hits reported revenue with no change in underlying demand.
• $1,697.2m of royalty-monetisation and development-funding liabilities rank ahead of equity in the Leqvio and vutrisiran/zilebesiran cash flows.

Going concern

No. $3.0bn of cash and marketable securities, GAAP profitable, $524.1m of operating cash flow in FY2025, $465.4m of free cash flow. A going-concern bear case is not arguable and is not argued.

8. What is not supported — stated rather than estimated

1. No consensus estimates. Alpha Vantage quota exhausted by parallel agents. No Street NTM revenue, no Street EPS, no external price target. The Consensus Criteria is INDETERMINATE and the 12-month target's "sanity band" against a professional's target could not be computed. It is left blank, not filled in.
2. No transcript Q&A. The mention-frequency corpus is prepared earnings-release text only. The prepared-remarks-versus-Q&A split that mention-frequency.md requires is not reportable from this corpus, and is stated as a limitation rather than fabricated.
3. FY2027 revenue is a house estimate, not guidance and not consensus. Guidance exists only for FY2026. The 12-month target's dependence on it is isolated and shown.
4. The obesity mention emergence is an open question, not a thesis. ALN-2232 is a Phase 1 asset with no disclosed data. No value is ascribed to it anywhere in this memo.
5. The magnitude of AMVUTTRA gross-margin compression is not disclosed. Management states the blended royalty rate payable on AMVUTTRA net sales is rising; the rate is not given. COGS at 20.0% of product revenue in Q1-2026 is the only observable.
6. The screen's "required CAGR 33.4%" could not be reproduced because the parameters behind it are not in any artefact present in this worktree. It is not used.
7. Beta is not regressed. No beta figure is asserted; the WACC of 10% is the framework default and is named as such.