Phase Space AI

Valuation

Insmed [INSM]

Insmed [INSM] — Valuation

As of 2026-07-29. Spot $103.16 (Alpaca snapshot). Framework v1.5.1. Tier 2.

---

1. COMPANY STATE — declared before any instrument is chosen

STATE C — scaling but economically observable

Why not State D. State D requires negligible revenue, unstable or deeply negative gross margin, financing dominating the outcome, or no reverse-DCF solution in range. None holds. TTM revenue is $819.5m and FY2026 consensus is $1,700.9m on 21 analysts. Gross margin is not merely positive, it is one of the highest in the corpus and has been stable for six years. Net cash is +$479.1m and there is no active ATM (terminated 2024-11-18), so financing is not the dominant variable. The solver converges everywhere.

Consequence. An EV/EBIT exit multiple is admissible — the State-D prohibition does not apply. The instrument is the State-C two-dimensional expectations surface, and the terminal margin must be BUILT, not assumed.

---

2. TERMINAL MARGIN — 32.0%, built through the opex bridge

The brief's condition for granting a loss-maker a positive terminal margin is a named, evidenced mechanism: a specific product, an approval, and a launch curve. All three exist, with sources:

• Product: BRINSUPRI (brensocatib 25mg/10mg tablets), the first approved therapy for non-cystic-fibrosis bronchiectasis anywhere.
• Approval: FDA 2025-08-12 (FY2025 10-K MD&A); EC November 2025; MHRA February 2026 (Q1'26 10-Q).
• Launch curve: $28.1m (Q3'25 part-quarter) → $144.6m (Q4'25) → $207.9m (Q1'26).

So the parameter is admissible. It is not INDETERMINATE. It is, however, evidence grade C, and the bridge says why.

m_gross,T                     82.0%
  less R&D                   (21.0%)
  less S&M + G&A             (28.0%)
  less amortisation           (0.2%)
  ------------------------------------
  m_EBIT,T                    32.8%  -> stated as 32.0%

Constraint check: 32.0% ≤ 82.0%. PASS.

Terminal gross margin, and why it is BELOW the reported figure. This is the unusual direction. Reported Q1'26 gross margin is 84.5% ($305.964m revenue less $47.420m cost of product revenues). It is temporarily flattered by an accounting policy: all BRINSUPRI product costs incurred before FDA approval on 2025-08-12 were expensed as R&D, and management states cost of product revenues will "benefit during 2026 and beyond, as we sell through inventory that was expensed prior to FDA approval of BRINSUPRI" (FY2025 10-K MD&A). The reported figure is therefore a ceiling, not a floor. FY2025 was 79.7% and FY2024 76.4%. Setting terminal at 82.0% reflects the mix shift toward BRINSUPRI — an oral small molecule, structurally cheaper than ARIKAYCE's liposomal inhalation suspension — while normalising the zero-cost inventory out.

Opex, and the honest admission. The gross-margin leg is grade A: 24 quarters of filed data. The opex leg has no Insmed precedent at scale whatsoever — the company has never printed a positive operating margin in a filing history stretching to 2000. The 21% R&D and 28% SG&A assumptions are drawn from what commercial-stage rare-disease operators run at $2–5bn of revenue, used as a sanity band and not as an anchor. Range: 25%–38%.

What was deliberately not done. No peer-median terminal margin was used. That is the specific defect this memo was commissioned to avoid: the scan credited 22 loss-making names with a positive peer-median terminal margin, SMR among them at an actual −2,190.6% operating margin, and every stored scan_v2/scan_v3 terminal margin is documented as meaningless (one universe-wide constant of 14.4% on 70 of 111 names).

---

3. THE EXIT MULTIPLE IS NOT A FREE PARAMETER — and this is where the name fails

The identity from the brief:

EV_T / EBIT_T  =  (1 − t)(1 − g/ROIC) / (WACC − g)

At t = 21%, g = 3%, WACC = 10%:

At WACC 9% the ceiling rises to 13.17x; at WACC 11% it falls to 9.88x. For a business funded at a 9.6% fixed secured-loan rate, no exit multiple above ~13x is internally consistent at any ROIC.

---

4. IMPLIED-PATH TEST — the primary long-horizon output

Solved with assets/reverse_dcf.py. Parameter solved for: revenue CAGR (and, separately, terminal margin). Held fixed: spot $103.16, shares 216.752m, net cash +$479.1m (lease-inclusive), base revenue $819.5m TTM, 5 years, WACC 10%.

4a. Solving for required CAGR at the built 32% terminal margin

Demonstrated / consensus-extended CAGR: 37.2%. Built as: TTM $819.5m → FY2027 consensus $2,776.4m (a +100.8% annualised rate over 1.75 years, i.e. launch math, taken as given), then 12%/yr for the remaining 3.25 years → ~$4.0bn in year 5. $819.5m → $3,988m over five years is 37.2%.

MARGIN (demonstrated − required) = 37.2% − 62.1% = −24.9pp at a 12x exit. At the identity-consistent ceiling of 11.29x: 37.2% − 64.1% = −26.9pp.

4b. Solving for required terminal margin at the 37.2% consensus-extended CAGR — the decisive output

At the highest exit multiple the identity permits, today's price requires a 78.3% terminal EBIT margin against a terminal gross margin of 82.0%. That passes the literal arithmetic constraint (78.3 ≤ 82.0) and fails the full expense bridge decisively, which is exactly the failure mode valuation.md warns about: "the ceiling is necessary but not sufficient... a 25% gross-margin business reaching a 15% EBIT margin leaves 10% of revenue for all R&D, selling and administrative cost. That can be impossible while passing the literal constraint." Insmed would be left 3.7% of revenue to run a global commercial organisation across three continents plus four Phase 3 TPIP programmes and two Phase 1 gene therapies. The highest operating margin in large-cap pharmaceuticals is roughly 45–50%.

The flip point. The built 32% terminal margin clears only at a ~28x exit multiple — 2.5x above the identity ceiling of 11.29x.

Conclusion of the implied-path test: there is no admissible exit multiple at which today's price is justified by the consensus-extended revenue path. The gap is not a matter of parameter choice. It survives the entire range the identity permits.

Sensitivity was run over the exit multiple, never over scenario probabilities, per the standing rule.

4c. The honest caveat on the base

The base is TTM revenue of $819.5m, per the rule "use TTM revenue, never the last fiscal year" — which exists because last-FY understated MU by 136%. Here the rule cuts the other way: a name mid-launch has a TTM that is stale by construction. Q1'26 annualised is $1,224m, and the FY2026 consensus base is $1,700.9m. Running from $1,700.9m over five years, the required CAGR at 12x falls to roughly 41% against a consensus-extended ~15–18% out-year rate — still a large negative margin, but a materially smaller one. Both bases are reported; neither is hidden. The conclusion does not change sign under either.

---

5. TWELVE-MONTH TARGET — $122.61, +18.9%. Percentile: UNIDENTIFIED

Why the percentile is UNIDENTIFIED

EV/EBIT history does not exist. Insmed has never printed a positive EBIT. There is no percentile to state.

EV/Sales history exists and is disqualified by a regime change. Measured over 860 trading days from 2023-02-23 the current 26.7x EV/TTM-sales sits at the 38.5th percentile (window median 34.4x, range 10.6x–97.5x). Measured over the 242 trading days since the BRINSUPRI approval on 2025-08-12 it sits at the 5.8th percentile (window median 60.2x, range 23.7x–97.5x).

Neither number is usable, and the second is the more misleading of the two. Across that 242-day window the TTM revenue denominator rose from $397.6m to $819.5m — +106% — while the price fell from a peak of $211.41 to $103.12, −51.2%. A percentile computed on a ratio whose denominator doubled inside the window is measuring revenue catch-up, not a market re-rating, and it will read "cheap" mechanically for as long as the launch continues. The BRINSUPRI approval took quarterly revenue from $107.4m to $306.0m in three quarters and added a 500,000-patient US indication with no competing approved therapy. Pre-approval observations price a single-product $430m ARIKAYCE franchise plus pipeline optionality; post-approval observations price a two-product commercial platform. These are different companies.

Per valuation.md — "if history is too short or spans a regime change, declare UNIDENTIFIED rather than substituting a peer median" — the percentile is UNIDENTIFIED. No peer median has been substituted.

What was used instead

Today's own EV/NTM-sales multiple, held constant, with the revenue window rolled forward. No re-rating is assumed in either direction — which is the only defensible stance when the percentile is unidentified.

EV / NTM sales today          9.40x   ($21,881m / $2,328.3m)
NTM revenue rolled forward    $2,776.4m   (FY2027 consensus, 21 analysts)
Implied EV                    $26,098m
plus net cash                 +$479.1m
Equity value                  $26,577m
/ 216,752,451 shares          =  $122.61
Upside to $103.16 spot        =  +18.9%

Sanity band against the name's own post-approval price range: low $91.88, median $146.25, high $211.41. A $122.61 target sits in the lower half of that range — consistent with the out-year estimate direction below, and consistent with the note in valuation.md that a house whose every target sits below spot is expressing a market view rather than valuing companies. This one sits above spot, and below the name's own post-approval median.

The named product-cycle events inside the 12 months

Estimate-moving, all dated and sourced in INSM_Catalyst_Calendar.md: the ARIKAYCE all-MAC sNDA submission (2H 2026), the Japan brensocatib regulatory decision (2026), TPIP PAH Phase 2b open-label-extension data (Q3 2026), and the TPIP PPF Phase 3 initiation (2H 2026).

Revision direction — split, and it is the most informative single fact in the file

The near term is being revised up and the out-year is being cut 82.9% in ninety days. That window brackets the CEDAR failure (April 2026), the Q1'26 print (2026-05-07) and the ARIKAYCE deceleration. The market is marking up the BRINSUPRI launch and marking down the terminal earnings power at the same time — which is precisely what the implied-path test above says must happen.

---

6. Criteria

No position verdict is issued. The book decides.