Phase Space AI

Valuation

MannKind Corporation [MNKD]

MannKind Corporation [MNKD] — Valuation

Two outputs, both mandatory per references/valuation.md: a 12-month target and the implied-path test. Reporting only one is a defect. Spot $3.8875 (2026-07-29).

---

1. The verified valuation inputs

Everything downstream depends on these. The screen's versions are shown to make the correction auditable.

Pro-forma caveat, stated plainly: the July placement and the $45m CVR payment are the only two events layered onto the 31 March 2026 balance sheet. Q2 2026 operating burn is not included because Q2 2026 has not been filed. Q1 2026 operating cash flow was −$5.4m; if Q2 is similar the pro-forma cash figure is ~$5–15m generous.

On the WACC. The framework default is 10%. A name-specific build: risk-free 4.3%, ERP 5.5%, beta ≈ 1.48 (realised vol 77.0% × assumed 0.3 correlation ÷ 15% market vol) → cost of equity 12.4%. Cost of debt 9.09% (the 10-Q's stated effective rate), with no tax shield — MannKind carries a full valuation allowance. At E $1,251m / D $325m: WACC ≈ 11.7%. Both 10% and 11.7% are run below.

---

2. The implied-path test — the Valuation Criteria

Instrument: assets/reverse_dcf.py. Solved for revenue CAGR; every other parameter held fixed and named.

2.1 Reproducing the screen, then correcting it

The screen's arithmetic is correct — it reproduces to 7.03% exactly. Its balance sheet is not.

2.2 The margin: demonstrated − required

criteria.md: "The margin in percentage points: demonstrated − required. This is the number the strategy ranks on." The required figure is 11.82%. The question is what "demonstrated" honestly means.

Headline margin: −3.6pp (TTM organic, the least noisy honest measure). The screen's +44.8pp does not merely shrink — it inverts. Two independent errors compound: an enterprise value understated by $288m raises the required rate from 7.0% to 11.8%, and a demonstrated rate that is a 2022–23 launch fossil overstates the delivered rate by ~44 percentage points.

2.3 Exit multiple, implied compression, and the anchoring problem

The screen used 22.6x EBIT, basis GROWTH_MATCHED, peer n = 130. This memo does not verify that anchor and notes two problems with it: a 130-name comparator set drawn across SIC 28 is not growth-matched to anything in particular, and the growth it was matched to (51.8%) is the fossil rate.

criteria.md requires the exit multiple to be drawn from a comparator set whose growth brackets the subject's growth at the exit year, or declared UNIDENTIFIED. MannKind's growth at year 5 is precisely the unknown being solved for, so no such set can be constructed ex ante. The EBIT-basis exit multiple is therefore declared UNIDENTIFIED, and the 22.6x runs are retained only as a like-for-like correction of the screen.

The defensible anchor is MannKind's own EV/Sales history, which valuation.md explicitly privileges. Built here from daily SIP closes, TTM revenue stepped in at each actual 10-K/10-Q filing date, and point-in-time share count and net debt from XBRL instant facts (21 quarters, 2021-Q1 → 2026-Q1) — not today's balance sheet held constant, because MannKind's went from net cash to $189m of net debt inside a year.

Current EV/Sales 3.99x (pro forma) sits at the 19th percentile of three years and the 40th percentile of one year. The name is cheap against its own history — but its own history was set when revenue was compounding above 40%.

Required CAGR on the sales basis, anchored on the name's own range (WACC 10%, 5 years):

Implied compression, stated as a number: at the 22.6x EBIT / 16.2% terminal-margin exit, terminal EV/Sales = 22.6 × 16.2% = 3.66x, against today's 3.99x — an implied compression of 0.33x, or 8.3%. That is a mild compression assumption, not a conservative one, given the business is decelerating.

2.4 Sensitivity — over the exit multiple and the terminal margin, never over scenario probabilities

Required revenue CAGR (%), EBIT basis, terminal margin held at 16.2%, EV $1,434m:

Required revenue CAGR (%) by terminal margin, at 22.6x and WACC 10%:

This is the highest-variance parameter and it is where the answer lives. The screen's 16.2% terminal margin is roughly four times the actual TTM operating margin of 4.1% and 1.5 times FY2025's 11.1%. At MannKind's realised current margin the price requires a 47.2% five-year revenue CAGR. The 16.2% assumption does most of the work in producing a "cheap" answer.

2.5 Verdict — PASS WITH ARGUMENT

• Not PASS: the required 11.8% exceeds demonstrated organic growth of 8.2% (and vastly exceeds the −4.7% most recent organic quarter).
• Not FAIL: the excess is attached to something specific, evidenced and dated, not to narrative — Furoscix ReadyFlow, FDA-approved 23 July 2026, in a franchise growing ~70% year-on-year, for which MannKind capitalised $129.6m of IPR&D that has now been de-risked by approval.
• PASS WITH ARGUMENT, with the argument's burden stated: to deliver 11.8% for five years with the UT block (64% of revenue) flat, Furoscix must grow from a ~$62m annualised run rate to roughly $300m by FY2031 — approximately 5x. If the UT block declines rather than staying flat, Furoscix must do more than that. That is the whole thesis, and it is one product.

---

3. The 12-month target

Built per valuation.md — near-term revenue base, named product-cycle events, multiple on the name's own trading range with the percentile stated. Not a DCF, not a peer median.

3.1 Step 1 — the NTM revenue base

No consensus was obtainable (Alpha Vantage returned its 25/day rate-limit message on 2026-07-29). Per the brief this blocks nothing, but it means this base is a house build, not consensus, and is labelled as such. It is assembled from disclosed run rates only.

Honesty note on that +13.6%: most of it is the scPharma acquisition annualising through Q2/Q3/Q4 2026, not organic growth. Organic NTM growth inside this build is roughly +2% to +4%.

3.2 Step 2 — named product-cycle events inside 12 months

Each appears in MNKD_Catalyst_Calendar.md with its date and its source.

1. Furoscix ReadyFlow US launch — FDA approved 23 July 2026. The only estimate-moving positive event MannKind controls.
2. Q2 2026 results (early August 2026, pattern-inferred, not confirmed) — the first read on whether Furoscix's Q1 sequential decline was payer-reset noise or the real run rate.
3. United Therapeutics files Tresmi for PAH and ILD — UT stated on its 25 Feb 2026 call it planned to file "within the year" (i.e. during 2026). A filing acceptance is a dated, checkable estimate-moving negative.
4. MNKD-201 Phase 1b top-line data — company-stated Q3 2026.
5. Sagard $50m milestone window closes 31 Dec 2026 — see §4.3 of the Research doc; treated as failing.
6. Resale registration statement for the July placement — due within 30 days of the 24 July closing, so by ~23 August 2026. Removes the legend from 12.85m shares; a mechanical overhang.

3.3 Step 3 — the multiple, on MannKind's own range

• Current EV/Sales 3.99x = 40th percentile of 1 year, 28th of 2 years, 19th of 3 years.
• Base multiple: 4.15x — the 1-year median. The 1-year window is chosen deliberately: it is the only window whose regime resembles today's (post-deceleration, post-leverage, post-acquisition). The 3-year median of 5.45x was set while revenue compounded above 40% and applying it would be paying for a growth rate the company no longer has.
• Base 4.15x sits above the bear anchor (1-year p25, 3.35x) and below the bull anchor (2-year median, 4.93x), so it does not violate the rule that a base multiple may not sit below every stated anchor.

3.4 The target

Shares in 12 months: 325.8m (321.76m + ~4m RSU/option vesting at the recent run rate). Net debt evolves from $189.1m pro forma by the case's free cash flow.

12-month target: $4.55 — +17.0% to spot.

Two things about that number, both important:

1. It is not multiple-driven. Holding today's 3.99x flat still gives $4.32, +11.2%. Only ~6pp of the 17% comes from re-rating to the 1-year median.
2. It is acquisition-driven, not organic. The revenue growth inside it is scPharma annualising and ReadyFlow launching. Strip those and the base case is roughly flat. The target says the market has under-priced a known acquisition lapping through; it does not say the underlying business is compounding.

No external professional target was available for the sanity band — MNKD is not in the reference book and no consensus was obtainable. That check is not performed, and is declared missing rather than substituted with a peer-derived figure.

3.5 Reconciling the two outputs

They do not conflict; they answer different questions over different horizons, which is exactly why valuation.md requires both.

• 12 months (+17%): the acquisition annualises, ReadyFlow launches, revenue prints ~$410m, and the multiple normalises to its own 1-year median. Perfectly plausible.
• 5 years (−3.6pp margin): for today's price to be correct, that has to keep happening at 11.8% a year for five years, out of one product, while the 64% of revenue that United Therapeutics controls does not shrink — against a partner that has publicly named its replacement and plans to launch it in 2027.

Invalidation level. The bear case's named cause is checkable at a price: $3.16 is the bear target and $2.28 is the 52-week low. A close below $3.16 on Tyvaso-DPI or Tresmi news would confirm the thesis-break rather than represent noise on a 77%-vol name.