Phase Space AI

Valuation

ResMed [RMD]

ResMed [RMD] — Valuation

Spot $206.91 (2026-07-30). Two outputs, both required: a 12-month target and the 5-year implied-path test.

---

1. COMPANY STATE — declared first

STATE A — mature and structurally stable. Full evidence in RMD_Research.md §6. Consequence per valuation.md: the terminal margin is built from RMD's own normalised economics plus an explicit forward bridge; industry data is a sanity band, never an override. The instrument is: fix the underwritten margin, solve for price-implied growth, show the neighbouring surface.

evidence_grade: A−.

---

2. Verified inputs — every one rebuilt, none taken from the screen

---

3. TERMINAL MARGIN — 35.5%, built from the opex bridge

Underwritten terminal EBIT margin: 35.5%.

The hard constraint first

m_EBIT,T  ≤  m_gross,T        →    35.5%  ≤  62.5%    ✓   (27.0pp of headroom)

Terminal EBIT consumes 56.8% of terminal gross profit, leaving 27.0pp of revenue for all R&D, selling and administrative cost against 26.9pp actually spent in 9M FY2026. The sufficiency test — not just the literal ceiling — is satisfied with essentially no stretch.

The bridge, every line from the 10-Q / 10-K

Arithmetic: 62.5 − 6.6 − 19.4 − 0.8 − 0.2 = 35.5. Cross-check on 9M FY26 actuals: 61.83 − 6.51 − 19.65 − 0.83 − 0.52 = 34.32%, which is exactly the reported 9M operating margin ($1,438.003m / $4,189.796m = 34.32%). The bridge reproduces the filed number before it is moved.

Above the trailing actual — stated, and justified

35.5% is +1.33pp above the TTM actual of 34.17% and +0.6pp above the latest quarter (34.90%). Per valuation.md rule 2, a terminal margin away from trailing requires a causal bridge, not a prohibition. The bridge is the mask mix shift: masks grew +15.5% against devices +8.8% in Q3 FY26 and +14.1% vs +8.7% over nine months, and masks carry the higher gross margin. Continuation of the observed mix drift alone delivers the 0.7pp of gross-margin expansion underwritten.

What the bridge would support if extrapolated: 36.0%. I underwrite 35.5% and state the 0.5pp haircut explicitly rather than applying it silently — it is for tariff exposure named in the 10-K's forward-looking-statements section and for the possibility that Philips' return brings device price pressure. That is a single, named, disclosed haircut, not stacked conservatism.

steady_state_check.py — what it flagged

python3 steady_state_check.py --ticker RMD --terminal-margin 0.355 --exit-multiple 14.0
  (asset sha256: 6fbbb0f9…)

Response, in the tool's own terms. Two things:

1. The tool compares against ANNUAL periods only, and RMD's newest annual is FY2025. The actual trailing figures are TTM 34.17% and Q3 FY26 34.90%. So the true gap is +0.6pp against the most recent observed quarter, not +2.8pp against a fourteen-month-old fiscal year. This is a granularity artefact of the check, and it is worth recording as a limitation of the instrument — on a June-FY name the annual series is up to four quarters stale.
2. The named mechanism is supplied and it is measurable, not narrative: consumable mix shift, with the quarterly device/mask growth differential in Research §1 and a segment gross-margin table. The check asks for exactly this and the answer exists.

The second flag is the more important one and the tool is right about it. A 14.0x exit against a 9.5x warranted multiple at 18.6% ROIC is a 1.48x premium. I do not dismiss it — see §4.

---

4. Exit multiple — derived, not asserted

EV_T/EBIT_T = (1−t)(1−g/ROIC)/(WACC−g). The exit multiple is not a free parameter.

• t = 21.1% — RMD's 9M FY2026 effective rate ($305.391m / $1,445.252m).
• ROIC = 25% — NOPAT TTM $1,892.5m × 0.789 = $1,493m; invested capital = equity $6,491.9m + debt $664.1m − cash & STI $1,666.0m = $5,490m → 27.2% measured. Stepped down to 25% at terminal. (steady_state_check.py measures 18.6% on a different, annual-basis capital definition; both are stated, and the sensitivity below spans the difference.)
• WACC = 9.0% — RMD's beta is 0.775 (AV OVERVIEW), the lowest in this three-name cluster.
• g = 4.0%.

(1−0.211) × (1 − 0.04/0.25) / (0.09 − 0.04)  =  0.789 × 0.84 / 0.05  =  13.3x

At WACC 8.5%: 14.75x. At the tool's 18.6% ROIC: 0.789 × 0.785 / 0.05 = 12.4x.

Base exit multiple adopted: 14.0x. It sits between the 13.3x identity result at 25% ROIC and the 14.75x at a 8.5% WACC, and it is below today's traded 15.33x — implied compression of 8.7%. Per criteria.md, the base may not sit below every stated anchor without argument; here it sits inside them.

Growth-matched comparator set — RMD's own, not shared with UFPT or USPH

These are three different businesses and must not share one multiple. RMD's comparator set is large-cap profitable medical devices with mid-to-high-single-digit exit-year growth:

n = 5 underwritten, growth range 9.0%–14.6%, dispersion 5.6pp. RMD's exit-year growth under the underwritten path is ~6% — which sits below the comparator range. Per the growth-matching rule this set does not bracket the subject at the exit year, so it is used as a sanity band only, and the exit multiple is taken from the derived identity above. Declaring that explicitly rather than using a median that the rule does not permit.

Consistency note versus the existing corpus: BSX's memo used a 16.0x exit for a 6% exit-growth business at 70% gross margin; RMD is underwritten at 14.0x for a similar exit growth at 62.5% gross margin and a lower measured ROIC. Deliberately more conservative than BSX, and the reason is stated. BSX's own comparator table lists RMD at 16.2x EV/EBIT — that figure came from the screen and is superseded by the 15.33x computed here on corrected net cash.

---

5. THE IMPLIED-PATH TEST — the Valuation Criteria

python3 reverse_dcf.py --spot 206.91 --shares 145.056 --net-cash 1001.9 \
  --revenue 5537.8 --years 5 --wacc 0.09 --terminal-margin 0.355 \
  --exit-multiple 14.0 --fcf-margin 0.3157 --hist-cagr 0.129
  (asset sha256: a46b1c2c…)

The --fcf-margin sign, and why omitting it would have mattered here

RMD's demonstrated TTM FCF margin is +31.57% — positive and very large. Per DATA_DEFECTS.md, positive FCF means omission overstates required CAGR. Measured directly, on this name:

That is larger than every previously measured instance in the corpus (STX 5.1pp, WDC 5.3pp, AVGO 3.56pp, META 2.61pp) and it is the single largest arithmetic difference in this memo. Omitting the flag would have turned a +9.2pp margin into +2.7pp — still a PASS, but a marginal one, and on a name one notch less cash-generative it would have flipped the sign. Direction: conservative here, but "conservative" is still "wrong", per the corrected entry in DATA_DEFECTS.md.

The tool's own bias table, printed when the flag is omitted, confirms the calibration: −30% FCF → 17.1%; −10% → 12.4%; +10% → 8.0%; +30% → 4.0%.

Sensitivity — over the exit multiple, not over scenario probabilities

The verdict does not flip anywhere in the plausible range. The flip point sits below 8x — a multiple lower than anything a 62% gross-margin, 34% operating-margin, low-beta compounder has traded at. That robustness is the result, and it is why the +9.2pp number should be trusted more than its precision suggests.

Terminal-value share

At the base path, terminal value is well above 60% of EV, so the reverse DCF is mandatory as the primary long-horizon output and a forward DCF would be supporting evidence only. Not built.

---

6. TWELVE-MONTH TARGET — $237, +14.8%

Built per valuation.md: near-term estimates × the name's own multiple history, percentile stated.

RMD's own EV/Sales history

Daily series 2020-07-27 → 2026-07-29 (n = 1,506 trading days), point-in-time TTM revenue lagged 45 days to filing, share count and corrected net cash held at current verified values so the series measures multiple movement.

RMD trades at the 7.4th percentile of its own six-year EV/Sales history and the 14.8th of its own three-year. This is the same shape as the BSX finding in this corpus (0th percentile of an eight-year EV/Sales history) — the sleep/medtech multiple has de-rated, hard, and the de-rating dates to the GLP-1 narrative. The 52-week range is $180.27–$291.57; spot sits 29.0% below the high.

The target

• NTM revenue $5,920m — taken from Street consensus, not from a house forecast. AV EARNINGS_ESTIMATES: FY2026 (Jun-26) revenue $5,650.6m / EPS $11.12; FY2027 revenue $5,967.8m (+5.6%) / EPS $12.08. The twelve months from here span FY2026 Q4 plus three quarters of FY2027, blending to ~$5,920m. Consensus FY2027 growth of 5.6% is materially below the 10.8% RMD just printed — the Street is already underwriting the GLP-1 deceleration, and I am not overriding it for the 12-month output.
• Revision direction, built in: trailing 30 days, 1 upward EPS revision and 0 downward on every horizon (FY2026, FY2027 and FQ1 FY2027). Thin coverage of revisions but unambiguously positive in sign.
• Multiple: 5.65x = the 25th percentile of the three-year window (current 14.8th). A partial mean-reversion within the recent regime — it does not assume the pre-GLP-1 multiple returns, and it is below the 3-year median of 6.57x and far below the 6-year median of 7.68x.
• Target EV = 5.65 × $5,920m = $33,448m; + net cash $1,001.9m = $34,450m; ÷ 145.056m shares.

Target: $237. +14.8% to spot.

The target is above spot, which valuation.md records as normal and expected (item B16: a process whose every target sits below spot is expressing a house view about the market, not valuing companies). It is driven entirely by the multiple percentile, not by out-forecasting the Street on revenue — at consensus revenue and today's 5.24x multiple the stock is worth $211.

Named events that move estimates inside 12 months, each dated in RMD_Catalyst_Calendar.md: Q4 FY2026 results and initial FY2027 guidance (early August 2026), the FY2026 10-K, and any oral-GLP-1 OSA regulatory action.

No external professional target is on file for RMD in this project, so the sanity-band check required by valuation.md cannot be performed. Recorded as unavailable, not as agreement.

The percentile is itself a finding

A 7.4th-percentile multiple on a business whose operating margin is at an all-time high and whose consumable line is accelerating is the definition of an expectations gap. The market is pricing the GLP-1 structural case; the reported numbers are not showing it. That gap is the trade, and it is why the 12-month and 5-year outputs agree on this name where they diverge on USPH.

---

7. Criteria scoring

The memo issues no position verdict. The book decides.

---

8. Downside case — named cause

Cause: oral GLP-1 approval for OSA solves the persistence problem, and new PAP patient starts turn negative. The 10-K names this exact escalation ("oral versions… may be approved for the treatment of OSA"), and management's own concession — 30–40% one-year GLP-1 persistence — is the thing an oral formulation is designed to fix. If new starts go to −5% and masks decelerate to +5% with RCS flat, consolidated growth falls to roughly 1%. At 12x that path prices RMD near $150, a −27.5% permanent impairment before any multiple overshoot.

Probability: 20% over five years. Not a going-concern case at any probability — net cash is +$1,001.9m, TTM FCF is $1,748.2m and the installed base amortises over years, not quarters. Logged to the ledger for Brier scoring.

Second-order risk, unpriced in the above: Philips' return from the consent decree. The 10-K says RMD "cannot predict the timing or nature of their substantial return." Not modelled as a step function, for the reason given in Research §3 — some of the vacated share has already gone to lower- priced entrants, so the reversal is not symmetric.