Phase Space AI

Valuation

DexCom [DXCM]

DexCom [DXCM] — Valuation

Spot $75.14 · 2026-07-29 · Shares 393.6m · Net CASH $1,030m · EV $28,545m · EV/Sales 5.92x

---

COMPANY STATE: A — mature and structurally stable (evidence grade B)

The brief flags DXCM as "arguably A or C." Both were tested; A is declared, and the reason C was rejected is recorded so the choice is auditable.

State A test: - Profitable: GAAP operating income positive in every year since 2019 (8 consecutive years), TTM GAAP operating margin 21.45%. - Operating margin low-variance across ≥5 years: this limb is NOT cleanly met. GAAP EBIT margin over FY2021–TTM: 10.86% / 13.44% / 16.50% / 14.88% / 19.56% / 21.45% — a 10.6pp band. Hence evidence grade B, not A. - No structural regime change: no business-model transition, no accounting-basis break, no transformative acquisition. The 2024–25 margin trough was a regime repair, not a regime change — an identified, itemised set of one-timers (325bp: scrap, freight, receiver recall, tallied independently by an analyst in Q4 2025 Q&A) layered on a rebate reset, all of which have since reversed. Gross margin is now +6.10pp YoY and above the pre-disruption FY2023 level on a non-GAAP basis.

Why not State C. C requires "pre-profit or thin-margin." DXCM earns a 21.45% GAAP operating margin, 36.0% adjusted EBITDA margin and generates positive free cash flow with $1,030m of net cash. It is not thin-margin on any reading. Misclassifying it C would licence an expectations-surface treatment that overstates the uncertainty in a business that publishes and beats formal margin guidance.

Consequence: fix the terminal margin from DXCM's own economics and stated trajectory; solve for price-implied growth; show the exit-multiple surface. Critically — the 5-year average margin is NOT the right anchor here. Averaging across a documented one-off trough would embed the trough into the terminal year, which is the §9.8 defect ("terminal margin capped below what the company demonstrably earns") arriving by a different route. The anchor is the current and forward-guided level.

---

Terminal margin: 27.0% GAAP EBIT

The opex bridge (terminal year FY2031, GAAP basis)

  m_gross,T                              66.0%
  − R&D                                 (11.0%)
  − SG&A                                (27.0%)
  − other / collaborative                (1.0%)
  ─────────────────────────────────────────────
  = m_EBIT,T                             27.0%

Hard constraint: m_EBIT,T 27.0% ≤ m_gross,T 66.0% — SATISFIED, 39pp of headroom.

Justification

• m_gross,T = 66.0%, set ABOVE the trailing actual of 61.84%. This is the one deliberately unprecedented assumption in the memo and it is flagged per the brief rather than clamped. The evidence:
• Q1 2026 GAAP gross margin 62.95%, non-GAAP 63.5%, versus 57.5% a year earlier — +600bp delivered in four quarters.
• FY2026 non-GAAP guidance 63–64%, deliberately not raised despite Q1 tracking ahead, for reasons management named as external ("fuel prices and shipping routes"), not operational.
• The mechanism is arithmetic, not narrative: the G7 15-day sensor is the same manufactured article amortised over 50% more wear days, and management describes the Q1 contribution as "initial benefit from the switchover." Rebate treatment is confirmed unchanged ("100% of all sensors… covered with no change in rebate trends"), so the wear-time gain is not given back in price.
• DXCM's own history reached 68.63% (FY2021) and 69.25% (FY2015), so 66.0% is below its own demonstrated maximum. The figure is unprecedented only against the last three years.
• Freight and scrap — 325bp of 2025 one-timers — are non-recurring by identification.
• R&D 11.0% vs TTM 12.4% ($599m/$4,818m). Q1 2026 R&D grew 2.7% against 15% revenue growth. Modest leverage assumed.
• SG&A 27.0% vs TTM 27.6% ($1,331m/$4,818m). Total Q1 2026 opex grew 8.8% against 15% revenue growth — a 6.2pp differential. Assuming only 60bp of SG&A leverage over five years is well inside what one quarter already demonstrated.

The mandated sanity test

Terminal margin 27.0% vs trailing actual 21.45% — ABOVE, by +5.55pp. Not the §9.8 defect. Cross-check against the company's own guidance: FY2026 non-GAAP operating margin is guided to 23–23.5%, and DXCM's GAAP-to-non-GAAP gap is only ~80bp, so guided FY2026 GAAP is ~22.3%. Getting to 27.0% by FY2031 requires +94bp per year. DXCM delivered +851bp in the single year to Q1 2026. The assumption is roughly one-ninth of the demonstrated annual rate.

---

Exit multiple: 17.0x EV/EBIT — growth-matched, two anchors

Exit-year growth

FY2026 guided +11–13%; terminal-year (FY2031) growth underwritten at ~10% (type-2 penetration maturing, CMS assumed resolved one way or the other by then).

Comparator set — growth-matched, recorded in analysis.json

Validity asserted: n=5 mature profitable device platforms; growth spans 8.8%–14.6% and brackets the underwritten 10% exit growth; dispersion 5.8pp, so the set carries information. PEN excluded, and the reason matters: BSX has agreed to acquire Penumbra, so PEN's 65.5x is a deal price, not a trading multiple. Including it would have lifted the median to 23.9x and flattered DXCM by ~2pp of required CAGR. Also excluded: PODD/GMED/TMDX/ISRG (20–32% growth, outside the bracket), BDX/LNTH/TFX/ICUI (≤2.4% growth, outside), BAX/IRTC (negative EBIT), MASI/HOLX (no AV statements).

Second anchor — the warranted-multiple identity

EV_T/EBIT_T = (1−t)(1−g/ROIC)/(WACC−g). DXCM NOPAT = $1,033m × 0.80 = $826m; invested capital ≈ equity less net cash ≈ $3.0bn → ROIC ~27.5%, a genuinely high-return business (contrast BSX at 10.2%). At t=0.20, g=5.0%, WACC=9.5%: 0.80 × (1 − 0.05/0.275) / 0.045 = 14.5x.

Base multiple selection

17.0x sits above the identity anchor and below the comparator anchor. It does not sit below every anchor. Implied compression from today's 27.6x: −38.4% — a large compression, and appropriate: per valuation.md rule 3, a name growing 16% today becomes a 10% grower by the exit year and should not hold a 16%-grower's multiple.

---

THE IMPLIED-PATH TEST — Valuation Criteria

reverse_dcf.py --spot 75.14 --shares 393.6 --net-cash 1030 --revenue 4818 --years 5 --wacc 0.095 --terminal-margin 0.270 --exit-multiple 17.0 --solve cagr

Solved for: revenue CAGR. Held fixed: terminal margin 27.0%, exit multiple 17.0x, WACC 9.5%, horizon 5y, net cash $1,030m, shares 393.6m, TTM revenue $4,818m.

THE MARKET REQUIRES: 15.24% revenue CAGR

Demonstrated (3-year): 17.01%. MARGIN: +1.77pp.

Sensitivity over the exit multiple

Flip point against the 17.01% demonstrated CAGR: 15.6x. At the comparator median of 20.6x the price requires only 10.9% — inside guidance.

VERDICT: PASS WITH ARGUMENT

The margin is positive against demonstrated history (+1.77pp) but the honest disclosure is this: required 15.24% is above FY2026's guided 11–13%, by 2.2–4.2pp. The price requires DXCM to grow faster over five years than it is guiding for the current year. That is not a PASS on demonstrated grounds alone.

The argument is specific, evidenced and named — not narrative:

1. Type-2 non-insulin reimbursement is a step function already in progress. >6m non-insulin lives covered across the three largest PBMs today; Prime Therapeutics from summer 2026 (disclosed); company target of >7m by end-2026; ADA Standards of Care already upgraded to recommend CGM for all people with diabetes. Q1 2026 already delivered "strong first-quarter share gains, with the biggest increase coming from people with type 2 diabetes who are not on insulin."
2. CMS is the largest single unlock and it is UNDATED. Roughly half of the type-2 non-insulin population sits in Medicare by the company's own statement. Management says only "a matter of time." This is the weak link in the argument and it is stated as such. The type-2 non-insulin RCT reads out at ADA 2026 Scientific Sessions and is described as the cornerstone evidence for any payer awaiting RCT-level data — that event is dated; the CMS decision is not.
3. International is compounding independently — +26% reported / +17% organic in Q1 2026, on newly opened access in France and Canada.
4. The 15-day transition raises revenue per patient-year and gross margin simultaneously, with rebate neutrality confirmed.

What is deliberately NOT in the argument: Stelo and the non-diabetic/OTC opportunity are carried at zero. DXCM discloses no Stelo revenue, units or subscribers in any of 13 quarters, and Stelo Q&A mentions have gone from 70 to 0. Unevidenced upside is not underwritten — if it converts, it is free.

Comparison with the screen

Screen: required 11.8% vs demonstrated 17.0% = +5.2pp. This memo: required 15.24%, margin +1.77pp — a 3.4pp overstatement in the screen. The whole difference is the exit multiple: the screen appears to have used ~20x (which returns 11.55%) against this memo's two-anchor-bracketed 17.0x. Net cash and TTM revenue both reconciled cleanly for DXCM, and av_vs_edgar.py returned 0 disagreements of 6 compared with trust=yes — DXCM was the only clean name of the three.

---

12-MONTH TARGET: $96.00 (+27.8%)

1. Near-term base. FY2026 guided revenue $5.16–5.20bn. NTM revenue (Q3 2026–Q2 2027) ≈ $5,350m.

2. Multiple anchored on DXCM's OWN trading history, with the growth-matched sub-window used.

DXCM's own quarterly EV/Sales history (n=32, 2018-06-30 → 2026-03-31):

Current 5.92x = the 9th percentile of DXCM's own eight-year history.

The 8-year median of 13.51x is NOT admissible — it was formed when DXCM grew 20–44%, and would imply a $168 target (+124%) by pricing a growth rate the company no longer has. That is the regime-change trap valuation.md names.

Anchor adopted: 6.91x, the trailing-8-quarter median (Q2 2024 – Q1 2026). That window is genuinely growth-matched: DXCM grew 2.0% to 21.6% YoY, averaging ~14%, across it — the same regime as today's 11–15%. 6.91x is the 12th percentile of the 8-year history.

3. Product-cycle events inside 12 months: Prime Therapeutics coverage live (summer 2026); type-2 non-insulin RCT full readout (ADA 2026 Scientific Sessions); Stelo redesign ship; new adhesive/patch to market; >7m non-insulin lives covered (end-2026); Smart Basal beyond pilot. CMS: undated, not modelled.

4. Target. (6.91 × 5,350 + 1,030) / 393.6 = $96.57. Rounded to $96.00, +27.8% to spot. Target multiple 6.91x = 12th percentile of own 8-year history.

Sanity band. A +27.8% target that requires only a return to the median of the last two years — not to the median of the last eight — on a business whose gross margin has expanded 610bp and whose operating margin has expanded 851bp over that same period. If instead the 8-year p25 (9.69x) were reached, the target is $121 (+61%); that is not claimed.

---

Criteria summary