Phase Space AI

Valuation

AtriCure [ATRC]

AtriCure [ATRC] — Valuation

Spot $40.89 · 2026-07-29 · Shares 50.9m · Net CASH $107m · EV $1,974m · EV/Sales 3.46x

---

COMPANY STATE: C — scaling but economically observable (evidence grade B)

Test and evidence: - Thin-margin: GAAP operating margin 2.44% TTM, and negative in FY2022–FY2025. Qualifies. - But economically observable — all five markers present: 1. Positive and high gross margin: 76.16% TTM, in a 71.6%–76.2% band for ten years. Never a question mark. 2. Identifiable contribution margin: consumables into scheduled cardiac surgery. 3. Visible expense scaling: SG&A guided "well below top-line growth"; R&D "leveling off" with LeAAPS enrollment complete. Q2 2026 opex ex-milestone +6.1% against revenue +12.8% — a 6.7pp differential, observed not projected. 4. A formally guided model: FY2026 revenue $600–610m, adjusted EBITDA raised to $85–89m, GAAP EPS $0.00–0.04, plus an Analyst-Day long-range plan management says it is "significantly ahead of." 5. Positive GAAP net income ($9m in Q2 2026) — the inflection is realised, not forecast.

Why not State D. D requires negligible revenue, unstable or deeply negative gross margin, or a reverse DCF returning no solution in range. ATRC has $570m of revenue, a 76% gross margin, and the reverse DCF solves cleanly at 18.91%. The instrument applies.

Why not State A. A requires low-variance operating margins across ≥5 years. ATRC's GAAP EBIT margin ran −12.92% to +2.44% over FY2022–TTM, crossing zero inside the window. Not A.

Consequence: the terminal margin must be BUILT through the opex bridge, never assumed, and the output is a two-dimensional expectations surface rather than a single solved parameter. Evidence grade B — the guidance is formal and has been beaten, but there is only one quarter of GAAP profitability on the record.

---

Terminal margin: 16.0% GAAP EBIT — built, with the haircut declared

The bridge, run first without judgement (terminal year FY2031, revenue ~$1.0bn at 12% CAGR)

TTM actuals: gross margin 76.16%, R&D 16.89% ($96.2m), SG&A 56.81% ($323.6m) → EBIT 2.46%. Ties exactly.

Rolling forward at the guided differentials — revenue +12%/yr, R&D +6%/yr, SG&A +7.5%/yr:

  m_gross,T                                       77.0%
  − R&D    16.89% × (1.06/1.12)^5  =             (12.9%)
  − SG&A   56.81% × (1.075/1.12)^5 =             (46.2%)
  ─────────────────────────────────────────────────────
  = m_EBIT,T  (unadjusted bridge output)          17.9%

Hard constraint: m_EBIT,T ≤ m_gross,T → 17.9% ≤ 77.0% — SATISFIED, 59pp of headroom. There is no danger here of the recorded breach where one name carried a 14.4% terminal margin against an 11.9% gross margin; ATRC has 61pp of gross margin to spend on operating cost.

The adopted figure: 16.0%, and why it is lower than the bridge

Per valuation.md rule 1 — "assert, never clamp, and persist the provenance" — the raw bridge output (17.9%), the transformation (−1.9pp), the reason, and the source are all recorded rather than applied silently:

Reason: the bridge assumes five uninterrupted years of the guided opex differential, and management has already told us it intends to break it on good news. Q2 2026 Q&A, on salesforce expansion "assuming that BoxX-NoAF is positive on that 30-day endpoint and that you would be hopefully launching"; and Q1 2026 Q&A, Wirick: "the accelerated pace of enrollment on BoxX-NoAF brings incremental costs — pulling enrollment in by a year into 2026 — that is incremental to our plan." ATRC has a live, stated incentive to reinvest a positive readout into commercial capacity, which is the correct use of capital and a drag on the terminal margin. This is valuation.md rule 2's "explicit causal bridge" for a below-bridge assumption — deliberate reinvestment at attractive ROIC — not an unargued haircut.

Adopted: m_EBIT,T = 16.0%.

Justification of each term

• m_gross,T = 77.0% vs TTM 76.16%. Above trailing by 0.8pp only. Management guides "modest improvement in full year 2026 gross margin over full year 2025", and Q1 2026's 77.39% was described as "exceptional as a result of extremely favorable mix" — so the trailing figure is flattered and 77.0% is appropriately restrained. Ten-year maximum 76.2%; this is barely unprecedented.
• R&D 12.9% vs TTM 16.89%. LeAAPS enrollment is complete (July 2025) and BoxX-NoAF completes around end-2026, after which both large trials are in follow-up rather than enrollment. Wirick, Q4 2025: "with the conclusion of enrollment in LeAAPS, we are seeing a step down in those clinical trial costs."
• SG&A 46.2% vs TTM 56.81%. This is the biggest single lever and the most contestable number in the memo. It requires SG&A to grow at 7.5% against 12% revenue for five years. Demonstrated: Q2 2026 total opex ex-milestone grew 6.1% against 12.8% revenue. The assumption is less aggressive than one delivered quarter, but sustaining it for twenty consecutive quarters is the real question — which is exactly why the 1.9pp haircut is taken here and not in the gross-margin line.

The mandated sanity test

Terminal margin 16.0% vs trailing actual 2.44% — ABOVE, by +13.56pp. Not the §9.8 defect. It is a large step-up, and the honest framing is that it is a scaling assumption: it takes ATRC from a 14% adjusted EBITDA margin (FY2026 guided) to roughly a 25% adjusted EBITDA margin at terminal, against 76% gross margins. Mature high-gross-margin SMID medtech comparators in the set earn GAAP EBIT margins of 9.0% (NVST), 17.6% (GMED), 18.8% (LNTH), 28.1% (EW) and 34.3% (RMD) — 16.0% sits at roughly the median of that band, which is the correct sanity check for a business that has just crossed into profitability.

---

Exit multiple: 15.0x EV/EBIT — growth-matched, two anchors

Exit-year growth

FY2026 guided +12–14%; terminal-year (FY2031) growth underwritten at ~11%.

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

Validity asserted: n=6, all profitable at the EBIT line; all SMID/mid-cap single-or-few-franchise medical device businesses with 55–78% gross margins (a matching operating model, not a matching SIC code); growth spans 0.6%–30.2%, comfortably bracketing the underwritten 11% exit growth, with 29.6pp of dispersion. This set carries real information — unlike the recorded 29.9x/29.9x/30.4x anchor. Excluded: PEN (65.5x is a deal price — BSX is acquiring it), PRCT and IRTC (negative EBIT), BAX (negative EBIT), TFX/ICUI (declining revenue and 45.7x/65.4x multiples on collapsed margins — the denominator, not the numerator, is doing the work), the mega-caps (BSX/MDT/ABT/SYK/ZBH/BDX — wrong scale and operating model), MASI/HOLX (no AV statements).

ATRC's own 137.5x EV/EBIT is meaningless — it is 3.46x EV/Sales divided by a 2.44% margin. This is precisely the situation valuation.md flags for low-margin businesses: an EV/EBIT multiple on a 2.4%-operating-margin base is close to the wrong instrument. The compression figure is therefore quoted on EV/Sales, where the comparison is well-defined: terminal EV/Sales = 15.0x × 16.0% = 2.40x, against today's 3.46x — an implied compression of −30.6%.

Second anchor — the warranted-multiple identity

EV_T/EBIT_T = (1−t)(1−g/ROIC)/(WACC−g). Today's ROIC (3.9%) is pre-inflection and uninformative, so the identity is run on the terminal economics: at a 16.0% EBIT margin on ~$1.0bn of revenue, NOPAT ≈ $126m on invested capital of roughly $500m → terminal ROIC ~25%. At t=0.21, g=5.0%, WACC=10.5%: 0.79 × (1 − 0.05/0.25) / 0.055 = 11.5x.

Base multiple selection

15.0x sits above the identity anchor and well below the comparator anchor. It does not sit below every stated anchor. WACC is set at 10.5% — the highest of the three names, reflecting a $2.1bn market cap and 47.6% realised volatility.

---

THE IMPLIED-PATH TEST — Valuation Criteria

reverse_dcf.py --spot 40.89 --shares 50.9 --net-cash 107 --revenue 570 --years 5 --wacc 0.105 --terminal-margin 0.160 --exit-multiple 15.0 --solve cagr

Solved for: revenue CAGR. Held fixed: terminal margin 16.0%, exit multiple 15.0x, WACC 10.5%, horizon 5y, net cash $107m, shares 50.9m, TTM revenue $570m.

THE MARKET REQUIRES: 18.91% revenue CAGR

Demonstrated (3-year): 17.40%. MARGIN: −1.51pp.

Sensitivity over the exit multiple

Flip point against the 17.40% demonstrated CAGR: 16.0x. One turn above the base.

VERDICT: FAIL — and the FAIL is declared non-robust

The required 18.91% exceeds the demonstrated 17.40% by 1.51pp, and — the more damaging comparison — exceeds FY2026 guidance of 12–14% by 4.9–6.9pp. The price requires ATRC to grow at roughly 1.4x its own guided rate for five years, while growth has decelerated in each of the last three periods (16.6% → 14.8% → 13.9%). Per criteria.md, PASS WITH ARGUMENT requires "a specific, evidenced reason (a named product cycle, mix shift, pricing action)" for the price to require more than demonstrated. The candidate arguments — LeAAPS and BoxX-NoAF market expansion — are genuinely specific and genuinely evidenced, but LeAAPS has no disclosed readout date at all, so the expansion cannot be placed inside the five-year window with any confidence. An argument that cannot be dated is not yet an argument. FAIL.

The disclosure that matters more than the verdict

This FAIL inverts on a 1.0x change in the exit multiple — from 15.0x to 16.0x — and the growth-matched comparator median is 21.9x, which is 6.9x above the flip point. At the comparator median the price requires only ~9.5%, below guidance, a comfortable PASS.

So the verdict is entirely multiple-determined, and stating that plainly is the point. criteria.md records the NTRA failure as running the sensitivity "on the parameter that could not change the answer" while conceding the flip point "has to be sought on the exit multiple, which is where the real judgement lives." Here the sensitivity is run on exactly the parameter that determines the answer, and the honest summary is:

ATRC fails at a conservatively-anchored 15.0x and passes at anything from 16.0x upward. The base sits 6.9x below the growth-matched traded median. A reader who believes 21.9x is the right anchor should read this as a PASS with a +7.9pp margin. The memo's own judgement is that a business with one quarter of GAAP profitability does not yet earn its cohort's median multiple, so the base is 15.0x and the verdict is FAIL — but the verdict carries less information than the sensitivity table above it.

Comparison with the screen

Screen: required 6.6% vs demonstrated 17.4% = +10.8pp. This memo: required 18.91%, margin −1.51pp — a 12.3pp overstatement, the largest of the three names. Decomposition: the screen must have combined a materially higher exit multiple (~24x returns 8.24%) with a higher terminal margin. Note the screen's demonstrated 17.40% reconciles exactly to the FY2022→FY2025 reported revenue CAGR, so the demonstrated side was computed correctly; the entire error is on the required side.

The two-dimensional surface (required by State C)

State C requires all three of: required margin at underwritten growth, required growth at underwritten margin, and the region clearing the hurdle. Reporting only one solved parameter overstates precision because growth and margin are economically correlated here — a faster BoxX-driven ramp would come with higher S&M.

Note the corner: at the unadjusted bridge output of 17.9% the required CAGR is ~16.7%, below the 17.40% demonstrated — i.e. the 1.9pp terminal-margin haircut is itself sufficient to flip this name from PASS to FAIL. That is exactly the ALAB situation recorded in known-silent-failures.md §9.8, where a margin adjustment alone moved the verdict. It is disclosed here rather than buried, and the haircut is retained because its causal reason (reinvestment of a positive readout into salesforce) is real.

---

12-MONTH TARGET: $42.00 (+2.7%)

1. Near-term base. FY2026 guided revenue $600–610m; FY2027 at ~12% on $605m ≈ $678m. NTM revenue (Q3 2026–Q2 2027) ≈ $645m.

2. Multiple anchored on ATRC's OWN trading history.

Current 3.46x = the 22nd percentile of ATRC's own eight-year history.

The 8-year median of 6.79x is not admissible — it was formed in 2020–2022 when ATRC was a 20%+ grower pre-profitability, a different regime in both dimensions.

Anchor adopted: 3.30x, the trailing-12-quarter median (19th percentile of the 8-year history). The trailing-8-quarter median (3.11x) is the more growth-matched window but the reason for going one notch above it is named and dated: BoxX-NoAF's H1 2027 readout is now a disclosed catalyst inside the 12-month horizon, and the FY2026 adjusted EBITDA guide was raised (to $85–89m from $80–82m) after that window closed. Both are new information relative to the 8-quarter average.

3. Product-cycle events inside 12 months: BoxX-NoAF full enrollment (around end-2026, disclosed); BoxX-NoAF data readout H1 2027 (disclosed); FDA IDE submission (2026); new product release (2027); STS post-op AF quality measure enforcement (2027). LeAAPS readout: undated, not modelled.

4. Target. (3.30 × 645 + 107) / 51.5 = $43.40. Trimmed to $42.00, +2.7% to spot. Target multiple ~3.20x = ~18th percentile of own 8-year history.

Sanity band — and this is the honest headline on ATRC. The 12-month target is essentially flat to spot, and that is the finding rather than a failure of the method. ATRC has already re-rated: $27.70 (May 2026) → $40.89 (July 2026), +47.6% in roughly two months, on the Q2 beat and the raised adjusted-EBITDA guide. Of the three names in this run it is the only one that is not at a multiple trough, and the +2.7% target says the easy re-rating has been taken. The remaining upside is the BoxX-NoAF readout, which is a binary in H1 2027 — a catalyst trade, not a valuation trade. valuation.md warns that a process whose every target sits below spot is expressing a house view; this target is marginally above spot, and it is low because the price moved, not because the model is bearish.

---

Criteria summary