PROCEPT BioRobotics [PRCT] — Financial Model Notes

As of 2026-07-29. Every figure below is traced to a primary source. Derived figures state their derivation.

1. Source hierarchy used

Not used: Alpha Vantage consensus (25/day shared quota exhausted at run time → Consensus Criteria INDETERMINATE, blocks nothing).

2. Quarterly revenue — the spine

RevenueFromContractWithCustomerExcludingAssessedTax, USD, 89–91-day periods. Q4 is never tagged separately; it is derived as FY − (Q1+Q2+Q3) and then cross-checked against the Q4 earnings release income statement, which prints it directly.

TTM to 2026-03-31 = 79,182 + 83,327 + 76,383 + 83,132 = $322,024k. Matches the screen to the dollar.

Q4'25 fell 8.3% sequentially from Q3'25 — a sequential decline in what is normally the seasonally strongest quarter for medical capital equipment. Cause is disclosed and specific (Research §2.2), not a mystery.

3. Revenue by type — assembled from quarters, because the 10-K omits it

The by-type disaggregation exists in every 10-Q and every earnings release. It does not exist in the FY2024 or FY2025 10-K — verified by enumerating the R-file inventory of both filings' FilingSummary.xml; neither contains a "Revenue" note or a "Disaggregation of Revenue" detail.

FY2025 vs FY2024 (Q4'25 release, "REVENUE BY TYPE AND GEOGRAPHY"), $k:

Q1'26 vs Q1'25 ($k): systems 27,239 vs 22,540 (+20.8%); handpieces/consumables 49,390 vs 42,488 (+16.2%); service 6,503 vs 4,134 (+57.3%).

Mix, Q1'26: systems 32.8%, consumables 59.4%, service 7.8%.

Modelling note. The consumable line decelerated from +60.0% (9M'25) to +16.2% (Q1'26) in two quarters. Any model that extrapolates the consumable line off FY2024–FY2025 will be modelling channel fill. The correct base for a forward consumable build is procedures × handpiece ASP, not historical consumable revenue growth:

• Q1'26 US procedures ≈ 12,200; handpieces ≈ 95% of procedures ≈ 11,590
• Q1'26 handpiece ASP ≈ $3,500 (disclosed; up 5% from Q4'25's $3,340)
• 11,590 × $3,500 ≈ $40.6m vs US handpieces-and-consumables of $43.0m — the ~$2.4m residual is "other consumables". The reconciliation holds.

FY26 handpiece revenue therefore ≈ US procedures × ~$3,500 × ~1.00. At the guided 39–48% procedure growth (60,200–64,100), that is $211–224m of US handpiece revenue. At the Q1 seasonal-share implied 56,700 procedures it is $198m. The spread between those two is $13–26m of FY26 revenue — roughly the entire distance between the guidance floor and a miss.

4. Unit metrics — disclosed, and the most valuable data in the filings

US install base and procedures, from the Q4'25 and Q1'26 release tables:

Annualised procedures per system: FY23 68.5 · FY24 67.6 · FY25 70.9. Flat.

Systems placed per year: FY23 148 · FY24 190 (+28%) · FY25 213 (+12%); Q1'26 47 vs Q1'25 42 (+12%).

System ASP: FY25 US system revenue $93.0m ÷ 213 = ~$437k; Q1'26 disclosed ~$485k for a new HYDROS. Q4'25 US system revenue was flat YoY ($27,600k vs $27,636k) on more units (65 vs 60) — an ~8% ASP decline in that quarter, before the Q1'26 recovery.

5. Margins and operating leverage

FY26 operating margin is derived, not guided: revenue $400m × 65% GM = $260m gross profit, minus $350m opex = −$90m operating loss = −22.5%.

The operating-leverage question in one line. FY25 delivered +9.3pp of operating-margin expansion because revenue grew 37.2% against opex growth of 28.4%. FY26 guidance needs +11.2pp from revenue +30% against opex +16.6%. Q1'26 delivered +0.5pp, because revenue grew 20.2% against opex growth of 20.9% — opex grew faster than revenue. The FY26 margin step is entirely back-loaded and entirely unevidenced so far.

Q4'25 gross margin was 60.6%, hit by (a) lower US consumable revenue in the destock and (b) a $1.5m one-time voluntary field action = 240bp (handpiece/system compatibility; no patient-safety issue; resolved by field upgrade). Both non-recurring; Q1'26 recovered to 64.9%.

6. Cash flow, balance sheet and the SBC question

SBC is 15.5% of FY2025 revenue and is guided to $59m in FY2026 — and adjusted EBITDA excludes all of it. The gap between the FY26 adjusted-EBITDA guide (−$17 to −$30m) and the FY26 GAAP net-loss guide (−$78.5 to −$91.5m) is almost entirely SBC plus D&A. This is the Twist pattern named in the brief: a non-GAAP headline that improves while the GAAP loss does not. It is disclosed, reconciled in the release, and not hidden — but a model built on adjusted EBITDA will overstate this company's economics by ~$60m a year.

Balance-sheet ratios are stable — the problem was never on the balance sheet.

Neither ratio deteriorated. The channel fill sat at the customer, not in PRCT's receivables — which is precisely why it was invisible in the financial statements and why it took a management disclosure, not a ratio, to surface it. This is the memo's methodological finding: an AAOI-style DSO screen would have found nothing here. The detection route that worked was units versus revenue — procedures against handpieces sold — which requires the operating-metric table, not the financial statements.

Accruals (retained on both archetypes per criteria.md): TTM net income −$102.5m vs TTM operating cash flow −$70.1m → accruals of −$32.3m = −6.6% of total assets. Negative (conservative). No earnings-quality flag on the Sloan measure.

Customer concentration: the Q1'26 10-Q's "Segment, Geographical, and Customer Concentration" note discloses geographic concentration only (US 87% / OUS 13%) and no single customer above the 10% disclosure threshold. There is no AAOI-style stocking distributor. The 10-K notes sales to "hospitals, or end customers, and to a lesser extent, distribution partners, ambulatory surgery centres and leasing companies", unquantified.

Sales-type leases: PRCT does place systems under sales-type leases. The amounts are immaterial — SalesTypeLeaseRevenue of $3.0m in FY2025 (1.0% of revenue) and $0 in Q1'26; net investment in lease $4.8m at 2026-03-31. This is not a channel for pulling forward system revenue at any meaningful scale, and it was checked specifically because it could have been.

7. Debt, dilution and capital structure

8. Model assumptions used in the valuation, and their basis

9. Defects found while building this model

Recorded because the framework's premise is that defects compound into future memos.

1. companyfacts filing dates are contaminated by restatements. SEC retains only the latest filing that restated a prior fiscal year, so a Q4 value derived from a 10-K inherits that (much later) filing date. Using it as the "publicly known from" date in a point-in-time multiple series pushed every TTM window containing a Q4 years forward and produced EV/Sales medians of 50–120x — confidently wrong, silently. Fix: fall back to the actual reporting lag whenever the recorded filing date exceeds 100 days after period end.
2. A four-quarter TTM window spans ~274 days between first and last period end, not ~365. A validity check written as 330 ≤ span ≤ 400 rejected every valid window and admitted the one invalid window that skipped a quarter. Both bugs produced plausible-looking output.
3. The annual report omits the revenue disaggregation the quarterly reports contain. A pipeline that parses only 10-Ks would never see the system/consumable split for this name.
4. The metric that mattered most is not in XBRL at all. Procedures, handpieces sold, install base and handpiece ASP live only in the 8-K Exhibit 99.1 tables. No XBRL-driven screen can see the utilisation flatness or the handpiece:procedure gap that is the entire finding of this memo. This is the strongest argument in the record for why Tier 2 exists.