Commure
B2B enterprise software with a services-heavy delivery layer; annual subscription by module, reportedly from ~$1M of commitment (Sacra). RCM is paid out of the provider's collected revenue — the correct side of the board's payer test, and the seat Waystar occupies in the listed market. Sold to health systems (HCA, Tenet, Jefferson Health, Providence) and thousands of physician-owned practices, EHR-agnostic across 60+ integrations incl. Epic, Oracle Cerner and MEDITECH. Assembled by M&A: PatientKeeper (from HCA, 2021), Athelas merger (2023), Strongline, Augmedix ($139M, 2024), Memora Health (2024).
Earnings, margins, COGS & capex
A fast-scaled private roll-up whose disclosure is thinner than its mark. Hatched inside General Catalyst, Commure bought PatientKeeper from HCA in 2021, merged with Athelas in Oct 2023 at a $6B combined mark under Athelas co-founder Tanay Tandon, took listed scribe vendor Augmedix private for $139M and added Memora Health. It claims 500+ organisations, 3,000+ sites, tens of billions of dollars of annual payments and RCM that 'completes more than 85% of work without human intervention'. Then the facts stop: no revenue figure since Jun 2025, no margin data, no audited statement — and a mark that went $6B (Oct 2023) to $7B (May 2026), +16.7% over ~31 months, while ARR is claimed to have doubled twice. Read one way that is multiple discipline; read the other, outside capital would not pay more, since the round was ~1% dilution from the owner.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~52¢ is cost of goods and ~0¢ operating expense, leaving ~48¢ of operating profit.
Revenue trend
Margins
acquired-asset proxy: Augmedix 45.1% (FY2022) -> 48.0% (FY2023) -> 49.3% (Q4 2023)
growth spend plus integration of five acquired stacks; unverified
FY2024 40.65% -> FY2025 42.04% -> Q2 2026 42.76% -> 1H 2026 42.95% — operating leverage, none of it attributed to AI in any release
COGS structure
Two stacks, and the mix is the argument. Inference (ASR plus LLM per encounter, model calls per claim, code and appeal letter) is cloud opex that falls with token prices. Human delivery is the other half: the RCM and scribe businesses inherit labour in the loop, Augmedix's 48.0% FY2023 gross margin is the closest audited read on it, and the workforce is majority Finance-and-Operations and 54.8% offshore. The counter-claim is automation eating that labour — '>85% of work without human intervention' — whose one named external datapoint is Ob Hospitalist Group: charge-entry time down 83% within three months and 85%+ of charges AI-coded across ~2,000 clinicians and 200+ sites (3 Dec 2025). One self-reported customer, no definition of 'work', no unit-cost disclosure.
Capex
Immaterial as fixed assets; capital intensity sits in M&A and offshore delivery capacity, plus the $200M CVF repaid from a capped share of the customer cohort it funds.
Latest earnings
n/a
None formal. Directional: positive cash flow by end-2026 and an IPO filing 'the following year' (2027) with a pre-listing tender offer (SiliconANGLE, 19 Jun 2025).
- Last mark
- $7B on a $70M round, 19-20 May 2026 (~1% dilution)
- Prior mark
- $6B at the Athelas merger, Oct 2023 — +16.7% over ~31 months
- Reported revenue
- 'more than $200M' (Jun 2025); absolute ARR unsourced per canon
- Capital
- ~$550-570M equity + $200M non-dilutive CVF = $750-770M; the company's '$750M total' sits at the bottom of that band and conflates the two instruments, and neither reconciles Sacra's ~$823M
- Scale
- 500+ organisations, 3,000+ sites, 130+ health systems (Jun 2025)
- Headcount
- ~2,373, -4.3% YoY (Mar 2026, Revelio estimate); 54.8% India and Bangladesh
Growth drivers
- Administrative cost as the market — ~$1T/yr in the US on the company's framing, paid out of collections rather than a discretionary innovation budget
- HCA as anchor — ambient across 188 hospitals, ~2,400 sites and 99,000+ registered nurses (23 Oct 2024), enterprise agreement then still being negotiated; plus Tenet, Jefferson Health, Providence
- Cross-sell across coding, claims, denials, ambient, practice management and monitoring into 500+ organisations already contracted
- Autonomous coding as the hard-dollar wedge, with Ob Hospitalist Group the named proof point
- EHR-agnostic reach — 60+ integrations, MEDITECH ambient, a Vizient GPO contract inherited with Augmedix
- General Catalyst as distribution — CVF capital that buys customers without dilution, and HATCo's ownership of Summa Health (~$515M, 2025) as a live deployment site
Bull & bear
Commure is the one large private health-AI asset selling into the correct payer — the provider's own collected revenue — and it already sits inside 500+ organisations doing work that is a cost centre, not a discretionary pilot. If the >85% automation claim survives an audit, this is a software business with a services P&L in transition, marked at ~35x a fourteen-month-old trade-press revenue datum by an owner that took ~1% dilution because it did not need the money.
- Payer alignment is the whole game: RCM is paid from collections, which is why Waystar runs 42-43% adjusted-EBITDA margins at 108% net revenue retention while ambient vendors argue about clinician minutes
- Scale is not aspirational — 500+ organisations, 3,000+ sites, HCA (188 hospitals, ~2,400 sites) and Tenet, tens of billions of dollars of payments already flowing through
- The claimed cadence is extraordinary if true: ARR doubling three years running to 'hundreds of millions' from ~$105M at Mar 2024, with the mark rising far slower than the revenue — the opposite of the Abridge shape
- Autonomous coding has a named external proof point: charge entry down 83% in three months, 85%+ of charges AI-coded at Ob Hospitalist Group
- Breadth turns one contract into many across 60+ EHR integrations, reaching buyers Epic-native rivals cannot
- Breakeven guided for end-2026 with a 2027 filing intended: if the S-1 shows software gross margins this is the first buyable expression of AI-native provider operations, and canon calls it the highest-signal IPO on the slate
Every load-bearing number is company-stated, undated or unreconciled, and the one price that exists was set by the owner at ~1% dilution. The mark went $6B to $7B in 31 months while ARR supposedly doubled twice, the acquired ambient asset disclosed 48% gross margins, most of the workforce sits in operations rather than engineering, and the incumbent that owns the record has begun shipping the same coding automation inside a contract the customer already signed.
- The $7B is not a clearing price: $70M is ~1% dilution led by the sponsor that created the company — the shape canon flags on Sword's $40M/$4B, worth as much
- $6B to $7B is ~6%/yr against claimed ARR doubling; at ~35x a Jun-2025 trade-press revenue datum it is still ~10x the 3.35x market-cap-to-TTM-revenue of the listed operator in the same seat ($4.05B on $1.21B, 31 Jul 2026 close)
- Margin evidence points to services: Augmedix disclosed 48.0% gross margin on $44.9M before Commure paid $139M for it, and 55.7% of headcount is Finance and Operations vs 23.8% Engineering
- Epic monetises the same work directly — Penny at 200+ organisations now, autonomous coding from Nov 2026 — across 43.7% of acute hospitals, while charging AI vendors only ~$1,700-1,900/yr plus a ~$500/yr listing, so there is no toll economics to shelter behind
- Olive AI is the base rate for this strategy: a $4B mark, $852M raised, shut on 31 Oct 2023 with assets sold to Waystar, Humata Health and Availity — a health-automation roll-up can go to zero while its category thesis stays correct
- Related-party density is unusual even privately — one firm as founder, mark-setter, off-balance-sheet financier and owner of a deploying hospital system: not improper, but reason to require the filing before the price
What it is worth
Last-round mark plus a revenue multiple, cross-checked against the listed operator in the same seat; no market price exists. $7B on a $70M round (19-20 May 2026) is ~1% dilution. Against 'more than $200 million in annual revenue' (SiliconANGLE, 19 Jun 2025 — a single trade-press datum; canon CM-PRIV-CM-02 holds Commure's absolute ARR unsourced, so this is a reported figure rather than a canon-grade denominator) that is roughly 35x or less, with a numerator eleven months newer than its denominator — the basis mismatch canon stamps on Abridge. The prior mark was $6B at the Oct 2023 Athelas merger, so the mark is up ~16.7% (~6%/yr) over ~31 months. Comparable: Waystar at $4.05B market cap on $1.21B TTM revenue = 3.35x, at 42.76% Q2 2026 adjusted EBITDA and 108% NRR — so Commure's mark is 1.7x Waystar's whole market capitalisation on roughly a sixth of the revenue. Both multiples are computed by us.
Epic's Penny and AI Charting attach at renewal across the Epic base, coding specialists take the highest-value module elsewhere, and the S-1 (or its absence) reveals healthcare-services economics — 45-50% gross margins, real churn, revenue quality that will not support ~35x. A down-round or trade sale into the $3-4B range follows: roughly the listed comparable's market cap for a business with none of its profitability. The base rate is Olive AI — a $4B mark and $852M raised, ended 31 Oct 2023 with assets sold to Waystar, Humata Health and Availity.
No filing within twelve months, or one showing 50-60% gross margins on a services-heavy model. The $7B holds nominally because nobody re-prices it, growth decelerates, liquidity stays theoretical — canon's 'lean watch' with no position possible, and the mark ages the way Abridge's June-2025 print has.
The S-1 shows software gross margins (65%+) on 'hundreds of millions' of ARR still compounding near the claimed cadence, with the automation ratio defined and audited and breakeven achieved as guided. On Waystar-like quality but multiples of its growth, a US listing clears well above $7B and the May-2026 mark proves to have been the conservative print of the cycle.
A private mark is not a mark-to-market, and this one is weaker than most: the lead is the sponsor that created the company, the dilution is ~1%, the dilution is ~1%, and no company release exists for the round at all, so the $7B is carried on trade reporting that specifies neither pre- nor post-money. CM-COH-04's caveats apply wherever this $7B is summed with other private marks — dated Jun 2025 to May 2026, summed, against market caps at the 31 Jul 2026 close. The honest position is canon's: watch the filing, not the round, because an S-1 would disclose for the first time whether the revenue-cycle layer earns AI economics or merely healthcare economics, and it re-rates the listed comparable either way. There is no way to own this today; the only expression a US-first book can underwrite is a US listing or an acquisition by a listed party. Nothing here is a price target or a recommendation.
SWOT
Strengths
- On the correct side of the payer test — paid from the provider's collected revenue, the seat Waystar occupies at a $4.05B market cap (31 Jul 2026 close)
- Real enterprise scale for a private company: 500+ organisations, 3,000+ sites, HCA and Tenet as anchors
- Broadest AI surface in provider operations, so land-and-expand is real rather than aspirational
- General Catalyst is founder, controlling investor, growth financier and provider operator at once — procurement shortcuts where procurement is the bottleneck
- EHR-agnostic, reaching the ~56% of US acute-care hospitals Epic does not hold (43.1% of beds) plus the independent-practice tail
Weaknesses
- No audited financials, no revenue print since Jun 2025, absolute ARR unsourced in canon
- Margin structure unproven and probably services-shaped — 48.0% at the acquired ambient asset, majority-operations headcount
- The $7B is set by a ~1% dilution round led by the incumbent owner, only +16.7% above the Oct 2023 mark
- Capital reporting mixes $200M of non-dilutive customer-value financing with equity
- Integration risk across five acquired stacks, with headcount down 4.3% YoY against a doubling-growth narrative
Opportunities
- An IPO — canon calls its S-1 the most valuable document this board could receive: the first disclosure from an AI-native operator of whether the RCM layer earns AI economics or merely healthcare economics
- Autonomous coding and denial appeals, where the deliverable is collected dollars rather than saved minutes
- Consolidating a fragmented field — ~60 ambient vendors alone (PHTI, March 2025) plus a long tail of point RCM tools
- The community-hospital and independent-practice tail plus Athelas monitoring as a second engine
- Acquisition by a listed RCM or EHR party; the buyer set (Waystar, Optum, FinThrive, Oracle Health) is real
Threats
- Epic: 43.7% of US acute-care hospitals, 56.9% of beds, $6.7B of 2025 revenue (company-stated to Becker's 28 Apr 2026, unaudited — Epic files nothing, no margin knowable), now selling the AI itself — AI Charting general since 4 Feb 2026, Penny live at 200+ organisations for professional billing coding, autonomous coding from Nov 2026 (ED and radiology first)
- A consolidating agentic-RCM field — Waystar (#1 in Black Book's Q1 2026 agentic-AI RCM benchmark; Iodine at a reported $1.25B, 1 Oct 2025), R1's Palantir-built R37, New Mountain's Smarter Technologies, FinThrive, AKASA, CodaMetrix
- Ambient commoditisation beneath it — Doximity Scribe $0 to every verified US clinician, athenaAmbient at no additional cost, Epic AI Charting bundled at an undisclosed price
- Concentration and related-party density — HCA-scale logos dominate while one sponsor founded the company, set the mark, finances customer acquisition and owns a deployment site
- Autonomy ceilings set outside the company — payer acceptance of automated coding and appeals (42 CFR 422.566(d) keeps a physician in the denial loop) and a shifting prior-auth regime
Moats, dependencies & bottlenecks
Moats
structural Funded from money the provider collects, so the budget survives the cost cycles that kill productivity pilots.
High once claims flow through it Coding, submitting, appealing and reconciling claims is harder to rip out than a scribe; the exposure is displacement at renewal.
60+ integrations plus MEDITECH and a Vizient contract reach buyers Epic-dependent vendors cannot serve.
cuts both ways Non-dilutive customer-acquisition capital and a captive deployment site are real; they also mean growth is partly bought and the mark partly self-set.
Payer-behaviour signal improves denial prediction; erodes against clearinghouses and payers holding more of it.
Dependencies
Integration surface AND direct competitor 43.7% of US acute hospitals, 56.9% of beds, $6.7B 2025 revenue (company-stated to Becker's 28 Apr 2026, unaudited — Epic files nothing, no margin knowable), and a negligible vendor toll — not a tollbooth but a distribution owner now selling the AI itself.
controlling investor, CVF financier, HATCo operator) Capital + governance + channel in one counterparty High (structural) Sets the mark, funds customer acquisition off balance sheet, owns a provider that deploys the product — concentrated dependency if the thesis or fund cycle turns.
Revenue concentration HCA is customer, ambient anchor and the 2021 seller of PatientKeeper into Commure. No source discloses concentration; that absence is the risk.
Core technology + variable COGS Inference is the software half of COGS (NVIDIA is reported among prior investors); 54.8% of headcount sits in India and Bangladesh, so wage or PHI-scrutiny shocks land in gross margin.
Financing / liquidity Canon PB-024: eleven named candidates including Commure, none priced at 6 August 2026, no confirmed S-1.
Advantages
- The correct payer: revenue-cycle spend comes out of collections
- Enterprise scale unusual for a private company — 500+ organisations, 3,000+ sites, HCA and Tenet
- Broadest AI module set in provider operations; expansion needs no new procurement cycle
- EHR-agnostic reach into non-Epic systems, community hospitals and the practice tail
- Sponsor-supplied non-dilutive growth capital plus a captive deployment environment
- Hard-dollar ROI framing rather than contested clinician-time savings
Weaknesses
- Every headline figure is company-stated; canon holds absolute ARR unsourced and the raise unreconciled
- Gross margin undisclosed and, on the best proxies, services-shaped (~48%)
- The mark is self-set at ~1% dilution and only +16.7% above the Oct 2023 valuation
- ~35x a fourteen-month-old trade-press revenue datum vs 3.35x for the listed operator in the same seat
- Integration risk plus shrinking headcount against a doubling-growth narrative
- Ambient is the weak flank — absent from the named top six of a ~$600M 2025 US ambient category despite owning Augmedix
Bottlenecks
- Disclosure itself — the >85% automation claim is untestable beyond one self-reported customer
- Converting labour-in-the-loop delivery into software gross margin fast enough to justify the mark
- Integrating five acquired stacks while headcount shrinks
- Health-system security and PHI review on cross-border processing, which lengthens sales cycles
- Payer-side acceptance limits on automated coding, claims and appeals
- Selling against a record owner that can bundle the same function into an existing contract
Top signals & trends
Top signals
Top-tier names, but ~1% dilution from the incumbent owner is a negotiated headline; Morgan Stanley and Kirkland read as pre-filing relationship capital.
The multiple compressed hard — either the earlier mark was fantasy or the growth claim is not underwritten at face value.
A company approaching an IPO that stops quoting revenue is telling you something.
The only forward statement here that can be falsified; canon's instruction stands — watch the filing, not the round.
The record owner is competing in Commure's highest-value module, inside a relationship the system already pays for.
The only named external evidence for the automation claim — one self-reported customer, but concrete and dated.
Trends
~$1T of annual US administrative spend, paid from collections — what makes the RCM seat structurally better than the ambient seat.
Waystar+Iodine, R1's R37, New Mountain's Smarter Technologies — Commure can be consolidator or consolidated.
Epic AI Charting general since 4 Feb 2026, Penny at 200+ organisations, an agent Factory previewed at HIMSS26; Oracle Health pushing native AI in Cerner accounts.
Doximity Scribe $0 and athenaAmbient bundled, against a ladder from Freed $1,188/yr to Ambience $4,000-5,000/yr and a measured $2,008/yr point estimate of marginal E/M revenue ($167.37/clinician/month, 95% CI $86.52-248.21; JAMA 2026;335(16):1408-1417, exploratory, called a conservative lower bound).
PHTI's March 2025 assessment found burnout improves while financial ROI is unproven — its April 2026 follow-up reports one health system raising revenue by over $1,000 per provider per month, single-site and uncontrolled — and the coding channel is worth ~$0 outside fee-for-service — favouring vendors whose deliverable is collected dollars.
Eleven named candidates including Commure, none priced at 6 August 2026 — the mark stays unmarked and liquidity theoretical.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
GPU capacity under inference; reported among prior investors alongside Y Combinator.
Anthropic, Google) ASR and LLM reasoning for notes, code assignment and appeal drafting (OpenAI/Anthropic private).
Cloud capacity; General Catalyst's AWS partnership names Commure among co-deployed portfolio technologies.
Bangladesh) 54.8% of headcount (Revelio estimate) — the human half of COGS.
60+ EHR integrations, MEDITECH ambient, a Vizient contract inherited with Augmedix.
Ambient across 188 hospitals, ~2,400 sites and 99,000+ registered nurses (23 Oct 2024); also the 2021 seller of PatientKeeper to Commure.
Named user of the RCM and ambient platform (19 May 2026).
Named large-system customers (Sacra).
Autonomous coding across ~2,000 clinicians and 200+ sites (3 Dec 2025).
130+ health systems as of Jun 2025, plus thousands of physician-owned practices.
Sponsor-owned provider used as a deployment environment — a related-party channel.
The listed operator in the same seat: $4.05B market cap on $1.21B TTM revenue (31 Jul 2026 close), 42.76% Q2 2026 adjusted-EBITDA margin, 108% NRR, Iodine at a reported $1.25B, #1 in Black Book's Q1 2026 agentic-AI RCM benchmark — and the buyer of Olive AI's remains.
Private record owner across 43.7% of US acute hospitals; Penny at 200+ organisations, autonomous coding from Nov 2026, AI Charting general since 4 Feb 2026. Existential, and unbuyable.
Private since a ~$8.9B take-private completed late 2024 (TowerBrook/CD&R); the managed-services giant now layering an agentic platform on top — Commure's pitch at greater scale.
Owns the largest claims and clearinghouse rails plus provider RCM services; structural data advantage, offset by distrust of a payer-owned vendor.
SmarterDx, Access Healthcare and Thoughtful AI combined — the same AI-plus-offshore-delivery model with an explicit roll-up sponsor.
~33% of the ~$600M 2025 US ambient category with the deepest enterprise footprint and a sharp 2026 price cut — it dominates the flank Commure entered by buying Augmedix.
Point solutions and incumbents attacking Commure's highest-value module and setting the price umbrella beneath it.
Private (Bain/H&F); owns the practice-management and ambulatory-billing seat Commure targets in the long tail and bundles athenaAmbient at no additional cost.
The funded ambient cohort; Abridge alone holds ~30% of the 2025 US ambient category with the deepest Epic-native integration.
#2 EHR bundling native clinical and revenue-cycle AI into its own base — Epic's displacement risk at smaller scale.