Abridge
B2B enterprise SaaS. Per-clinician annual subscription (~$2,500/clinician/yr — an estimate corroborated by Sacra; Abridge publishes no pricing) sold directly to health systems and academic medical centers; deep native integration ('Abridge Inside') with Epic; expanding from documentation into revenue-cycle coding and inpatient workflows ('care intelligence').
Valuation ~3.2x from $0.85B (Feb 2024) to $2.75B (Feb 2025), then ~2x again to $5.3B (Jun 2025) in ~4 months; the reported Apr 2026 extension held the $5.3B mark rather than marking up. Series C valuation and the Apr 2026 extension are reported figures, not company-confirmed audited data.
The thesis on this name
State of AI for Healthcare
PRIVATE — the watchlist's explicit PASS, and the sharpest price-against-measurement test on the board. Abridge leads the category on every operational measure: a $5.3B post-money mark set by a $300M Series E led by a16z with Khosla in June 2025 (fourteen months old at this vintage), a further ~$316M Series E extension reported in April 2026 at a price no source discloses, ~$1.07B raised — a sum this board computes from Abridge's own round announcements plus that reported extension, not a figure any source states — ~$100–117M of contracted ARR on a Q1-2025 basis, 150+ enterprise health systems at the Series E and 250+ reported by mid-2026, roughly 30% of a ~$600M US ambient-scribe category in 2025, Best in KLAS 2026 ambient at 94.7, and an NVIDIA partnership announced 12 Jun 2026 to co-develop a clinical-conversation foundation model on Nemotron. What makes it a pass is the measurement. JAMA (Rotenstein et al., online 1 Apr 2026; 2026;335(16):1408-1417) studied 8,581 ambulatory clinicians — 1,809 adopters against 6,772 concurrent non-adopters — across five academic health systems from Jun 2023 to Aug 2025, and put marginal E/M revenue at $167.37 per adopting clinician per month (95% CI $86.52–$248.21): $2,008/yr at the point estimate, $1,038–$2,979/yr across the interval. That figure is EXPLORATORY, not one of the four prespecified main outcomes (total EHR time −13.4 min and documentation time −16.0 min per 8 scheduled patient hours, no significant change in after-hours EHR time, +0.49 visits per week); the authors leave the mechanism open between extra visits booked in nontemplated time and a level-4/level-5 coding-composition shift; and they call the estimate a conservative LOWER bound that cannot generalise to cost-benefit considerations. Against a ~$2,500/clinician/yr enterprise list — a board estimate, since Abridge publishes no pricing — that is 1.24x at the point estimate and 0.84x–2.41x across the interval. A wide-interval observation, then, not demonstrated overpricing, and the call does not rest on it. It rests on the $0 substitute (Doximity's free Scribe; athenahealth's athenaAmbient at no additional cost) and a record owner licensing the same function since 4 Feb 2026; on ~45–53x that fifteen-month-old ARR base; on the Epic equity-and-revenue-share dependency, which makes the moat and the constraint the same object and limits how far Abridge can disrupt its own distributor; and on the 32% of adopters who used the scribe on at least half their visits, which is the population average an enterprise buyer pays for. Pass the round; track the ASP.
Earnings, margins, COGS & capex
Fast-scaling private growth company. ARR moved from ~$60M (end-2024) to ~$100-117M (mid-2025), with valuation doubling to $5.3B inside a single year on the strength of enterprise health-system logos and native Epic distribution. Economics are growth-over-profit: capital is being deployed into product (inpatient, RCM/coding), model quality, and land-and-expand within large systems. Real margin/FCF data is not public; the key financial risk is inference/COGS drag; an Epic-channel revenue-share is reported but nowhere disclosed.
Revenue trend
Margins
structurally software-like, pressured by per-encounter LLM inference cost
growth-stage burn
venture-funded
COGS structure
Primary variable cost is model inference — transcription (ASR) + LLM summarization per patient encounter, run on cloud GPUs and/or foundation-model APIs, plus clinical/ML review and infra. This is the structural difference vs classic zero-marginal-cost SaaS: heavy-use clinicians (many encounters/day) raise unit COGS, so falling GPU/token prices are a direct gross-margin tailwind. Epic 'Abridge Inside' distribution is reported to carry a revenue-share, but none is disclosed: Epic's published vendor toll is a flat ~$1,700-1,900/yr Vendor Services membership plus a ~$500/yr Connection Hub listing, with no revenue share disclosed since App Orchard closed in Dec 2022 — so it should not be modelled as a margin cap.
Capex
Minimal physical capex (asset-light). No owned data centers disclosed; compute consumed as cloud opex. Capital is deployed into headcount (R&D, clinical, GTM) and model development, not fixed assets.
Latest earnings
n/a
No public guidance. Directional company signals: 150+ health systems at the Series E (Jun 2025), 250+ reported by mid-2026; projecting ~100M patient conversations documented in 2026; ARR trajectory through nine figures.
- Valuation
- ~$5.3B post-money (Series E, Jun 2025); the Apr 2026 extension disclosed no price
- Contracted ARR
- ~$117M (Q1 2025)
- Total funding
- ~$757.5M across announced rounds through Series E (Jun 2025); ~$1.07B incl. the press-reported ~$316M Apr 2026 extension — a computed sum, not a stated total
- Health systems
- 150+ at the Series E (Jun 2025); 250+ reported by mid-2026
- Enterprise price (estimated)
- ~$2,500 / clinician / year — an estimate corroborated by Sacra; Abridge publishes no pricing. Vendor and aggregator blogs put enterprise at $400-600/month, so ~$2,500 is the vendor-favourable end of a contradicted band.
- KLAS
- #1 Best in KLAS, Ambient AI (2025 & 2026)
Growth drivers
- Native Epic integration ('Abridge Inside', from Haiku to Hyperdrive, incl. inpatient and ED workflows) as a distribution moat into the largest EHR install base
- Land-and-expand inside mega-systems (Kaiser Permanente 40-hospital / 600+ medical-office rollout, Mayo 2,000+, Johns Hopkins, Duke, UPMC 12,000 clinicians, Yale New Haven, UCHealth)
- Expansion from documentation into higher-value revenue-cycle management / medical coding ('care intelligence'), lifting ACV and stickiness in fee-for-service settings — the E/M coding channel is worth ~$0 under capitation, including at anchor customer Kaiser Permanente
- Clinician-burnout / documentation-burden tailwind and health-system ROI (throughput, retention, and coding capture where billing is fee-for-service)
- Falling inference costs improving unit economics per encounter over time
- #1 Best in KLAS ambient-AI ranking (2025 & 2026) driving enterprise procurement credibility
Bull & bear
Abridge is compounding into the default care-intelligence layer for US health systems: it owns the leading brand, the deepest Epic distribution, and the reference logos, and is expanding from documentation into revenue-cycle economics where the ROI is hard-dollar in fee-for-service settings and the switching costs deepen.
- Category leadership + #1 Best in KLAS (2025 & 2026) makes it the safe enterprise choice, and health-system procurement rewards the incumbent leader
- Native Epic integration is a structural distribution advantage into the dominant EHR base
- ARR roughly doubling YoY with 150+ large systems at the Jun 2025 Series E and 250+ reported by mid-2026 signals genuine product-market fit, not just hype
- Expansion into RCM/coding turns a 'nice-to-have scribe' into a revenue-generating system-of-record — in fee-for-service settings — with rising ACV and retention
- Falling inference costs convert a current COGS headwind into a future gross-margin tailwind
- Over $1B cumulative funding and top-tier backers (a16z, Khosla, Lightspeed, Bessemer, Redpoint, Spark, CVS Health Ventures, Kaiser Permanente Ventures, NVIDIA NVentures, Eli Lilly) fund the land-grab window
Abridge is a well-funded leader in a rapidly commoditizing feature category that its most important partner (Epic) and largest competitor (Microsoft) can absorb — at ~45-53x contracted ARR on a Q1-2025 basis with undisclosed, inference-pressured margins, the valuation leaves no room for the compression that is already starting.
- Epic has shipped native ambient documentation (AI Charting, GA 4 Feb 2026) and can turn a partner into a subsumed feature — the same dependency that is Abridge's moat is its kill-switch
- Microsoft/Nuance Dragon Copilot lists at a reported $369-604 per provider per month (trade-press, no primary source) and holds ~33% reported category share; a well-capitalized incumbent that also supplies the transcription inside Epic AI Charting can commoditize the core scribe function
- A crowded field (Suki, Nabla, Ambience, DeepScribe, Commure, Doximity's free scribe, Oracle Health) drives ASPs down before Abridge reaches profitability
- Per-encounter LLM inference COGS holds gross margin below classic SaaS, and a reported-but-undisclosed Epic revenue-share would compound it
- ~$5.3B (Jun 2025) on ~$100-117M contracted ARR (Q1-2025 basis) demands sustained doubling; any deceleration or a lost mega-logo re-rates hard
- Foundation-model providers moving down-stack could disintermediate the application layer
What it is worth
Private-round + ARR multiple (no public price). Last disclosed mark ~$5.3B post-money (Series E $300M, Jun 2025); the reported ~$316M Series E extension of Apr 2026 disclosed no price. Against ~$100-117M contracted ARR on a Q1-2025 basis that is ~45-53x — a frontier-AI multiple on reported figures with no primary source, assuming sustained ~2x annual growth and eventual margin expansion.
Epic and/or Microsoft commoditize the core scribe, growth decelerates and margins stay capped by inference COGS and any Epic revenue-share; the ~45-53x multiple compresses and a down-round/re-rate to $2-3B follows.
Growth halves from hyper-growth but stays strong (~$150-200M ARR); valuation roughly holds at $5-6B as the market waits for margin proof; liquidity via later round or IPO.
If ARR compounds toward ~$300M+ with RCM expansion and improving gross margin, a 30-40x forward multiple sustains or exceeds $5.3B and an IPO clears well above — the leader premium holds.
As a private company there is no market price; secondaries trade via platforms like EquityZen/Forge. Multiple is rich even for AI, so the valuation is 'supported' only under a durable-category-leader thesis where Epic stays a partner and RCM expansion lifts ACV/margin. No digital-health IPO cleared in H1 2026, so near-term liquidity is secondary/round-driven. Not a buy/sell call.
SWOT
Strengths
- Clear category leader in health-system ambient AI by logos, ARR, and #1 Best in KLAS ranking (2025 & 2026)
- Deep native Epic integration ('Abridge Inside') — distribution most rivals cannot match
- Marquee enterprise deployments (Kaiser, Mayo, Johns Hopkins, Duke, UPMC) create reference-driven procurement flywheel
- Over $1B raised (incl. reported Apr 2026 extension) gives multi-year runway to outspend on model quality and GTM
- Physician-founder (Dr. Shiv Rao, cardiologist) credibility with clinical buyers
Weaknesses
- No public path to profitability; margins undisclosed and pressured by per-encounter inference COGS
- Product is a feature-adjacent capability that Epic and Microsoft already ship and can bundle (Epic AI Charting, GA 4 Feb 2026; Microsoft Dragon Copilot)
- A reported but undisclosed Epic-channel revenue-share would reduce economics on its strongest distribution path
- Heavy customer concentration risk in a handful of mega-systems
- Valuation (~45-53x contracted ARR, Q1-2025 basis) prices in near-flawless execution
Opportunities
- Move up-stack from notes into revenue-cycle/coding, prior-auth, and clinical decision support — larger, stickier TAM, with the coding component addressable only where billing is fee-for-service
- Nurse and inpatient documentation, specialty expansion (50+ specialties), and non-physician care settings
- Falling GPU/token costs structurally expand gross margin per encounter
- International and payer/ambulatory-adjacent expansion
- Consolidation candidate or eventual IPO given category leadership — though no S-1 is confirmed for Abridge or for any of the eleven tracked 2026-27 digital-health IPO candidates, and none has priced
Threats
- Epic AI Charting — Epic's native ambient documentation, generally available since 4 Feb 2026 and licensed inside the existing EHR relationship at an undisclosed price — the single largest existential threat
- Microsoft/Nuance Dragon Copilot (DAX) — deep-pocketed incumbent that lists at a reported $369-604 per provider per month (trade-press, no primary source)
- Foundation-model providers (OpenAI/Anthropic/Google) moving down-stack into clinical apps
- Commoditization — many funded rivals (Suki, Nabla, Ambience, DeepScribe, Commure, Doximity, Oracle Health) push pricing down
- Regulatory/liability exposure on AI-generated clinical documentation and PHI handling
Moats, dependencies & bottlenecks
Moats
Strong but rented powerful while Epic partners, fragile now that Epic also sells the function itself (AI Charting, GA 4 Feb 2026) Best go-to-market in the category, but the partner controls the channel and is reported — though nowhere disclosed — to take revenue-share.
High once fully deployed Deep integration into clinician workflow, EHR, and increasingly coding/RCM makes rip-and-replace costly.
#1 Best in KLAS (2025 & 2026) and Kaiser/Mayo/UPMC references de-risk procurement for the next buyer.
Scale of real encounters (50+ specialties, 15+ languages) aids model quality, but foundation-model commoditization narrows the edge.
Over $1B raised lets it outspend startups, but not Microsoft or Epic.
Dependencies
Distribution + integration partner AND active competitor Most important channel, and the most likely disruptor: Epic's own AI Charting has been generally available since 4 Feb 2026. Revenue-share terms are reported but undisclosed; Epic's disclosed vendor fees are flat.
OpenAI, Anthropic, Google GOOGL; cloud) Core technology + COGS input Inference cost and model access drive unit economics; NVIDIA is also an investor via NVentures.
Mayo, Johns Hopkins, Duke, UPMC) Revenue concentration A few enterprise logos likely drive a large share of ARR; churn of one is material.
Pre-profit; continued scale currently relies on equity funding and eventual IPO/secondary liquidity. No digital-health IPO cleared in H1 2026.
Advantages
- First-mover category leadership and strongest brand in health-system ambient AI
- Deepest Epic integration of any independent vendor
- Blue-chip customer references and #1 Best in KLAS (2025 & 2026)
- Largest war chest among independent AI-scribe startups (over $1B raised)
- Expansion into revenue-cycle economics beyond documentation, where the return is hard-dollar in fee-for-service settings
Weaknesses
- No disclosed path to profitability; margins undisclosed
- Strategic reliance on Epic — both moat and existential risk
- Commoditizing core product with many funded rivals and free alternatives
- Extreme valuation (~45-53x contracted ARR, Q1-2025 basis) sensitive to any growth deceleration
- Customer concentration in a small set of mega-systems
Bottlenecks
- Per-encounter inference COGS capping gross margin, plus a reported-but-undisclosed Epic revenue-share
- Clinical accuracy/liability bar — errors in AI-generated notes carry patient-safety and legal exposure
- Enterprise sales cycles and IT/security review length in health systems
- Talent competition for clinical ML in a field where Microsoft/Google also hire
- Dependence on third-party model roadmaps for core capability
Top signals & trends
Top signals
Top-tier investors underwriting continued hyper-growth; also raises the execution bar.
Enterprise adoption accelerating with marquee logos; projecting ~100M conversations in 2026.
Incumbent commoditization pressure on core scribe pricing; the list figure is trade-press reported, with no primary source.
The distribution partner is also arming rivals and itself.
Moves up-value-chain to defensible, hard-dollar ROI where billing is fee-for-service; the E/M coding channel is worth ~$0 under capitation, including at anchor customer Kaiser Permanente.
Trends
Expands TAM and normalizes procurement; Abridge is the reference leader.
Directly improves per-encounter gross margin over time.
Threatens to bundle the category into the system of record.
Suki, Nabla, Ambience, Doximity's free scribe and Microsoft's OEM position inside Epic AI Charting pressure ASPs.
Vendors that own coding/RCM raise ACV and stickiness.
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 compute for inference; also an investor via NVentures.
Foundation LLMs underpinning summarization/reasoning (Google public; OpenAI/Anthropic private).
Hosting and GPU capacity; Azure = MSFT, GCP = GOOGL.
Anchor deployment across 40 hospitals + 600+ medical offices (~24,600 physicians, reported); also an investor via KP Ventures.
Enterprise deployment 2,000+ physicians plus nursing pilots.
Named academic-medical-center customer.
UPMC / Duke Health / Yale New Haven / UCHealth UPMC scaled enterprise-wide to 12,000 clinicians; among the 250+ large-system customers reported by mid-2026.
The incumbent giant; deepest enterprise+Epic footprint, lists at a reported $369-604 per provider per month (trade-press, no primary source). Primary strategic threat.
Private EHR incumbent; Abridge's key partner and an active competitor since AI Charting reached GA on 4 Feb 2026.
Well-funded rival; on Epic Toolbox; RCM/coding overlap.
Voice-first assistant; strong mobile ratings; reported $299-399/clinician/month ($3,588-4,788/yr), above Abridge's estimated ~$2,500/yr.
Telehealth-strong AI copilot, GDPR+HIPAA; funded independent.
Specialty-tuned models; strong KLAS scores.
Consolidating health-AI platform including ambient scribe.
Doximity Scribe offered free to its large physician network — commoditization pressure at the low end.
#2 EHR building native clinical AI; bundling risk in Cerner systems.
SMB/clinic-focused low-cost scribes expanding upmarket.