Nabla
B2B subscription per clinician per month (reported list ~$119 = $1,428/yr; aggregators also report a ~$239 tier). Three routes to market - direct enterprise, EHR marketplaces (athenahealth), and an OEM module, 'Nabla Connect' (Oct 2025), that drops ambient AI into any EHR - which is the structural difference from the Epic-native leader. Founded 2018 in Paris; headquartered in New York.
Earnings, margins, COGS & capex
Private, pre-profit, capital-light, with one unusual feature for the category: the only product on the ladder with a positive randomised efficacy result, priced below the point estimate of the measured value anchor, so its commercial argument does not lean on the contested half of the ROI case. The value anchor is $167.37 of additional marginal E/M revenue per adopting clinician per month (95% CI $86.52-$248.21) - an EXPLORATORY outcome of a five-site JAMA cohort of 8,581 ambulatory clinicians, not one of the four prespecified outcomes - annualising to $2,008/yr (interval $1,038-$2,979). Nabla's ~$1,428/yr is 0.71x the point estimate and 0.48-1.38x across that interval: below it, but straddling parity at the top. The authors call the figure 'a conservative lower bound' and state the analyses 'cannot generalize to cost-benefit considerations' - which is what a price/value ratio does.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~99¢ is cost of goods and ~0¢ operating expense, leaving ~1¢ of operating profit.
Revenue trend
Margins
one of the lowest revenue-per-seat rungs on the ladder; falling token prices are the main lever
spending ahead of revenue
equity-funded from $120M cumulative
COGS structure
ASR on the encounter audio plus LLM generation of the note, both metered, so unit COGS scales with the 40M annual encounters rather than seats billed. Three Nabla-specific pressures: 35+ languages multiply the QA surface beyond an English-only scribe; the company runs its own model stack rather than reselling one frontier API; and at ~$119/month there is roughly 57% of the per-seat headroom a ~$2,500/yr Abridge seat gives (an estimate corroborated by Sacra; Abridge publishes no pricing), so heavy-use clinicians are more dilutive here than anywhere on the ladder. Marketplace and OEM channels carry undisclosed revenue-share economics.
Capex
Minimal. No owned data centres; inference and training run as cloud opex. The capital base is people and models.
Latest earnings
n/a
No financial guidance. Directional only: 190+ organisations, 100,000+ clinicians, 40M annual encounters (21 Jul 2026); a 2026 roadmap of deeper EHR integration and coding support; and an agentic roadmap on AMI world models with the stated ambition of being 'the first to bring FDA-certifiable agentic AI systems to healthcare' - an aspiration, not an authorisation.
- Valuation
- not disclosed at the last round; ~$180M reported at the Jan 2024 Series B. No mark asserted.
- Total funding
- $120M company-stated at the $70M Series C, 17 Jun 2025 (HV Capital lead; Highland Europe, DST Global, Cathay Innovation, Build Collective)
- Reported list price
- ~$119 / clinician / month = $1,428 / yr (reported list, no primary source - private company)
- Price vs measured value
- 0.71x the $2,008/yr point estimate; 0.48-1.38x across the 95% interval ($1,038-$2,979/yr)
- Randomised result
- -9.5% time-in-note (95% CI -17.2 to -1.8, P=0.02); 238 physicians, 14 specialties, UCLA Health, NEJM AI
- Deployment scale
- 190+ organisations, 100,000+ clinicians, 40M encounters, 35+ languages (company-stated, Jul 2026)
Growth drivers
- The category's only positive randomised result (-9.5% time-in-note, 95% CI -17.2 to -1.8, P=0.02), in a market where PHTI's March 2025 assessment found buyers see gaps in the productivity and financial evidence
- Price position — reported list ~$1,428/clinician/yr = 0.71x the $2,008/yr point estimate and 0.48-1.38x across the measurement's 95% interval - below the point estimate but straddling parity, with only Freed at 0.59x lower among the paid rungs
- EHR-agnostic distribution — Epic and Oracle Cerner plus athenahealth Marketplace, NextGen, Greenway, CharmHealth, Opus and Nabla Connect
- Widening from the note into dictation (M Health Fairview's stated reason for choosing Nabla) and into coding and revenue-cycle support (Toledo, Apr 2026)
- Exclusive first access to AMI Labs' world models (18 Dec 2025), co-founder Alex LeBrun now AMI's CEO - though no formal equity or licensing agreement exists yet
- Brian Manning as CEO (21 Jul 2026), previously President and CRO of Bamboo Health, which he helped scale past $140M ARR
Bull & bear
The only ambient vendor shown to work in a randomised trial, priced at 0.71x the $2,008/yr point estimate of measured value and 0.48-1.38x across the measurement's 95% interval - so it wins a purely evidentiary bake-off, and the price comparison straddles parity rather than settling against it. It reaches the non-Epic majority of ambulatory clinicians through marketplaces and an OEM module rather than one partner's channel, and doubled annual encounters in about 13 months on one-ninth of the leader's capital.
- Evidence asymmetry is the thesis: Nabla cut time-in-note 9.5% (95% CI -17.2 to -1.8, P=0.02) while DAX Copilot moved -1.7% and missed significance - and PHTI says buyers see gaps in exactly this evidence
- Price leaves room the rungs above do not have: 0.71x the $2,008/yr point estimate and 0.48-1.38x across the measurement's 95% interval, while the enterprise tier above it (Abridge to Ambience's full suite) runs 1.24-2.49x at the point estimate, so a price war compresses rivals' structure first
- Distribution is not rented from one partner: Epic and Oracle Cerner plus athenahealth Marketplace (30+ groups in year one), NextGen, Greenway, CharmHealth, Opus and Nabla Connect
- Scale compounding on modest capital: 130+ organisations / 85,000+ clinicians / 20M encounters (Jun 2025) to 190+ / 100,000+ / 40M (Jul 2026) on $120M raised
- The value anchor may be far too low, making the 0.71x point ratio (0.48-1.38x across the interval) a bargain: PHTI reported in April 2026 that a health system saw a 5% rise in Level 5 encounters after AI scribes, 'raising revenue by over $1,000 per provider per month' - about 6x the JAMA point estimate. Single-system, self-reported and uncontrolled, so it cannot carry a valuation; but it comes from the institution cited for 'ROI unclear', and the JAMA authors call their own estimate conservative partly because only ~32% of adopters used the scribe on half or more of visits
- The AMI link is a free option on the next architecture: exclusive first access and a shared founder at a lab that raised roughly $1B+ in Mar 2026 - and with no formal agreement yet, none of it has been paid for
Nabla holds the best evidence in a category whose buyers do not buy on evidence. ~4% share against Microsoft's 33% and Abridge's 30%, one-ninth of the leader's capital, one of the lowest revenue-per-seat rungs on the ladder, and a price floor beneath it of exactly zero. Its founder-CEO now runs a different company. And the anchor that makes its price look disciplined is an exploratory, E/M-only estimate its own authors say cannot be generalised to cost-benefit.
- Being right does not confer share: the randomised win is published and Nabla still sits at ~4%, because KLAS shortlists, Epic-native integration and EHR bundling decide these deals
- Capital asymmetry is structural: $120M against Abridge ~$1,061M and Ambience $345M - and the category's own research finds large systems 'just as likely to switch vendors as to stay', with customers calling scribing 'becoming commoditized'
- The price that makes the arithmetic close caps the business: ~$1,428 per clinician-year needs ~1.75x the seats to match an estimated ~$2,500 Abridge seat, before channel dilution
- The floor is free: Doximity reported over 800,000 active prescribers on workflow tools, nearly half using its clinical AI - a bundle of AI search AND Scribe, not a Scribe-only count - roughly the order of the whole paid market's installed base
- Founder attention is split at the worst moment: LeBrun owns the model roadmap while serving as CEO of AMI Labs, through the consolidation window
- The value anchor is fragile both ways: $167.37/month is EXPLORATORY, monetises billed E/M visits only, does not monetise the 16.0 min/8h documentation-time saving, reflects a population in which only ~32% of adopters used the scribe on half or more of visits, and is worth ~$0 at capitated buyers. Its authors state the analyses 'cannot generalize to cost-benefit considerations', so the 0.71x that flatters Nabla is a wide-interval observation, not a demonstrated bargain
What it is worth
No public market and no current private mark, so there is nothing to multiply. What can be stated: $120M cumulative funding, company-stated at the $70M Series C of 17 Jun 2025 (HV Capital lead, with Highland Europe, DST Global, Cathay Innovation and Build Collective), with NO post-money disclosed. The only reported mark is ~$180M around the $24M Series B of Jan 2024 - trade-press sourced, ~2.5 years stale, a negotiated price in one round rather than a clearing price. No valuation may be inferred from a raise size, and none is asserted here.
The floor rises to meet it: Epic's in-contract AI charting, athenaAmbient's bundled rung and Doximity's free tier make a $119 seat unwinnable on price, Microsoft's re-priced licence takes the mid-market and Abridge's capital takes the enterprise. Share drifts below 4%, the next round is flat or down, and the best randomised evidence in the category turns out to be the best asset in a business the market never paid for.
Nabla stays the credible number-three-or-four independent: mid-single-digit share, scale compounding on modest capital, a low price that keeps it in every bake-off without winning the mega-system deals. The durable outcome is acquisition by an EHR, a platform consolidator or a payer/retail-health buyer wanting ambient plus multilingual coverage.
The randomised result becomes the default answer for evidence-led buyers while marketplace and OEM channels compound outside Epic's orbit; encounters keep doubling, coding and agentic workflows lift ACV above the $1,428 documentation seat, and Nabla re-rates to a mark well above anything reported so far.
The honest comparable frame is the category's own multiples, themselves stale: Abridge at $5.3B on ~$117M of Q1-2025 contracted ARR (roughly 45-53x, a Jun-2025 mark against an ARR base fifteen months older) and Ambience at $1.25B on ~$30M ARR (May 2025). If Nabla's ~4% share of a ~$600M category is roughly right, its implied revenue is an order of magnitude below Abridge's, so a round priced inside that range would still be a fraction of the leaders'. Two facts pull opposite ways at the next round: Nabla holds the only positive randomised efficacy result in the category, and one of the lowest revenue-per-seat rungs on the ladder. Nothing here is a price target or a recommendation.
SWOT
Strengths
- The category's only positive randomised result — -9.5% time-in-note (95% CI -17.2 to -1.8, P=0.02) vs DAX Copilot -1.7% (P=0.66)
- Reported list ~$1,428/yr = 0.71x the $2,008/yr point estimate and 0.48-1.38x across the measurement's 95% interval - below the point estimate but straddling parity, with only Freed at 0.59x lower among the paid rungs
- EHR-agnostic distribution — Epic and Oracle Cerner plus athenahealth Marketplace, NextGen, Greenway, CharmHealth, Opus and Nabla Connect
- Ambient documentation and dictation on one platform - M Health Fairview's stated reason for choosing it
- Clinical bench outsized for its funding — CMO Dr. Ed Lee (ex-CIO, The Permanente Federation), advisers Yann LeCun and Tony Fadell
Weaknesses
- ~4% of a ~$600M US category against Microsoft/Nuance 33% and Abridge 30%
- $120M raised against Abridge ~$1,061M and Ambience $345M - one-ninth of the leader's capital
- The low seat price is low revenue per clinician — ~1.75x the seats needed to match an estimated ~$2,500 Abridge seat
- Founder-CEO LeBrun moved to executive chairman / chief AI officer and is simultaneously CEO of AMI Labs
- No disclosed revenue, margin, burn, cash or current valuation
Opportunities
- The non-Epic ambulatory market — independent groups, community and safety-net systems, marketplace-sourced practices
- Coding and revenue-cycle support, where the return is hard-dollar rather than time
- Agentic workflows on AMI world models: referrals, scheduling, insurance checks, pre-visit lab orders
- Falling inference cost, which relieves the ladder's lowest-priced seats disproportionately
- Multilingual reach - 35+ languages differentiates for safety-net and international buyers
Threats
- The floor is $0, not $119 — Doximity Scribe is free to verified US clinicians, athenaAmbient carries no additional cost inside athenahealth's core fee, and Epic licenses AI charting inside the EHR relationship
- Microsoft reset Dragon Copilot licensing on 1 May 2026 (one per-user licence, ambient billed per encounter), reported list $369-$604/provider/month
- The value denominator may shrink — the JAMA figure monetises billed E/M visits only at the 2025 Medicare PFS and is worth ~$0 wherever the clinician is not paid fee-for-service (Kaiser Permanente, the VA, staff-model plans, full-risk MA/ACO)
- Payers contest the coding lift — Cigna's R49 auto-downcoding of 99204-05 / 99214-15 / 99244-45 was paused after AMA advocacy and Maryland fined Cigna $80,000 (13 Mar 2026), but the audit right is untouched and CMS's MA coding-intensity adjustment still claws back
- Safety exposure inside the trial that vindicates it — UCLA investigators reported clinically significant inaccuracies in some notes, with one mild patient-safety event, concluding it 'requires active physician oversight, not passive acceptance'
Moats, dependencies & bottlenecks
Moats
and for now unique a result, not a barrier A rival with 9x the capital could fund several trials; as of Aug 2026 none has published one that worked.
Reaches buyers outside the Epic-native fight. A channel owner can change terms, but none is a kill switch.
0.71x the point estimate and 0.48-1.38x across the measurement's 95% interval, so it straddles parity - defensive, not accretive.
Real switching cost, but the category's research finds large systems as likely to switch as stay.
35+ languages differentiates for safety-net and international buyers; frontier models close that gap for rivals each quarter.
Dependencies
Epic, Oracle Health (ORCL), athenahealth, NextGen, Greenway, CharmHealth Distribution channel and integration surface; each also a potential competitor Epic - 43.7% of US acute-care EHR share (per KLAS as relayed), $6.7B of 2025 revenue (company-stated to Becker's, 28 Apr 2026 - unaudited and unverifiable because Epic files nothing) - licenses AI charting inside its own relationship; athenaAmbient ships inside the core fee.
cloud GPU capacity, the ASR/LLM stack, AMI Labs' world models Core technology and per-encounter COGS Unit economics track token and GPU prices at one of the ladder's lowest revenue-per-seat rungs. The AMI link is first-access only - no formal agreement as of Mar 2026.
commercial payers) The buyer's economic justification Worth ~$0 at capitated buyers and contested by payers. A CY2028 -2.5% efficiency adjustment extended to E/M would trim the monthly figure by about $4.18 - real, but not an evaporation.
Pre-profit with $120M raised against rivals holding 3-9x more, and no digital-health IPO cleared in H1 2026 to reset private marks.
Leadership and model roadmap LeBrun owns the AI strategy while serving as CEO of AMI Labs - the technical roadmap sits with a part-time principal.
Advantages
- The only vendor in the category with a positive randomised efficacy result (NEJM AI, UCLA Health)
- Priced at 0.71x the $2,008/yr point estimate and 0.48-1.38x across the measurement's 95% interval - below the point estimate but straddling parity, with only Freed at 0.59x lower among the paid rungs
- EHR-agnostic reach: marketplace plus OEM ('Nabla Connect') distribution into any EHR
- One platform for ambient documentation and dictation - a stated procurement differentiator
- Model-roadmap option via exclusive first access to AMI Labs' world models, at zero disclosed cost
Weaknesses
- ~4% US category share against Microsoft/Nuance 33% and Abridge 30%
- $120M raised versus ~$1,061M (Abridge) and $345M (Ambience)
- Among the lowest revenue-per-clinician rungs on the ladder, diluted further by channel economics
- No disclosed revenue, margin, cash or current valuation
- Founder-CEO now leads another company; the CEO seat changed hands mid-consolidation
Bottlenecks
- Enterprise procurement — security review, integration queues and legal cycles gate revenue more than product quality, and KLAS shortlists favour the incumbent
- Per-encounter inference plus multilingual QA cost against one of the lowest revenue-per-seat rungs on the ladder
- Clinical accuracy and required oversight — UCLA found clinically significant note inaccuracies and one mild safety event, so scaling encounters scales a liability surface
- Regulatory headroom is unbuilt — no 510(k) clearance, no PMA approval and no De Novo authorisation is claimed, and none is required for a documentation tool making no diagnostic or treatment claim - but the 'FDA-certifiable' agentic ambition implies a cost neither budgeted nor demonstrated
- Capital: $120M cannot fund frontier model work, implementation staffing and a national GTM build at once
Top signals & trends
Top signals
238 physicians, 14 specialties at UCLA Health, Nov 2024-Jan 2025; the effect did not replicate across both arms (Nabla's absolute move was 4:30 to 3:49). Burnout and task load improved for users of either scribe - the differentiation is documentation time, not wellbeing.
Company-stated, unaudited. Encounters roughly doubled in 13 months, but encounters are usage rather than billings.
An operator who scaled Bamboo Health past $140M ARR fits this phase - against a founder who owns the model roadmap now running another company.
Priced below the point estimate of measured value, which is protective; near the bottom of the ladder, which caps ARPU. Reported list, no primary source.
Menlo Ventures via Becker's, Oct 2025 - VC work through trade press, ~10 months old. The same research found large systems 'just as likely to switch vendors as to stay'.
The marketplace/OEM strategy converts in the segment an Epic-native competitor reaches least well.
Trends
Roughly 60 ambient scribes were on the market at PHTI's March 2025 count, splitting ~$600M of 2025 US revenue - adoption is settled, vendor selection is not.
Doximity Scribe at $0, athenaAmbient bundled inside the core fee, Epic AI charting licensed in-contract, Microsoft's 1 May 2026 reset. A low price helps until the floor is free.
Where the return turns hard-dollar, ACV and stickiness rise - but every competitor is moving the same way.
Cigna's R49 downcoding was paused after AMA advocacy and prohibited in Maryland, but the audit right survives and the MA coding-intensity adjustment claws back mechanically.
Measured effects run ~6x below the category's advertising - Dragon Copilot markets '13 additional appointment slots per provider per month' against a measured 2.1.
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.
Inference and training as opex; Azure = MSFT, GCP = GOOGL. Which provider Nabla runs on is not disclosed.
Private - Yann LeCun's lab, which raised roughly $1B+ in Mar 2026. Exclusive first access to its world models (18 Dec 2025); co-founder Alex LeBrun is AMI's CEO. No formal agreement yet.
Per-encounter ASR plus note generation are the dominant variable costs; Nabla runs its own model stack.
Systemwide across 10+ hospitals and 60+ clinics (5 Feb 2026) on Epic; chose the combined ambient + dictation platform.
The largest safety-net system on Nabla: a 50-clinician pilot across twelve specialties, then 400 sign-ups in the first week systemwide. Pilot metrics are vendor-reported and uncontrolled.
Both the site of the NEJM AI randomised trial and a company-named customer - the reference and the evidence are the same institution.
Iowa / Carle Health / Toledo / Aultman / CVS Health Named deployments; Toledo scaled on revenue-cycle grounds (Apr 2026), Aultman via Oracle Cerner (Jan 2026). Iowa and Carle outcome figures are vendor case studies, not measurement.
Share leader at ~33% and the arm that did NOT work in the randomised trial (-1.7%, P=0.66) - Nabla's strongest competitive line. Reported list $369-$604/provider/month; licensing reset 1 May 2026.
Private category leader: ~30% share, ~$1.06B raised, a $5.3B mark (Jun 2025), ~$117M contracted ARR on a Q1-2025 base, Best in KLAS two years running, 150+ health systems at the Series E and 250+ reported by mid-2026. Wins on logos and capital; loses on randomised evidence and price.
Private; 43.7% of US acute-care EHR share (per KLAS as relayed), $6.7B of 2025 revenue (company-stated to Becker's, 28 Apr 2026 - unaudited and unverifiable because Epic files nothing). Licenses AI charting inside the existing EHR relationship rather than giving it away - a rung the CIO already contracts for.
Private; $1.25B mark on a $243M Series C (30 Jul 2025), $345M raised, ~13% share, ~$30M ARR (May 2025). Top of the price ladder ($2,800-3,200 base; $4,000-5,000 full suite).
Free to verified US clinicians, funded by pharma marketing at an 89.1% FY2026 GAAP gross margin. Over 800,000 active prescribers on workflow tools, nearly half using its clinical AI (search AND Scribe bundled).
Private; $168M raised, ~10% share, $299-399/clinician/month reported list - priced above the measured value anchor at every point in its interval (1.20-4.61x).
The #2 EHR building native clinical AI - simultaneously a Nabla channel (Aultman, Jan 2026) and the platform most able to bundle the function away.
Private, sponsor-owned. Ships ambient at no additional cost inside the core fee while also being one of Nabla's marketplace channels - bundled-versus-partner is the live tension.
Private; $99/clinician/month = $1,188/yr, 0.59x the point estimate - the only PAID rung below Nabla (the free rungs are Doximity Scribe and bundled athenaAmbient). SMB-weighted, so it presses where Nabla's self-serve tier sells.
Private challengers: Commure a $7B-marked roll-up and the highest-signal IPO on the slate; DeepScribe specialty-tuned with top-decile KLAS satisfaction; Heidi self-serve and international.