Innovaccer
B2B enterprise platform, mid-conversion. Historically a data-platform subscription (Gravity, Atlas); now explicitly moving to transaction pricing per successfully completed task — a prior authorisation, a denial appeal — rather than a licence (MedCity News, 15 Apr 2026). CaduceusHealth (21 May 2026) bolts on a labour-based revenue-cycle services book. A third line runs the other way: as a CMS WISeR participant Innovaccer is paid by the federal payer 'a percentage of the expenditures associated with averted wasteful, inappropriate care as a result of their reviews' (CMS's own language, canon REG-WISR-004). No CPT code pays for any of it, and nothing in the stack is a regulated device — no 510(k) clearance, De Novo grant or PMA approval claimed or required.
Earnings, margins, COGS & capex
The most operationally legible company on canon's IPO watchlist and among the least financially legible. It has an ARR figure at a named source (~$200M, +54%, Fortune 23 Jul 2026), a named federal contract (CMS WISeR, Ohio / MAC jurisdiction J15), marquee customers who are also investors, five completed acquisitions, and a pricing model it is publicly re-architecting from licences to per-task transactions. It has never published a gross margin, an operating result, a burn figure, a runway, customer concentration or net revenue retention. The comparison canon permits: at ~$200M of ARR Innovaccer is roughly twice Abridge's ~$100–117M of contracted ARR on a Q1-2025 basis (CF-ABR-03 — fifteen months older than the Jul-2026 ARR it is set against, so the ratio is directional) and about two-thirds of OpenEvidence's reported ~$300M annualised run-rate as of Jul 2026 (CM-PRIV-OE-03; trade press, no primary source). Those are the two private names with a reported revenue base, and neither derived multiple is offered as an analytical finding. Innovaccer has no multiple at all, because the denominator has no numerator at a source.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~79¢ is cost of goods and ~0¢ operating expense, leaving ~21¢ of operating profit.
Revenue trend
Margins
Diluting by construction near-term — per-task pricing puts inference straight into COGS and CaduceusHealth adds labour-based revenue; the offsetting tailwind is falling cost per inference.
Actively managed — 340 roles cut May 2026, a third restructuring in four years, against a $250M three-year platform commitment.
Equity-funded; ~$651M raised on canon's basis, plus a ~$75M ESOP buyback in Jan 2026.
What a listed, profitable healthcare-workflow-automation business discloses. Innovaccer discloses none of it — that gap is the distance between the two.
COGS structure
Three cost bases under one platform brand, blend unpublished. (1) Data platform — Gravity ingests and normalises EHR, claims, CRM and HR/finance data; cost is storage, pipelines and the bespoke integration engineering every health system's data model demands, front-loaded per customer and amortised over a multi-year contract. (2) Agents — priced per completed task, so each prior authorisation, denial appeal or coding check burns inference at the moment revenue is earned. Genuinely variable COGS, and where transaction pricing either compounds margin as inference cheapens or compresses if task complexity outruns unit cost. The NVIDIA collaboration (28 Oct 2025 — NeMo Guardrails and Framework, Riva Parakeet NIM, Triton Inference Server, TensorRT-LLM, for voice documentation, OCR and model training) reads as much like a cost-per-task programme as a capability one. (3) Services — CaduceusHealth is a 1997-founded revenue-cycle operation serving nearly 4,000 providers and administering $5B of annual gross patient charges; a headcount business whose margin improves only if the agents displace the labour. WISeR differs again: CMS requires that 'all recommendations for non-payment or non-affirmation will be determined by appropriately licensed clinicians', making a clinician panel a mandatory, non-automatable cost of that contract.
Capex
Minimal and asset-light — no owned data centre, GPU capacity as cloud opex. Capital goes to engineering and clinical headcount, the $250M three-year agent-platform commitment, and M&A — five deals closed (Cured Jan 2024, Pharmacy Quality Solutions Mar 2024, Humbi AI Jan 2025, Story Health Sep 2025, CaduceusHealth May 2026), with two to three more reported as planned.
Latest earnings
n/a. The nearest thing to a miss is against its own words: 'on track to hit $250 million in ARR this year' (Jan 2025) versus ~$200M crossed by 23 Jul 2026.
No financial guidance. Directional only: an IPO threshold of $400–500M ARR, a $250M three-year platform commitment, and a reported intent to complete two to three further acquisitions.
- ARR
- ~$200M crossed (Fortune, 23 Jul 2026) — company-stated, unaudited
- ARR growth
- ~+54% YoY from ~$130M
- Valuation
- NOT CARRIED — unverified at any named source (canon PB-024); the circulating figure is anonymous-source in origin and never company-confirmed
- Total raised
- ~$651M on canon's basis (PB-024); $675M per Fortune — gap unreconciled
- Last round
- $275M Series F, 9 Jan 2025, ~35% of it secondary liquidity for seed and Series A holders
- Top-10 health systems served
- 6 of 10 (Jan 2025) → 7 of 10 (Jul 2026)
- Stated reach
- 200+ health systems and payers, 95% of community pharmacies, 80M patient lives (company, 21 May 2026)
- Acquisitions
- 5 — Cured (Jan 2024), Pharmacy Quality Solutions (Mar 2024), Humbi AI (Jan 2025), Story Health (Sep 2025), CaduceusHealth (21 May 2026)
- CaduceusHealth scale acquired
- ~4,000 providers, $5B of annual gross patient charges; deal value not disclosed by the company
- Workforce action
- 340 roles cut, reported 15 May 2026 — third restructuring in four years; prior round ~245 roles, Jan 2023
- Federal contract
- CMS WISeR participant, Ohio / MAC jurisdiction J15 CGS; model runs 1 Jan 2026 – 31 Dec 2031
- FDA authorisations
- None claimed, none required — nothing in the stack is marketed as a regulated device
Growth drivers
- The pricing conversion — per-completed-task billing turns each automated prior authorisation and denial appeal into a metered revenue event, which is how a $200M base reaches the CEO's own $400–500M IPO threshold without doubling the customer count
- Revenue cycle as the newest and largest attach — CaduceusHealth brings ~4,000 providers and $5B of annual gross patient charges into Flow, and RCM is the one health-system budget with a self-evident payback
- Prior-authorisation regulation as a forcing function — CMS-0057-F imposes 7-day standard / 72-hour expedited decision clocks from 1 Jan 2026 and four API mandates (Patient Access, Provider Access, Payer-to-Payer, Prior Authorization) from 1 Jan 2027 (canon REG-PA-001); CMS-0062-P extends electronic prior authorisation to drugs (REG-PA-003)
- The WISeR contract as a federal reference — CMS selected Innovaccer for Ohio and MAC jurisdiction J15 CGS, routing prior-authorisation requests through its portal from 5 Jan 2026, in a model running to 31 Dec 2031 (canon REG-WISR-001/002)
- Investor-customers converting into top-of-market references — Kaiser Permanente and Banner Health invested in the Series F and appear among named deployments with Ascension, Trinity Health, Prisma Health, Risant Health and Franciscan Health (MedCity News, 15 Apr 2026)
- A measurable agent outcome to sell on — Risant Health's prior-authorisation turnaround falling from roughly 45 minutes to under one minute (MedCity News, 15 Apr 2026)
Bull & bear
Innovaccer is the only company on canon's eleven-name IPO watchlist that has put a real revenue base on the record at a named source, and it sits on the two budgets that survive a margin squeeze — revenue cycle and prior authorisation. It has a federal payer as a customer, seven of the top ten health systems as references, and a pricing model that charges for completed work rather than seats. If any of the eleven files, this is the one whose S-1 could actually pass the board's payer test.
- It has a number. ~$200M ARR growing ~54% is roughly twice Abridge's ~$100–117M of contracted ARR on a fifteen-month-older Q1-2025 basis (canon CF-ABR-03) and about two-thirds of OpenEvidence's reported ~$300M annualised run-rate as of Jul 2026 (CM-PRIV-OE-03; trade press, no primary source) — the only figure of its kind on the watchlist
- The payer is right. Revenue cycle is paid out of the provider's collected revenue — the test canon uses to justify holding Waystar — and prior authorisation is paid by the health plan. Neither needs a CPT code to exist, which is the trap that catches every diagnostic and ambient vendor on this board
- Per-task pricing converts ROI from a promise into an invoice. Canon's PHTI record is that health-system leaders do not believe the financial-ROI claims made for clinical AI; a vendor billing only on a completed prior authorisation argues on the buyer's terms
- Regulation is a tailwind with dates on it: decision clocks live since 1 Jan 2026, four API mandates due 1 Jan 2027, drug ePA proposed in CMS-0062-P. Payers must automate whether or not they want AI
- CMS itself is a customer — the WISeR prior-authorisation channel for Ohio and MAC jurisdiction J15 through 31 Dec 2031, compensated on averted expenditures. No competitor on the watchlist holds a six-year federal reference
- The acquisition machine works on a fragmenting field: five deals closed, CaduceusHealth adding ~4,000 providers straight into Flow, and canon PB-065 shows the listed cohort shrinking by acquisition faster than the IPO class replaces it — scarcity accrues to whoever lists first with a real book
A company with no verifiable valuation, one disclosed financial metric, a growth rate that already ran behind its own guide, a diluting margin mix it will not quantify, three restructurings in four years, and a federal contract that pays it to avert payment while it sells software to the providers filing the claims. The public market has already priced this category, and it priced it at a few hundred million dollars per name.
- There is no mark. The circulating figure originates with an unnamed 'source familiar with the deal' at the Jan 2025 Series F and has been laundered through aggregator pages ever since. A number nobody will attach their name to is not a price
- The listed comparables are brutal and canon-sourced: Health Catalyst $154M, Evolent Health $348M, Phreesia $663M at the 31 Jul 2026 close (CM-COH-01) — roughly $1.17B combined for the three closest public analogs of Innovaccer's own product lines. Any private mark has to explain that gap on evidence Innovaccer does not publish
- One disclosed metric, and it is the flattering one. ARR without gross margin, retention, concentration or burn cannot distinguish a compounding platform from a services roll-up growing by acquisition — and a large slice of recent scale arrived attached to a deal
- Margin runs the wrong way on purpose: per-task pricing pushes inference into COGS at the moment of revenue, CaduceusHealth adds a labour book, and WISeR mandates licensed-clinician review of every non-affirmation — a cost the rule forbids automating away
- The cost base has outrun revenue repeatedly: ~245 roles cut Jan 2023, 340 cut May 2026, against a $250M three-year forward commitment and an undisclosed burn
- The WISeR position is a live political hazard. A defunding amendment cleared committee in Sept 2025 (canon REG-WISR-005, per KFF), and 42 CFR 422.566(d) has required physician review of adverse determinations since long before AI, capping how much review can ever be automated (REG-PA-004)
What it is worth
NO valuation figure is carried, by canon instruction (PB-024) and on the evidence. Innovaccer is private, files nothing consolidated, has never confirmed a mark, and the only number in circulation is attributed to an unnamed 'source familiar with the deal' at the Jan 2025 Series F and then repeated by aggregator pages that add no source of their own. What can be done instead is triangulation on things that ARE sourced: the disclosed ARR base and growth rate, the market caps of the closest listed analogs, and the disclosed economics of the one profitable listed competitor.
The disclosure that eventually arrives is worse than the narrative. Per-task pricing proves margin-dilutive once inference and mandatory clinician review are loaded in, the CaduceusHealth services book drags blended gross margin, growth decelerates as the top-ten health systems saturate and Epic ships adequate native equivalents, and a fourth restructuring lands. In that world the reference set is the listed one — Health Catalyst $154M, Evolent $348M, Phreesia $663M at the 31 Jul 2026 close — and the private mark nobody will attach a name to gets re-cut in a down round or a structured secondary. The WISeR contract, the single most differentiated asset here, is also the most politically exposed: a committee-approved defunding amendment already happened once, in Sept 2025.
No priced event. Growth continues in the 40–55% band, the company stays private below its own $400–500M ARR bar, liquidity comes from further secondaries and ESOP buybacks, and the valuation remains what it is today: unverifiable at any named source and therefore uncarryable. Another acquisition or two lands; the margin question stays unanswered because nothing compels an answer.
Re-rated on disclosure rather than narrative: ARR compounds past $400M on the per-task model, an S-1 discloses AI-attributable revenue separately at a gross margin that survives the services mix, the payer is the provider's collected revenue and the health plan rather than a CPT code that does not exist, and the WISeR contract renews or widens. That is exactly the screen canon's board applies to a first listing — and Innovaccer is the only watchlist name with the revenue base to attempt it.
The sourced anchors, in order of usefulness. (1) REVENUE: ~$200M ARR crossed, up ~54% from ~$130M (Fortune, 23 Jul 2026) — company-stated, unaudited, ARR not GAAP. (2) LISTED ANALOGS: at the 31 Jul 2026 close canon's cohort (CM-COH-01, aggregator-sourced quotes) carries Health Catalyst at $154M, Evolent Health at $348M and Phreesia at $663M — roughly $1.17B combined for the three closest public expressions of Innovaccer's data-platform, value-based-care and patient-access lines. (3) THE PROFITABLE COMPARABLE: Waystar at a $4.05B market cap with Q2 2026 revenue of $319.7M (+18.1%), adjusted EBITDA of $136.7M and a FY2025 adjusted EBITDA margin of 42.04% (canon CM-COH-01, PB-018) — roughly 3x market cap to annualised Q2 revenue, before debt, for a business that discloses everything Innovaccer does not. (4) PRIVATE PEERS WITH A REPORTED REVENUE BASE: Abridge at ~45–53x on a Q1-2025 contracted-ARR base of ~$100–117M against its $5.3B post-money mark of Jun 2025 (CF-ABR-02/03), and OpenEvidence at a reported ~$300M annualised run-rate as of Jul 2026 against its Jan-2026 $12B mark — ~40x on Jul-2026 revenue and ~80x on revenue at the mark date (CM-PRIV-OE-03; trade press, no primary source). Both are reported-revenue-implied rather than analytical findings, and evidence of what the private market has paid rather than evidence about Innovaccer. (5) WHAT IS NOT AN ANCHOR: the raise. ~$651M raised (canon PB-024; $675M per Fortune) is capital consumed, not value created, and no valuation is inferred from it. Innovaccer is deliberately absent from canon's six-mark private cohort (CM-COH-04, ~$33.05B of last-round marks dated Jun 2025 – May 2026) because its mark cannot be sourced; including it would corrupt the aggregate. Liquidity is a further private round, a secondary, or an IPO the CEO has conditioned on $400–500M ARR — roughly double the current base. Canon PB-024 records that none of the eleven watchlist candidates had priced at 7 August 2026, no S-1 is confirmed for any, and that any will price is not assertable. Not investment advice and not a price target.
SWOT
Strengths
- A revenue base disclosed at a named source — ~$200M ARR growing ~54% (Fortune, 23 Jul 2026) — more than any other name on canon's eleven-candidate IPO watchlist has put on the record
- Federal customer status — CMS selected Innovaccer for WISeR in Ohio and MAC jurisdiction J15, a model to 31 Dec 2031 whose participants are paid a share of averted expenditures (canon REG-WISR-001/002/004) — the rarest reference in health AI, a government payer that buys the product
- Top-of-market placement — 7 of the top 10 US health systems, 200+ health systems and payers, 80M patient lives, with Kaiser Permanente and Banner Health as both Series F investors and named deployments
- Five agent categories on one data substrate, so a single Gravity integration is amortised across multiple budget lines rather than one
- Pricing aligned to the buyer's arithmetic — billing per completed task answers the objection canon's PHTI record raises against seat-priced clinical AI (CM-PHTI-02 — burnout improves, financial ROI unproven)
Weaknesses
- No verifiable valuation at any named source — the fact canon isolates about this company, and one no other watchlist name carries. There is no mark to defend, discount or benchmark
- ARR is the only financial disclosure — no gross margin, operating result, burn, runway, retention or concentration, so a 54% growth rate cannot be assessed for quality of revenue
- The trajectory ran behind its own guide — '$250M ARR this year' in Jan 2025 against ~$200M crossed in Jul 2026
- Margin structure diluting by design — per-task inference in COGS plus an acquired labour-based services book, layered onto what was sold as a software platform
- Serial restructuring: 340 roles cut May 2026, ~245 in Jan 2023, a third round in four years
Opportunities
- The prior-authorisation compliance wall — CMS-0057-F's four 1 Jan 2027 API deadlines plus CMS-0062-P extending electronic prior authorisation to drugs (canon REG-PA-001/003) force payer-side automation on a statutory clock
- WISeR expansion — six performance years across six states against a pool KFF's analysis of 2024 Medicare claims estimates at $12.3B, 5.3% of traditional-Medicare Part B spending (canon REG-WISR-003); a participant that performs in Ohio is the natural candidate if CMS widens scope
- Revenue cycle is the deepest unautomated budget in US healthcare, and CaduceusHealth's $5B of administered gross charges is a live book to run agents against rather than a greenfield sale
- Consolidation as strategy — five acquisitions done and two to three more reported as planned, in a fragmented category where point vendors are cheap relative to a platform with distribution
- HTI-5, if finalised as proposed, would amend 45 CFR 171.102 so 'access' and 'use' of EHI expressly include autonomous AI systems (canon REG-IB-007), making refusal information blocking against a $1,327,209-per-violation ceiling. Proposed only, and canon records zero enforcement actions to date (REG-IB-002)
Threats
- The listed comparables are the harshest evidence on the board — at the 31 Jul 2026 close canon's cohort carries Health Catalyst at $154M, Evolent Health at $348M and Phreesia at $663M — the three closest public analogs to Innovaccer's data-platform, value-based-care and patient-access lines, roughly $1.17B combined (canon CM-COH-01, aggregator-sourced quotes)
- Epic builds rather than buys and owns the substrate — 43.7% of US acute-care hospitals and 56.9% of beds in 2025 per KLAS via trade relay (canon CM-KLAS-01), on $6.7B of company-stated 2025 revenue (CM-EPIC-01). A data-unification pitch is weakest where one vendor already holds the data
- The WISeR contract is a liability as well as an asset — Innovaccer is paid by CMS a share of averted expenditures while selling revenue-cycle software to providers whose claims it reviews, and a defunding amendment cleared the House Appropriations Committee in Sept 2025 before being excluded from enacted FY2026 appropriations (canon REG-WISR-005, per KFF)
- No reimbursement anywhere — 26 CPT codes describe clinical AI as of Jan 2026, three Category I and all diagnostic (Bipartisan Policy Center, secondary), against 1,524 FDA AI authorisations from FDA's own list (primary) — 508:1 on a mixed primary/secondary basis (canon REG-CODE-001/002). Every dollar is discretionary operating budget
- Hyperscalers sell the substrate directly — Fortune's profile places Innovaccer against Microsoft, Google, AWS and Salesforce, each able to price data unification as a loss leader for cloud consumption
Moats, dependencies & bottlenecks
Moats
The unified data substrate (Gravity) and the cost of replacing it Normalising EHR, claims, CRM and HR/finance data across 200+ health systems and payers and 80M lives is hard and sticky — switching means a re-integration project, not a contract termination. But Epic sits underneath 43.7% of US acute-care hospitals and 56.9% of beds (canon CM-KLAS-01) and never has to unify anything.
Strong while it runs six performance years to 31 Dec 2031, politically exposed A federal payer as customer, a designated WISeR review position for Ohio / MAC jurisdiction J15 with prior-authorisation requests routable through the Innovaccer WISeR Provider Portal — not an exclusive channel, since providers may still submit direct to CGS or via esMD — and compensation tied to averted expenditures, in a model CMS says 'does not change Medicare coverage or payment policy' (canon REG-WISR-001/002/004). It is also a demonstration model: a committee-approved defunding amendment in Sept 2025 did not survive into enacted FY2026 appropriations, and the next cycle is a new roll.
High while the relationships hold Kaiser Permanente and Banner Health invested in the $275M Series F and appear among named deployments, alongside Ascension, Trinity Health, Prisma Health, Risant Health and Franciscan Health. Real distribution — and the related-party overlap that clouds reference independence.
Billing on completed prior authorisations and resolved denials produces an auditable outcome record — Risant Health's ~45 minutes to under one minute is the shape of it — and a challenger has to run pilots to produce a comparable one. It also transfers execution risk to the vendor: unfinished work is unbilled.
Five acquisitions in under three years, funded from ~$651M raised. But Commure carries a $7B mark on canon's private-cohort record (CM-COH-04) and is the better-capitalised consolidator in the same field, and roll-up capacity is a property of the next round, not of the business.
useful as speed a boundary a regulator can redraw Nothing in the stack is a regulated device, so no 510(k) clearance, De Novo grant or PMA approval is required. That is deployment speed, not exclusion — canon's FDA list runs 1,524 authorisations at 96.19% 510(k) (REG-FDA-001/003), and a category with no clearance requirement has no clearance barrier for anyone else either.
Dependencies
discretionary opex, not reimbursement No CPT code pays for data unification, an agent or a denial appeal (canon REG-CODE-001/002: 26 clinical-AI codes, three Category I, all diagnostic — Jan 2026, Bipartisan Policy Center, secondary — against 1,524 FDA authorisations from FDA's own list, primary; 508:1 on a mixed basis). Spend is justified against administrative cost, the first line cut when provider margin compresses.
Contract revenue AND market-forcing function WISeR is an Innovation Center model, not a statute; CMS-0057-F's 1 Jan 2027 API deadlines and CMS-0062-P's drug ePA proposal drive payer demand; and a defunding amendment cleared committee in Sept 2025 before being dropped. Three exposures to one agency.
Integration channel AND competitor Every agent needs the chart, the schedule and the claim. Epic held 43.7% of US acute-care hospitals and 56.9% of beds in 2025 on $6.7B of company-stated revenue and builds rather than buys; Oracle Health lost 56 hospitals and 14,676 beds in 2025 (canon CM-KLAS-03), a motive to bundle. Epic is un-hedgeable — no security exists (CF-EPIC-06).
GOOGL, AMZN) capability partner and competitor The 28 Oct 2025 collaboration builds on NeMo Guardrails and Framework, Riva Parakeet NIM, Triton Inference Server and TensorRT-LLM. Under per-task pricing, inference price IS gross margin. Microsoft's M12 is a Series F investor and Microsoft is named in the competitive set in Fortune's profile.
Revenue concentration Medium-high (unquantified) Seven of the top ten US health systems is a strength and a concentration at once. Concentration has never been disclosed, and two of the largest named accounts are also shareholders.
Financing / liquidity Canon PB-024: eleven named 2026-27 IPO candidates, none priced at 7 August 2026, no S-1 confirmed for any, and that any will price is not assertable. The CEO's own threshold is $400–500M ARR, roughly a doubling. Interim liquidity has come from the Series F's ~35% secondary tranche and a ~$75M ESOP buyback.
Advantages
- The only name on canon's eleven-candidate IPO watchlist with a revenue figure disclosed at a named source (~$200M ARR, Fortune, 23 Jul 2026)
- A federal contract with CMS as paying counterparty, running to 31 Dec 2031
- Distribution at the top of the market — 7 of the top 10 US health systems, 200+ health systems and payers, 80M patient lives
- One data substrate amortised across five agent categories, so the second and third product into an account carry almost no integration cost
- Pricing that bills on completed work rather than seats — the direct answer to the unproven-ROI objection canon records against seat-priced clinical AI
- Five completed acquisitions and demonstrated integration capacity in a consolidating category
Weaknesses
- No valuation verifiable at any named source — the defining fact canon isolates, and disqualifying for any mark-to-market
- ARR is the sole financial disclosure; margin, burn, runway, retention and concentration are all absent
- Growth that ran behind the company's own stated trajectory ($250M ARR 'this year' in Jan 2025 versus ~$200M crossed Jul 2026)
- A margin mix diluting by design — inference COGS plus an acquired labour-based services book
- Three restructurings in four years, the latest 340 roles in May 2026
- Structural conflict between the WISeR payer-side contract and the provider-side revenue-cycle business
Bottlenecks
- Data integration is still the gating cost — every new health system is a bespoke normalisation project against a system of record the vendor does not control, capping how fast a platform sale converts to platform revenue
- Mandatory human review sits on the highest-value workflow — CMS requires all WISeR non-affirmation recommendations be determined by licensed clinicians, and 42 CFR 422.566(d) has required physician review of adverse Medicare Advantage determinations since long before AI
- No disclosed unit economic — without a cost-per-completed-task against the price per task, neither buyer nor investor can tell whether transaction pricing expands or compresses margin at scale
- Services drag from the RCM acquisition — a managed-services book with ~4,000 providers only becomes platform margin if the agents genuinely displace the labour running it
- The company's own IPO bar — $400–500M ARR against ~$200M crossed in Jul 2026 means roughly two more years of 50%+ compounding, a long time to hold an unmarked private position
- Conflict management on WISeR — being paid a share of averted expenditures by the payer while selling revenue-cycle tooling to providers is a positioning problem no engineering resolves
Top signals & trends
Top signals
The most consequential disclosure on the watchlist — a named source, a rate and a base. Still ARR rather than GAAP, still unaudited, and still about half the CEO's own $400–500M IPO threshold.
Canon REG-WISR-001/002, primary at CMS. A federal payer as customer through 31 Dec 2031. KFF's analysis of 2024 Medicare claims puts the in-scope pool at $12.3B, 5.3% of traditional-Medicare Part B spending, of which skin substitutes alone are $10.3B (REG-WISR-003 — KFF attribution mandatory).
Canon PB-024 carries the date. Real distribution into revenue cycle and a real services-margin question in one transaction. Deal value undisclosed by the company; a trade-press figure circulates and is not asserted here.
Entrackr and Inc42, six days before the CaduceusHealth announcement. Framed by the CEO as a shift to 'AI-native' operations; alongside ~245 roles in Jan 2023 it reads as a cost base repeatedly outrunning revenue.
Strategically the right answer to the ROI objection and financially the riskiest: revenue now depends on agents finishing work, and inference cost lands in COGS at the moment of billing. No cost-per-task figure has been published.
Canon PB-024. Eleven candidates, none priced at 7 August 2026, no confirmed filing for any, and that any will price is not assertable. Innovaccer is excluded from canon's six-mark private cohort (CM-COH-04) precisely because its mark cannot be sourced.
Roughly $1.17B combined for the three closest public expressions of Innovaccer's own product lines. Aggregator-sourced quotes per canon's caveat. The market's standing verdict on the category Innovaccer leads privately.
Trends
Prior authorisation, denial management and patient access have a payer and a measurable cost baseline; canon's clinical record (CM-PHTI-02) is that financial ROI for clinical AI remains unproven to buyers. Innovaccer sits on the administrative side of that split.
CMS-0057-F: 7-day standard / 72-hour expedited decisions from 1 Jan 2026, four API mandates from 1 Jan 2027; CMS-0062-P extends electronic prior authorisation to drugs. Compliance spending is non-discretionary.
WISeR runs six performance years across six states with participants paid a share of averted expenditures (canon REG-WISR-001/004). Revenue for the vendor and a political exposure at the same time, with a committee-approved defunding attempt already on the record.
It makes ROI auditable and aligns vendor with buyer — and it puts inference cost directly into gross margin while making revenue depend on completion rates nobody publishes.
Epic at 43.7% of US acute-care hospitals and 56.9% of beds ships native analytics and AI and never has to solve data unification; the hyperscalers named in Fortune's profile can price the substrate as cloud pull-through.
Canon PB-065: Exact Sciences to Abbott (Mar 2026), iCAD to RadNet (Jul 2025), Sapiens to Advent (Dec 2025), Exscientia to Recursion, R1 RCM LBO'd (Nov 2024), ProAssurance to The Doctors Company (Jun 2026). Scarcity accrues to whoever lists first with real disclosure — though canon is equally clear no candidate had priced at 7 August 2026.
HTI-5 would amend 45 CFR 171.102 so 'access' and 'use' include autonomous AI systems (canon REG-IB-007), exposing refusal to a $1,327,209-per-violation ceiling. Proposed only, and canon records zero information-blocking enforcement actions to date.
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.
Collaboration announced 28 Oct 2025 across NeMo Guardrails, NeMo Framework, Riva Parakeet NIM, Triton Inference Server and TensorRT-LLM, for voice documentation, OCR and model training. Under per-task pricing this is a gross-margin dependency, not just a capability one.
M12 is a Series F investor and Azure a hosting and model supplier; Microsoft also appears in the competitive set named in Fortune's profile. Supplier, shareholder and competitor at once.
Google Cloud and AWS supply compute and reasoning models into the same category they sell into. No exclusive relationship disclosed.
Epic, Oracle Health, Meditech and payer claims feeds are the mandatory inputs to Gravity — and in Epic's and Oracle's case the suppliers of the raw material are also the competitors.
CMS requires all WISeR non-affirmation recommendations be determined by appropriately licensed clinicians — a regulated, non-automatable labour input to the federal contract.
WISeR participant for Ohio and MAC jurisdiction J15 CGS, 1 Jan 2026 – 31 Dec 2031, compensated on a percentage of expenditures associated with averted wasteful, inappropriate care. The only federal payer relationship on canon's watchlist.
Series F investor and named platform customer. Risant Health, Kaiser's non-profit health-system platform, is separately named with prior-authorisation turnaround falling from ~45 minutes to under one minute.
Series F investor and named customer — the second investor-customer, and the second source of related-party overlap in the reference list.
Ascension / Trinity Health / Prisma Health / Franciscan Health Named health-system deployments of the agent platform (MedCity News, 15 Apr 2026). Contract values and scope undisclosed.
Ambulatory provider groups via CaduceusHealth Nearly 4,000 providers with $5B of annual gross patient charges administered, acquired 21 May 2026 — the installed base Flow's revenue-cycle agents are deployed against.
95% of community pharmacies; 80M patient lives Company-stated reach (21 May 2026); Fortune states 80M patient records and 7 of the top 10 US health systems. Neither has an audited denominator.
Private and un-listable (canon CF-EPIC-06). Owns the system of record at 43.7% of US acute-care hospitals and 56.9% of beds in 2025 on $6.7B of company-stated 2025 revenue, and ships native analytics, registries and AI. The one competitor for whom Innovaccer's core problem — unifying data — does not exist.
Private, $7B mark on canon's private-cohort record (CM-COH-04, dated 19 May 2026). The best-capitalised consolidator assembling provider-workflow assets across documentation, patient messaging and RCM, chasing the same single-operating-layer pitch. Also on canon's eleven-name IPO watchlist.
The listed revenue-cycle-automation incumbent Flow now competes with directly. Q2 2026 revenue $319.7M +18.1%, adjusted EBITDA $136.7M, FY2025 adjusted EBITDA margin 42.04% (canon PB-018), at a $4.05B market cap — profitable, disclosed, and roughly 6x Innovaccer's ARR base.
Private prior-authorisation and utilization-management specialist, and a fellow CMS WISeR participant assigned Texas to Innovaccer's Ohio (canon REG-WISR-002). Direct competitor in the single workflow where Innovaccer holds a federal contract.
Foundry is sold into health systems as exactly the data-unification-plus-operational-workflow layer Gravity claims, with far greater capital and a federal-contracting track record. Not a healthcare pure-play, which is why it is absent from canon's pure-play cohort.
The closest listed analog to Innovaccer's original business — a healthcare data platform with analytics and improvement services. $154M market cap at the 31 Jul 2026 close (canon CM-COH-01, aggregator-sourced). Less a competitive threat than the market's priced verdict on the category.
Private population-health and value-based-care data platform, historically Innovaccer's most direct head-to-head in health-system analytics selections. Narrower product surface, no comparable agent layer disclosed.
The #2 US EHR at 21.9% share, losing 56 hospitals and 14,676 beds in 2025 (canon CM-KLAS-01/03) — precisely the motive to bundle analytics and agents into the EHR contract at aggressive pricing.
Listed value-based-care and specialty-utilization-management services, $348M market cap at the 31 Jul 2026 close. Competes for the same value-based-care budget and shows what the public market pays for services-weighted revenue here.
Listed patient intake, access and engagement — the direct public analog to Flow's scheduling and patient-engagement front end. $663M market cap at the 31 Jul 2026 close.