Nourish
Fee-for-service claims to third-party payers. The revenue event is a billed clinical encounter — a medical-nutrition-therapy assessment (CPT 97802, per 15 min) and its follow-ups (97803; 97804 group; G0270/G0271 on a change in condition) — adjudicated by the patient's commercial, Medicare Advantage or Medicaid managed-care plan, with ~94% of patients paying $0 out of pocket (company-reported, Apr 2025). Patients arrive direct-to-consumer through search and app plus referral from what the company describes as tens of thousands of providers at 250+ health systems. Dietitians were W-2 employees at the Series B; whether that holds across the 10,000+ network claimed at the Series C is undisclosed. There is no software licence, no per-seat SaaS line and no per-clinician subscription: the unit sold is a clinician-hour reimbursed at an administered rate. This is a labour business with software attached, not a software business with labour attached.
Earnings, margins, COGS & capex
The financially distinctive fact about Nourish is not its AI — it is that a bill gets paid. Every encounter maps to a procedure-code family that has existed for two decades (97802 / 97803 / 97804, plus G0270/G0271), so the company sells into a claim line rather than a discretionary budget. That is exactly the test the board's thesis applies — canon counts 26 clinical-AI CPT codes at Jan 2026, only 3 Category I and all diagnostic, a 508:1 ratio of FDA AI authorisations to Category I codes (REG-CODE-001/002) — and it is the test nearly every AI-first health company on the board fails. Nourish passes it by not being an AI company: the codes it bills pay a human clinician for her time. The corollary is the whole bear case. A code that already exists has a price already set by someone else, in fifteen-minute units, for a service the payer treats as preventive. Revenue therefore scales with clinician-hours and payer mix, not software leverage, and the AI's entire financial job is to lift sessions per clinician-hour and cut no-shows and documentation drag. Nothing — revenue, margin, rate, clinician cost, utilisation — is disclosed, so the $1.75B has no computable multiple, unlike Abridge (~$117M contracted ARR against $5.3B) or OpenEvidence (~$300M annualised at Jul 2026 against $12B, canon CM-PRIV-OE-03). The one honest triangulation is the listed comparable: Omada Health (OMDA) did $283.30M of TTM revenue growing 49.4%, guides $312-322M for 2026, reports 1M+ members and positive adjusted EBITDA in Q1 2026, and is capitalised at $1.18B in canon's 31 Jul 2026 cohort — about 4.2x trailing sales. Nourish's reported mark is ~1.48x Omada's entire market cap with no revenue published at all. A private mark can be sanity-checked against a public price; it cannot be validated by one.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~11¢ is cost of goods and ~0¢ operating expense, leaving ~89¢ of operating profit.
Revenue trend
Margins
Provider-shaped, not software-shaped: the clinician hour is COGS against an administered per-encounter rate. Improves only via sessions per clinician-hour, no-show rate, clean-claim rate and payer mix; degrades with RD wage inflation or any downward rate action.
Two heavy pre-revenue lines — clinician recruiting/credentialing and payer contracting — plus consumer acquisition.
Claims working capital: payer receivables and denial rework sit behind revenue. Single-plan claim edits are a live cash risk — Community First Health Plans denies 97802/97803 without a diagnosis from 3 Aug 2026 on its Texas Medicaid and CHIP lines, one regional plan acting alone by bulletin.
COGS structure
The cost of goods sold is a licensed human being's hour, and that is the most important financial fact here. A registered dietitian nutritionist is a credentialed clinician (CDR registration, state licensure, payer credentialing per plan and per state), and Nourish described its dietitians as W-2 employees at the Series B — so salary, payroll tax and benefits sit in COGS whether or not the schedule fills, and idle capacity is a direct margin leak rather than a marketplace's problem. Against that sits a per-encounter reimbursement the company does not set: MNT bills in fifteen-minute units (one unit = 8 minutes minimum), so a 60-minute intake and a 30-45-minute follow-up convert to a bounded number of units at the plan's fee schedule. No rate is published by the company and none is asserted here. Secondary lines are ordinary — cloud and LLM inference for the copilot and the patient app's meal tracking and agent features (small against payroll), wearable and lab integrations, and the revenue-cycle apparatus of eligibility checks, submission, denial rework and appeals, which for a claims business is a real cost centre. The AI's mandate follows: every point of no-show reduction, every minute of documentation removed and every point of clean-claim rate converts one-for-one into gross margin, because the denominator — the paid clinician hour — is fixed.
Capex
Negligible on the balance sheet: no PP&E, no owned compute, software and cloud in opex. The functional equivalent is standing up supply and access ahead of demand — recruiting, credentialing and licensing thousands of dietitians across 50 states, and negotiating in-network status behind 200M+ covered lives. Both are front-loaded, both take months to a year to convert, and both are why a business with a scheduled-hour marginal cost needs $100M.
Latest earnings
n/a — no guidance issued, no consensus exists
No financial guidance of any kind. Forward statements are directional: scale the RD network, deepen plan and health-system partnerships, build out the 'AI-native metabolic clinic' positioning introduced at the Series C. Clinical outcomes are published as averages with no control arm, sample size, follow-up window or peer review: 8% body-weight reduction, 1.3-point A1C reduction, 31-point LDL reduction, 23-point systolic BP reduction, and 'over $2,000 per patient in annual cost savings for health plans' (company-reported, 19 May 2026).
- Last mark
- ~$1.75B post-money reported (Series C, $100M, Menlo Ventures lead, 19 May 2026) — carried by Axios Pro; NOT in the company's own release
- Implied dilution
- $100M primary ≈ 5.7% of a $1.75B post — a negotiated headline off a small slice, not a clearing price
- Prior mark
- >$1B reported by Fortune at the $70M Series B (23 Apr 2025) — also not company-stated. Implies ~1.7x in ~13 months
- Total raised
- $215M — $8M seed (Feb 2023) + $35M Series A (27 Mar 2024) + $70M Series B (23 Apr 2025) + $100M Series C (19 May 2026)
- Clinician network
- 10,000+ RDs (19 May 2026) from 3,000+ (23 Apr 2025) — ~3.3x in thirteen months
- Covered lives
- 200M+ contracted across commercial, Medicare and Medicaid plans, all 50 states
- Referral footprint
- Tens of thousands of providers across 250+ health systems
- Patient cost share
- ~94% pay $0 out of pocket (company-reported, Apr 2025)
- Engagement
- Hundreds of thousands of monthly active app users; 'millions of appointments'
- Revenue / ARR
- Not disclosed anywhere — no multiple computable against the $1.75B
- Listed comparable
- Omada Health (OMDA): $1.18B cap on $283.30M TTM revenue +49.4% (~4.2x sales) — Nourish's mark is ~1.48x Omada's entire market cap
- Founded / founders / base
- 2021; co-founders Aidan Dewar (CEO), Sam Perkins, Stephanie Liu; Y Combinator alumnus. Headquarters UNRESOLVED — MedCity says New York City, Tracxn says Austin
Growth drivers
- GLP-1 volume — the strongest tailwind. Every patient starting semaglutide or tirzepatide is a candidate for nutrition support (muscle preservation, protein and micronutrient adequacy, GI tolerance, post-taper maintenance); Nourish claims GLP-1 patients lost 33% more weight when paired with one of its RDs (company-reported Apr 2025, no published methodology)
- Payer breadth — 200M+ contracted covered lives across commercial, Medicare Advantage and Medicaid managed care; each new plan converts a cash-pay prospect into a $0-out-of-pocket one
- Referral supply — tens of thousands of providers across 250+ health systems turns acquisition from paid consumer marketing into a clinical referral graph a rival cannot buy
- Clinician capacity — in a clinician-hour business, 10,000+ RDs added ahead of demand IS the growth mechanism
- Condition expansion beyond weight — published outcomes span A1C, LDL and systolic BP, and the earliest positioning covered ~40 conditions including eating disorders, perinatal, renal, oncology and GI, widening the billable population past the GLP-1 cohort
- AI leverage on a fixed input — the RD copilot (note automation, insight surfacing) and the patient-side agent raise revenue per clinician-hour rather than adding a revenue line
Bull & bear
Nourish is the rarest thing on this board: a health-AI-adjacent business whose bill actually gets paid, by an insurer, under codes that already exist, for a service the patient wants and the plan can price. It sits downstream of the largest new spending event in US chronic care, it has assembled the two assets that cannot be shortcut — in-network contracts over 200M+ lives and 10,000+ credentialed clinicians — and it aims AI where AI is worth money in this model: throughput per clinician-hour, not a clinical claim needing FDA review.
- Payer identity is the board's whole thesis, and Nourish's payer is the insurer's claims system. Canon puts the clinical-AI code inventory at 26 CPT codes with only 3 Category I, all diagnostic, and a 508:1 authorisation-to-code ratio (REG-CODE-001/002); the MNT family has been paying for twenty-plus years and needs no new coverage decision
- It is not selling into the broken buyer. Canon's provider margin is 2.5% median CYTD at April 2026 (CM-KH-01), which is what makes an IT-budget sale fragile — Nourish's revenue arrives from the plan, and the health system's role is to refer patients it cannot staff for
- The GLP-1 wave creates demand Nourish does not pay for: every prescription written by someone else is a qualified referral, and the company reports GLP-1 patients paired with its RDs losing 33% more weight — nutrition support is the annuity underneath a drug people stop taking
- The moats are the slow ones. Contracting 200M+ lives and credentialing 10,000+ multi-state clinicians is quarters-to-years of unglamorous work that no better model or cheaper API shortens — which is precisely why this is not the ambient-scribe market where roughly 60 vendors compete (canon CM-PHTI-01)
- AI is aimed at the right variable. Because the paid clinician hour is the fixed denominator, documentation automation, no-show reduction, between-visit engagement and clean-claim rate each convert straight to gross margin — a narrow, measurable, unregulated use with no 510(k) clearance, De Novo grant or PMA approval implicated
- H.R. 7148 removed the near-term policy overhang: home-based and audio-only MNT are covered with no geographic or originating-site restriction through 31 Dec 2027, giving the delivery model two years of certainty it did not have in 2025
Strip the AI language and this is a staffing company billing fifteen-minute units at a rate someone else sets, marked at $1.75B off $100M of primary by a single lead, with no revenue figure, no margin figure and no audited outcome anywhere. The two things that make it work — payer coverage and GLP-1 volume — are both other people's decisions, and the field behind it is full of funded rivals whose clinicians are not on payroll.
- There is no denominator. No revenue, ARR, gross margin, rate, utilisation or burn is disclosed, and the single growth statistic does not say what tripled. A mark that cannot be falsified cannot be defended — and the $1.75B itself lives in one named outlet's reporting, not the company's release
- The margin is capped by arithmetic. Sessions bill in fifteen-minute units against an administered schedule while the dietitian is a paid W-2 hour; no copilot converts that into software economics, and every hire ahead of demand is idle payroll. The honest comparison is a virtual clinic, not Doximity's 89.1% gross margin (canon CF-DOCS-07)
- Coverage is a policy variable, not an asset. Medicare MNT pays 3 hours in year one and 2 hours a year after, only for diabetes, CKD or a transplant within 36 months, and only on a physician's referral an NP or PA cannot write. Commercial coverage is revisable at renewal, single plans tighten claim edits by bulletin, and telehealth rests on an extension expiring 31 Dec 2027
- The GLP-1 tailwind is borrowed and contested. Employers and plans are actively restricting anti-obesity drug coverage on cost, prescribers are building nutrition support in-house, and the cohort Nourish serves best exists because somebody else's benefit design allowed it
- Competition is dense and structurally cheaper. Fay raised $50M at a reported $500M and Berry Street $50M, both on independent-practitioner models with no payroll on unfilled slots; Foodsmart serves 2.2M members with food-benefit integration; Culina Health is payer-backed; Omada and Hinge sell cardiometabolic and MSK into the same committees with public balance sheets
- The exit path is unproven and Nourish is not even on canon's list. PB-024-cash-catalyst names eleven 2026-27 digital-health IPO candidates, NONE priced and no S-1 confirmed — and Nourish is not among them. Meanwhile canon's taxonomy ruling (CM-COH-03) is that telehealth platforms like TDOC and AMWL are not health-AI pure plays, and TDOC's $1.22B cap at the 2026-07-31 close is what the public market pays for a scaled virtual-care business it has stopped believing
What it is worth
Last-round post-money mark only, and even that is reported rather than company-disclosed — no public price, no audited statement, no revenue figure, therefore no computable multiple. Triangulated three ways: implied dilution on the round; the listed comparable (Omada Health, an adjacent virtual cardiometabolic business with audited revenue and public guidance); and the private peer set, where Fay's reported $500M is the only other priced mark in insurance-covered virtual nutrition. Explicitly NOT triangulated against canon's private cohort — Nourish is not a component of CM-COH-04 and is not added to it.
A material markdown, and the arithmetic is unforgiving: the reported $1.75B is ~1.48x the entire market capitalisation of Omada Health, a listed peer doing $283.30M of audited revenue growing 49.4% at a $1.18B cap (canon CM-COH-01, 31 Jul 2026 close), about 4.2x trailing sales. Holding the private mark at the public multiple needs roughly $420M of Nourish revenue, and the company has never published a dollar. From there any two of the following do it: a large national plan cuts MNT rates or narrows its network at renewal; employer and plan restrictions on GLP-1 coverage thin the referral funnel; plan-level claim edits and the 31 Dec 2027 telehealth expiry bite collection and then delivery; RD wage inflation compresses a gross margin that was never software-shaped; Fay and Berry Street take the supply side with a structure that does not pay for idle hours. When private capital tightens, the marks with no denominator reprice first — and the public tape has already shown what it pays for scaled virtual care it stops believing: Teladoc at $1.22B.
Roughly holds $1.75B. Capacity and covered lives compound, GLP-1 referral volume stays healthy, the rate environment is stable through the current telehealth extension — but revenue stays undisclosed, gross margin stays provider-shaped, and no liquidity event materialises. The next priced event is another private round or a secondary at a similar or modestly higher level, set again by a small primary slice rather than by a market.
Materially above $1.75B. The 'more than tripled' is revenue, the business is genuinely profitable, and the metabolic-clinic repositioning moves billing up the acuity curve from fifteen-minute MNT units toward chronic-care management and risk-bearing contracts where price is negotiated rather than administered. 10,000+ credentialed clinicians and 200M+ contracted lives become infrastructure GLP-1 platforms, employers and plans rent rather than rebuild; independent actuarial validation of the '$2,000 per patient in plan savings' claim converts the outcomes deck into contract leverage; and a first credible listed pure-play in insurance-covered virtual nutrition re-rates the private set the way canon expects a listed agentic vendor to re-rate the health-AI cohort.
~$1.75B post-money on a $100M Series C led by Menlo Ventures, announced 19 May 2026, with Thrive Capital, Index Ventures, J.P. Morgan Growth Equity Partners, Maverick, Y Combinator, BoxGroup, Atomico, Daybreak and Operator Partners participating; total raised $215M. THE VALUATION IS RELAY-GRADE: Axios Pro's 19 May 2026 item is the only named outlet carrying it and the company's own release states the round and syndicate but not the price. Treat it as negotiated between one lead and one board — $100M of primary is ~5.7% of a $1.75B post, so a small slice of paper sets the whole headline, the same defect canon flags on Sword Health's $4B off a $40M round (CM-COH-05) and Hippocratic's $3.5B off ~3.6% dilution. A raise size never implies a valuation and none is inferred here. No revenue multiple is computable: the company has never published a dollar across four funding announcements, and its only growth statistic does not name its metric. The discipline available is the public tape, and it is sharp. Omada Health carries a $1.18B market cap in canon's cohort at the 31 Jul 2026 close (CM-COH-01; independently $1.16B at $19.54 on the 6 Aug close) on $283.30M of TTM revenue growing 49.4% — roughly 4.2x trailing sales for an audited, adjusted-EBITDA-positive business with 1M+ members guiding $312-322M. Nourish's reported $1.75B is ~1.48x that entire market cap; read back through Omada's multiple it implies something on the order of $400-430M of Nourish revenue, a figure no source supports and the company has never published. The point of the comparison is that gap, not the estimate. Hinge Health sits at $5.78B and Teladoc at $1.22B in the same cohort, with canon ruling Teladoc a telehealth platform rather than a health-AI pure play (CM-COH-03) — the same taxonomy question this company raises. On the private side Fay's reported $500M (Feb 2025) is the nearest priced peer, and Nourish's mark is ~3.5x it on a network roughly 10x the size. Liquidity is unresolved: $215M raised, no disclosed profitability, no S-1, and canon's eleven-name 2026-27 digital-health IPO slate (PB-024-cash-catalyst) has produced no priced deal and does not include Nourish. Realistic paths are another private round, a secondary, or a strategic acquirer. Not investment advice and not a price target.
SWOT
Strengths
- It bills an existing claim line. Nutrition care maps to MNT codes that predate clinical AI by two decades — against canon's 3 Category I clinical-AI CPT codes and 508:1 authorisation-to-code ratio (REG-CODE-001/002), a company whose payer is the insurer's claims system is structurally unlike every ambient or agentic vendor selling into a provider IT budget
- Payer contracts covering 200M+ lives across commercial, Medicare Advantage and Medicaid in all 50 states — the slowest, least glamorous, least copyable asset in the business, and the reason ~94% of patients pay nothing
- The largest claimed employed dietitian network in the US at 10,000+, grown ~3.3x in thirteen months — where the constraint is credentialed clinician-hours, supply IS the product
- A referral graph rather than a pure acquisition funnel — tens of thousands of providers across 250+ health systems, lowering acquisition cost and conferring clinical legitimacy a consumer app cannot buy
- Positioned downstream of the GLP-1 wave with a company-reported 33% weight-loss uplift for GLP-1 patients paired with an RD, and a widening condition set (A1C, LDL, blood pressure) that outlives any one drug cycle
Weaknesses
- Zero financial disclosure — no revenue, ARR, gross margin, reimbursement rate, RD cost, utilisation, burn or runway. The $1.75B has no denominator, and the one growth number given does not name its metric
- The margin ceiling is structural, not executional — a W-2 clinician hour against an administered per-unit fee cannot produce software economics; AI improves throughput at the edges without changing the shape of the cost curve
- Price-taker on both sides — Nourish sets neither the fee schedule nor the covered indications, so one large plan cutting MNT rates, tightening visit limits or narrowing its network reprices a large share of revenue unilaterally
- Concentration in the GLP-1 narrative — much of the 2025-26 re-rate drafts on drug volume, and coverage retrenchment or a shift to cheaper orals hits the funnel before it hits the clinical need
- Clinical evidence is self-published averages — 8% weight loss, 1.3-point A1C, 31-point LDL, 23-point systolic, '$2,000 per patient' savings — with no control arm, sample size, follow-up window or peer review, which is the evidence quality a payer's own actuaries discount
Opportunities
- Become the nutrition rail rather than a nutrition provider — 10,000+ credentialed, multi-state, in-network RDs is infrastructure GLP-1 platforms, employers and plans would rent rather than rebuild
- Move up the acuity curve — the 'AI-native metabolic clinic' framing points at higher-value billing (chronic care management, remote monitoring, prescriber-adjacent services) than fifteen-minute MNT units
- Value-based contracts. If the '$2,000 per patient in plan savings' claim survives independent actuarial review, fee-for-service units can be traded for shared savings — the only path from an administered rate to a negotiated one
- Medicaid and Medicare Advantage expansion where nutrition-sensitive disease concentrates and quality measures reward intervention — the segment Elevance, Highmark and Kaiser food-as-medicine programmes have legitimised
- Health-system channel depth — 250+ systems already refer, and systems at a 2.5% median CYTD operating margin (canon CM-KH-01, April 2026, incl. allocations) have every reason to outsource an outpatient dietetics service they staff at a loss
Threats
- Reimbursement action. Medicare MNT is hard-capped — eligibility limited to diabetes, CKD or transplant within 36 months, 3 hours in year one and 2 hours per year after, and only a physician can refer (an NP or PA cannot). Commercial coverage is plan-by-plan and revisable at renewal, and a single plan can move first: Community First Health Plans denies 97802/97803 without a diagnosis from 3 Aug 2026 on its Texas Medicaid and CHIP lines
- Telehealth cliff. Home-based and audio-only MNT rests on the flexibilities H.R. 7148 (signed 3 Feb 2026) restored retroactively and extended only through 31 Dec 2027; a lapse strikes the delivery model, not merely a rate
- A crowded, well-funded field with a lower-cost structure — Fay ($50M Series B at a reported $500M, Feb 2025, Goldman Sachs-led) and Berry Street ($50M, Northzone-led, 1,000+ RDs) both run independent-practitioner models with no payroll on idle hours
- The listed comparable is a discipline. Omada Health is capitalised at $1.18B on $283.30M of TTM revenue growing 49.4% — an audited, adjusted-EBITDA-positive business at ~4.2x trailing sales, against which a $1.75B private mark with no disclosed revenue, at ~1.48x Omada's whole market cap, is a bet rather than a comparison
- RD wage inflation and finite supply. Tripling a credentialed clinical workforce in thirteen months bids against health systems for the same pool; wage growth lands straight in COGS with no offsetting rate increase available
Moats, dependencies & bottlenecks
Moats
contracts renew, and renewal is where rates get cut The hardest asset to replicate and the reason ~94% of patients pay $0: every plan is a separate negotiation, credentialing cycle and fee schedule across 50 states. But it is a moat against entrants, not against the counterparty — the plan that grants in-network status sets the rate and can narrow the network at renewal.
Strong on scale, ambiguous on lock-in Supply is the product in a clinician-hour business, and 3,000+ to 10,000+ in thirteen months is real operational capability. The ambiguity is retention and cost: dietitians are portable, rivals recruit from the same credentialed pool, and if the W-2 structure described at the Series B still holds, idle capacity is a direct margin leak.
250+ health systems, tens of thousands of providers A referral relationship is stickier than a paid click and cheaper than both, and it solves the credibility problem a consumer nutrition app cannot. Systems at a 2.5% median CYTD operating margin (canon CM-KH-01) have a standing reason to outsource outpatient dietetics rather than staff it.
weak until independently validated Millions of appointments with linked wearable and lab data is a genuinely scarce nutrition dataset and is what makes the copilot and patient agent work. But the published outcomes are uncontrolled company averages; until a payer's actuaries or a peer-reviewed study reproduce them, it is an operating advantage, not a commercial moat.
Nourish captures 'dietitian covered by insurance' intent directly and runs an app with hundreds of thousands of monthly actives, which feeds the referral flywheel. But search distribution is rentable, AI answer engines are re-cutting the channel, and Fay and Berry Street compete for the same intent.
Dependencies
Medicare Advantage and Medicaid plans (UNH, ELV, CI, CVS/Aetna, CNC, HUM and the Blues) Revenue counterparty — the payer of essentially every encounter The fourth payer that makes the business work and the single point of failure that could unmake it. Rate setting, covered indications, visit limits, prior authorisation, network breadth and claim-edit policy are all the counterparty's unilateral levers. Concentration by plan is undisclosed, which is itself a gap.
state Medicaid, plan medical policy) Regulatory / coverage Medicare MNT is diagnosis-restricted (diabetes, CKD, transplant within 36 months), hour-capped (3 hours year one, 2 hours after) and physician-referral-only — an NP or PA cannot refer. Home-based and audio-only delivery rests on H.R. 7148, signed 3 Feb 2026 and running only to 31 Dec 2027. Plans tighten independently: Community First denies 97802/97803 without a diagnosis from 3 Aug 2026 on its Texas Medicaid and CHIP lines. None of it is Nourish-specific and none is in Nourish's control.
LLY, prescribers and PBMs) Demand generation The strongest source of qualified patients is a drug someone else prescribes and a benefit someone else funds. Employer and plan restrictions on anti-obesity medication, or a shift to cheaper orals with a different support profile, change the funnel before they change the clinical need.
Core input / COGS The US credentialed RDN pool is finite and Nourish bids for it against health systems, Fay, Berry Street, Foodsmart and every employer programme. Wage inflation lands directly in COGS with no matching rate increase available, and multi-state licensure plus per-plan credentialing add months between hiring and billing.
$215M raised, no disclosed profitability, no S-1, and canon's eleven-name 2026-27 digital-health IPO slate (PB-024-cash-catalyst) has produced no priced deal and does not include Nourish. Realising $1.75B needs another private round, a secondary, or a strategic buyer — a plan, a PBM, a GLP-1 telehealth platform or a listed virtual-care consolidator.
Advantages
- The payer is the insurer, under codes that already exist — the structural condition the board says determines where margin accrues, and the one most AI-first health companies cannot meet
- Two slow assets already built: 200M+ contracted covered lives and 10,000+ credentialed multi-state clinicians
- ~94% of patients pay $0 out of pocket, removing the biggest conversion barrier in consumer health — a direct consequence of the contracting work
- A referral base of 250+ health systems and tens of thousands of providers, lowering acquisition cost and conferring clinical legitimacy
- AI applied to the one variable that pays — output per paid clinician hour — with no device pathway and no 510(k) clearance, De Novo grant or PMA approval implicated by the product as described
Weaknesses
- No disclosed revenue, margin, rate, utilisation or burn — the $1.75B has nothing to test it against, and the mark is reported rather than company-stated
- Gross margin is capped by the arithmetic of a W-2 clinical hour against an administered per-unit fee and cannot reach software economics
- Price-taker on rates and coverage; renewal is a downside-only event on the rate line
- Clinical evidence is uncontrolled, self-published averages with no sample size, follow-up window or peer review
- Heavy narrative dependence on the GLP-1 cycle, whose coverage is being actively restricted by the employers and plans that fund it
Bottlenecks
- Clinician supply, not demand — revenue is capacity x utilisation x rate, and capacity is credentialed human hours that take months to recruit, license across states and enrol with each plan
- The administered rate itself — MNT pays in fifteen-minute units at a price Nourish does not set, leaving session count, no-show rate and payer mix as the only controllable variables
- Eligibility and referral gating — coverage is condition-linked, Medicare requires a physician referral an NP or PA cannot write, and plans tighten claim edits by bulletin — making upstream diagnosis and referral capture a revenue-integrity problem, not a back-office one
- Patient adherence and drop-off. Nutrition-counselling attrition after the first or second session is the classic failure mode; lifetime value lives in visits four through twelve, which is exactly what the patient-side AI agent exists to defend
- Multi-state licensure and per-plan credentialing — the tax every scaled virtual medical group pays, growing linearly with headcount rather than amortising like software
Top signals & trends
Top signals
Thirteen months after the Series B, with J.P. Morgan Growth Equity Partners, Thrive, Index, Maverick, Y Combinator, BoxGroup, Atomico, Daybreak and Operator Partners following on. $100M on a $1.75B post is ~5.7% dilution — a negotiated mark, and the valuation appears in Axios Pro's reporting rather than the company's own release.
In a clinician-hour business this is the closest observable proxy for billable capacity — ~3.3x in thirteen months, consistent with the unnamed 'more than tripled' claim. It says nothing about utilisation of that capacity, which is undisclosed.
A company that discloses appointment counts, covered lives, headcount and four clinical outcome deltas but never a dollar has made a choice. Relays describe it as profitable; no company statement and no named outlet's own reporting confirms that, so it is not carried.
Reads as a deliberate move up the acuity and billing curve toward cardiometabolic management — and as the language a 2026 fundraise needs. The published A1C, LDL and systolic outcomes support the clinical half of the claim; nothing published supports the economic half.
The category is funded and the cheaper cost structure sits with the challengers. Nourish's employed-clinician model buys quality control and pays for idle capacity; a marketplace does the reverse.
Canon CM-COH-01 at the 31 Jul 2026 close; independently $1.16B at $19.54 on the 6 Aug close. Nourish's reported $1.75B is ~1.48x that entire market cap with zero published revenue. Alongside canon's CM-COH-03 ruling that TDOC and AMWL are telehealth platforms rather than health-AI pure plays, it is the reminder that the public multiple for scaled virtual care is not the private one.
Trends
Muscle preservation, protein and micronutrient adequacy, GI tolerance and post-taper maintenance are real needs a drug alone does not address. Omada scaling past 150,000 GLP-1 members and launching prescribing plus a flex-care option is independent evidence the companion-care category is being funded, not just pitched.
Elevance, Highmark and Kaiser Permanente programmes have legitimised nutrition as a covered plan intervention, which is what turns a wellness line item into a claim. It also invites plans to build in-house or consolidate onto one national vendor — a dynamic that cuts both ways.
Employer and plan restrictions on GLP-1 coverage compress the referral funnel while plans independently raise the claim bar — Community First Health Plans denies 97802/97803 without a diagnosis from 3 Aug 2026 on its Texas Medicaid and CHIP lines, one regional plan acting alone. Both are counterparty decisions arriving without negotiation.
H.R. 7148, signed 3 Feb 2026, retroactively restored Medicare telehealth flexibilities through 31 Dec 2027 — no geographic or originating-site restriction, home-based MNT covered, audio-only permitted. Two years of certainty is a real tailwind for a model that assumes the patient is at home; that it is two years and not permanent is the structural risk execution cannot offset.
Canon's evidence base is that AI's measured value in care delivery is time and burden, not diagnosis — PHTI's March 2025 verdict was burnout relief with financial ROI unproven (CM-PHTI-02), and the JAMA multisite anchor monetises only marginal E/M revenue. A business whose AI targets clinician throughput is aligned with what the evidence supports.
Canon marks TDOC at $1.22B and AMWL at $0.18B in the 2026-07-31 cohort and rules both out as health-AI pure plays (CM-COH-03). Omada's audited $283.30M TTM at a $1.18B cap is the fair comparable. Private marks in this segment are being set against a public tape that has already discounted the category once.
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.
Credentialed registered-dietitian labour pool The primary input and the primary cost — CDR-registered RDNs with state licensure and per-plan credentialing; 10,000+ claimed as of 19 May 2026, described as W-2 employees at the Series B. Contested with health systems, Fay, Berry Street and employer programmes.
Hosting and inference behind the RD copilot (note automation, insight surfacing) and the patient-side agent and meal tracking. Azure = MSFT, GCP = GOOGL, AWS = AMZN. Small against clinician payroll, and falling.
The app integrates wearables and labs, which is what makes the outcomes dataset (A1C, LDL, weight, blood pressure) possible. Inputs from continuous-glucose and device vendors (DXCM, ABT) and lab networks (DGX, LH); no exclusive relationship disclosed.
Eligibility verification, claim submission, denial management and appeals — the machinery turning a completed session into collected cash. Built or bought (Waystar, Availity, Optum), it is a cost centre a SaaS company does not carry, and plan-level claim edits raise its stakes.
The economic customer for most encounters. Nourish states partnerships with 'hundreds of the nation's leading health plans' behind 200M+ covered lives; individual names and any revenue concentration are undisclosed. Peers in the set: ELV, CI, CVS/Aetna, HUM, CNC and the Blues.
Medicare Advantage and Medicaid managed-care plans Named since the Series B. These are the populations where nutrition-sensitive chronic disease concentrates and quality measures reward intervention — and where MNT coverage rules are tightest.
Hundreds of thousands of monthly active app users; ~94% pay $0 out of pocket. They choose the service and drive retention — which is why patient-side AI engagement is a revenue feature — but they do not settle the bill.
250+ health systems and tens of thousands of providers referring. Functionally a distribution channel rather than a paying customer, but for systems at a 2.5% median CYTD operating margin (canon CM-KH-01), outsourcing outpatient dietetics is a cost decision as much as a clinical one.
Employers (via plan and point-solution channels) The buyer Omada, Hinge and Foodsmart sell to directly. Nourish's disclosed route runs through plans and referrals rather than direct employer contracting — a narrower but less churn-prone channel.
The most direct rival: insurance-covered virtual dietitian care on an independent-practitioner marketplace rather than employed clinicians. $50M Series B at a reported $500M valuation (Feb 2025, Goldman Sachs-led with General Catalyst and Forerunner), ~$75M total after a $25M Series A. Lower fixed cost per unfilled slot; Nourish's own trade coverage names it as the comparison.
1,000+ registered dietitians plus practice-management tooling for independents, founded 2023; $50M raised (Northzone-led, with Sofina, FJ Labs and operator angels) in the same window as Fay's round. Attacks the supply side — if RDs prefer running their own book on Berry Street's rails, Nourish's recruiting cost rises.
The listed comparable and sharpest reference point. Virtual cardiometabolic and GLP-1 companion care sold to employers and plans; $283.30M TTM revenue +49.4%, 2026 guide $312-322M, 1M+ members, 150,000+ on GLP-1s, positive adjusted EBITDA in Q1 2026, $1.18B market cap in canon's 31 Jul 2026 cohort. Audited, priced daily, selling into the same benefits committee — and worth two-thirds of Nourish's private mark.
Telenutrition plus food-benefit management serving 2.2M members across employer plans, Medicaid managed care, Medicare Advantage and commercial insurers; ~$200M raised. The most payer-native competitor and the one bundling food logistics with counselling — a wider contract than Nourish's.
Insurance-covered virtual clinical nutrition, notable as an early payer-backed entrant; ~$20M total raised after a $7.9M Series A. Sub-scale against Nourish but validated by the same coverage mechanism, and named alongside Fay in Nourish's own trade coverage.
'Digital food pharmacy' — condition-specific meal planning with ingredient and prepared-food delivery, ~$34M Series A. Competes for the same food-as-medicine plan budget from the logistics side rather than the clinician side.
Not a nutrition competitor, but the category benchmark for a listed clinician-plus-software virtual specialty clinic sold to employers and plans; canon carries it at $5.78B at the 2026-07-31 close (CM-COH-01). Sets the public expectation for what this model is worth at scale and competes for the same benefits-committee slot.
The cautionary comparable more than the live threat: a scaled virtual-care incumbent with chronic-condition assets, carried by canon at a $1.22B market cap at the 2026-07-31 close and explicitly ruled a telehealth platform rather than a health-AI pure play (CM-COH-03). What the public market pays for undifferentiated virtual care once it stops believing the growth story.
They own the moment of prescription — the top of Nourish's funnel — and each has an obvious incentive to attach nutrition support rather than refer it out. Mostly cash-pay rather than claims-billed today, which is the distinction Nourish must defend.
Elevance, Highmark and Kaiser Permanente food-as-medicine initiatives, plus health systems' existing outpatient dietetics departments. The insourcing option: the counterparty that pays Nourish's claims can staff the service instead, and the systems that refer already employ dietitians.