Qure.ai
B2B and B2G licensing: per-site or per-X-ray-unit annual licences and per-study pricing sold to public screening programmes (TB, lung cancer), ministries, donors and NGOs, imaging chains and hospitals, plus pharma-funded disease-awareness programmes (AstraZeneca's Lung Ambition Alliance) and OEM/PACS channel deals. In the US it is the standard radiology-AI model — a per-facility or per-study subscription from an operating budget, with no dedicated CPT code retrieved for any Qure.ai product at 7 August 2026.
Earnings, margins, COGS & capex
Clearance-rich, revenue-light, and losing money faster than it grows. On the only revenue record retrieved — the Indian entity's RoC filing via Entrackr — FY25 revenue was ₹175.5 crore (≈$20–21M), +24.5%, against ₹279 crore of costs and a ₹90 crore loss that nearly doubled. Three structural facts dominate. (1) Over 99% of revenue is non-Indian even though the company is Indian; the home market pays ₹1.3 crore. (2) The asset marketed is not the asset an independent counter sees: 26 FDA-cleared indications across 9 products (company, 26 Feb 2026) versus 9 authorisations on the FDA AI list, all software-code (canon REG-FDA-018/023). Different units — an indication is a finding inside a cleared device — and conflating them is the basis error canon corrects for GE (130 vs 33) and Aidoc (34 vs 33). (3) Because India-listed Fractal Analytics equity-accounts ~31%, a partial public loss window now exists, and it points to a group loss materially larger than the India filing.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~82¢ is cost of goods and ~0¢ operating expense, leaving ~18¢ of operating profit (~51¢ net).
Revenue trend
Margins
Cloud at ₹18 crore (~10% of revenue) is the visible variable-cost proxy; the real drag is deployment and validation labour inside the ₹133 crore employee line, which does not fall with volume.
Deteriorating — costs +39% against revenue +24.5%. The company is buying US entry and new indications ahead of the revenue they unlock.
Worse than FY24's −34%, and the associate window suggests the group figure is worse still.
Up from ₹109 crore — a people business wearing a software label, which is also what makes 26 indications affordable at this revenue.
Equity-funded; ₹35 crore of cash and bank balances against a ₹90 crore annual loss.
COGS structure
Three pools, one of which behaves like software. INFERENCE: running a CADe model over a chest X-ray or head CT is cheap and getting cheaper — ₹18 crore of cloud spend bounds all cloud cost near a tenth of revenue. REGULATORY AND CLINICAL VALIDATION: the real cost of the product. Each claimed indication rests on a submission and its validation dataset; canon prices the bare US filing at the FY2026 standard 510(k) user fee of $26,067 ($6,517 small business), which is why canon calls clearance stock 'a filing fee, not a rent' — the fee is trivial, the evidence is not, and it lands in the employee line. DEPLOYMENT: PACS integration, national-programme workflow design and field support across 100+ countries is labour, and it is what makes a wide footprint expensive rather than free. That mix is why an apparently software-margin business printed −45.30% EBITDA.
Capex
Immaterial as reported. No owned compute or imaging hardware; GPU capacity is cloud opex. Capital goes to engineering and clinical headcount, regulatory submissions and the US commercial build-out. The Series D named 'complementary med-tech company acquisitions' as a use of proceeds; no completed acquisition was retrieved.
Latest earnings
n/a
None published. The CEO told Reuters on 20 May 2025 that the company looked to 'break even and be profitable next financial year' and that 'maybe in two-and-a-half years or two years is the earliest we can do an IPO'. An ambition, not a filing — no DRHP or S-1 was retrieved, and the FY25 accounts filed five months later showed the loss nearly doubling.
- FDA-cleared indications (company claim)
- 26 across 9 products for X-ray and CT (26 Feb 2026)
- FDA authorisations (canon basis)
- 9 on the FDA AI list, all software-code — canon REG-FDA-018/023, vintage 16 Jun 2026, decisions through 30 Mar 2026
- CE-certified indications (company claim)
- more than 65
- Last priced round
- $65M Series D announced 25 Sep 2024, led by Lightspeed and 360 ONE Asset — post-money NOT disclosed
- Round participants
- Merck Global Health Innovation Fund, Kae Capital, Novo Holdings, HealthQuad and existing investors; Fractal Analytics and Peak XV are longer-standing holders
- Third-party valuation estimate
- just under $264M, Nov 2024 (Tracxn; carried by Reuters 20 May 2025) — a relay, not a disclosed post-money
- Listed-holder link
- ~31% held by Fractal Analytics (NSE/BSE, listed 16 Feb 2026) as an equity-accounted associate, per IPO analyses of its RHP
- Capital raised
- ~$121M / ~$123M–$141.3M / $125M across Entrackr, Tracxn and Reuters — no primary total retrieved
- FY25 revenue and loss
- ₹175.5 crore ≈ $20–21M, +24.5%; loss ₹90 crore ≈ $10.6M, +87.5%
- Geography
- >99% of FY25 revenue earned outside India; CEO: US ~25%, India <5% (May 2025)
- Footprint (company claim)
- 90+ countries and 3,000+ sites at the Series D → 105+ countries and 45M+ lives (9 Jul 2026)
Growth drivers
- Clearance breadth as the US sales weapon — six qXR-Detect indications cleared 26 Feb 2026 (510(k) Class II, CADe) across six regions of interest — lung, pleura, mediastinum/hila and heart, bone, hardware, other — framed by the company as completing the chest X-ray suite and taking claimed FDA-cleared indications to 26 across 9 products
- The screening buyer — TB and lung-cancer case-finding budgets at ministries, the Global Fund, Unitaid and NGOs — pays for population throughput, needs no CPT code, no hospital IT committee in the path
- Pharma-funded demand — AstraZeneca's Lung Ambition Alliance (since 2022), AI chest X-rays at multi-million scale across 20+ countries, extended in 2026 into Indian public facilities including a 20-site Telangana programme
- Head CT and stroke (qER) as a second franchise reaching hospital urgency rather than screening yield — the line that puts Qure.ai against RapidAI and Viz.ai directly
- US build-out — an EVP of Sales for North America (16 years across Microsoft and Nuance) and a first CMO (ex-Google) named 9 Jul 2026, on claims of 105+ countries and 45M+ lives
- Falling ultraportable X-ray cost, which creates screening sites — and CAD demand — where no radiologist exists
Bull & bear
Qure.ai owns the deepest chest X-ray clearance portfolio in radiology AI and sells it to the one buyer in global health that pays for throughput rather than per-radiologist productivity. It has the widest deployed footprint of any independent, a pharma partner funding scans at multi-million scale, a second franchise in head CT, and an India cost base that makes 26 indications affordable at ~$20M of revenue. The US — where the money is — is almost entirely ahead of it: the chest suite completed only in February 2026, the sales leadership only in July.
- Clearance stock is real on canon's own basis: 9 FDA authorisations, 100% software-code — cleaner than any imaging OEM — and part of the 71 held by four private independents that exceeds GE (33) plus Siemens (32) combined
- The Feb 2026 six-indication clearance completes the chest X-ray suite for the US. Chest X-ray is the highest-volume imaging exam in the world, and owning the whole finding set rather than one triage indication is the difference between a point tool and a workflow product a reading group will integrate
- The screening buyer is structurally better for a young vendor than a US hospital: population scale, a WHO-endorsed clinical warrant, and no need for a CPT code to exist — which matters, because canon's 508:1 ratio says one mostly does not
- AstraZeneca's Lung Ambition Alliance work already runs AI chest X-rays across 20+ countries and extended into Indian government facilities in 2026 — a pharma commercial budget buying volume that would otherwise need a health-system sale, and a template repeatable in respiratory and cardiology
- Over 99% of revenue is earned outside India and the CEO puts the US at ~25% and rising: hard-currency revenue from developed and middle-income buyers, with the ex-India line growing 39.6% in FY25 against 24.5% consolidated
- The exit is the cleanest part: no US-listed vehicle for radiology-AI software exists and all four holders are private, while acquirers pay up for distribution (GE paid $2,293M for Intelerad, which holds zero FDA AI entries, against $357M for two AI-clearance deals). A complete cleared chest finding set is a cheap, high-fit bolt-on for an OEM, PACS vendor or imaging roll-up
A ~$20M-revenue business losing ₹90 crore a year at the India entity — and, on its listed shareholder's associate line, materially more at group — with growth decelerating from 83% to 24.5%, selling into the one health-AI market whose donor funding was cut sharply and whose price its own funders are cutting up to 65%. The US market it must win is already held by two better-capitalised private incumbents and pays no code for a chest X-ray read. The headline asset, '26 FDA clearances', is nine device authorisations counted a different way, and the only valuation anyone quotes is a data-provider estimate the company has never confirmed.
- Unit economics do not work and are deteriorating: ₹279 crore of cost on ₹175.5 crore of revenue, −45.30% EBITDA, loss +87.5% while revenue grew 24.5%, ₹35 crore of cash and bank balances at year-end — and growth slowed by two-thirds in the year the company was supposedly scaling
- The listed-holder window is worse than the filing: Fractal's ~31% stake absorbed ₹45 crore of Qure.ai losses in H1 FY26, which grossed up implies roughly ₹145 crore of six-month group loss. Directional — the equity line can carry fair-value amortisation — but it points the wrong way
- The flagship market's budget is collapsing: ~50% of international donor TB funding 2015–2024 was American, the FY2026 request cut TB by $228M to $178M, and the Global Fund's 2024–26 cycle lost $1.4B (11%) by July 2025. A vendor whose demand comes from case-finding campaigns is levered to an appropriation, not a purchase order
- Donors are also setting the price — Unitaid/CHAI's 17 Jul 2026 agreement puts TB-CAD at $2,500–2,850 per X-ray unit per year across 138 countries. Whether or not Qure.ai participates, that is the number a ministry now quotes at it, in the market where it has the most units
- No code, no annuity in the US: 508:1 authorisations to Category I CPT codes; the only AI NTAP expired after two fiscal years at a $1,040 per-case cap; the SaMS reclassification moved the aggregate rate +0.53% and CMS itself calls the protection interim. Canon's own case study is Viz.ai — 'the cleanest case study in why a reimbursement win is not an annuity'
- Disclosure quality is itself a risk: no post-money, no ARR, raise totals differing ~17%, a '60–70% every year' growth claim against a filed +24.5%, and a break-even-'next financial year' statement made five months before accounts showing the loss nearly doubling. Against ~$20M of revenue the unconfirmed $264M estimate would be ~13x trailing sales — illustrative only, and above the ~11.4x TRAILING sales canon records for listed HeartFlow (and further above the ~9.4–9.6x forward multiple on HeartFlow's own raised $228–232M FY2026 guide) on nine times the revenue
What it is worth
No market capitalisation exists — private, no listed equity, no disclosed post-money. Approach: state the last priced event and what it does and does not establish; attribute the only third-party valuation datum as a relay; anchor on the two hard-ish records (the Indian RoC filing and Fractal's equity-accounted associate line) and show what multiple an unconfirmed mark would imply, labelled illustrative; triangulate against the canon-verified position of the four private radiology-AI independents and what acquirers actually pay.
A materially lower mark or a distressed sale. TB donor funding keeps falling while Unitaid/CHAI-style agreements reset the reference licence to ~$2,500–2,850 per unit per year, compressing revenue per site across the densest part of the footprint; the pharma channel is cut in a portfolio review; US traction is slow because Aidoc, RapidAI and Viz.ai are already inside the accounts and no code pays for a chest X-ray read; growth decelerates further from +24.5% while the group loss runs ahead of the ₹90 crore India figure against ₹35 crore of cash; and the company raises flat-to-down or sells to an OEM that, on canon's evidence, pays for distribution rather than clearance stock. In that path the 26-indication portfolio becomes a cheap bolt-on inside somebody else's reading pipeline, and Fractal's carrying value is where the markdown first becomes visible.
Roughly flat to modestly higher. Revenue compounds in the 20–35% range off ~$20M as US clearances convert slowly and ex-India programme revenue grows; losses persist because the US build-out is front-loaded and the associate line says group burn is heavier than the India filing shows; no priced round tests the mark; and the company remains what canon says it is — evidence that the radiology-AI clearance leaders are all private, rather than an investable expression.
Materially above the last unconfirmed mark. The completed chest suite converts into US enterprise contracts through the new North America organisation; qER takes stroke-network share from RapidAI and Viz.ai; pharma-funded and government screening revenue more than offsets donor cuts; revenue re-accelerates off a ~$20M base while the India cost structure keeps the loss flat rather than compounding; and a strategic acquirer bidding for a complete, cleared chest finding set prices scarcity rather than revenue. A liquidity event at Aidoc, RapidAI or Viz.ai would set the comparable that re-rates the whole private radiology-AI set.
Last priced event: the $65M Series D announced 25 Sep 2024, led by Lightspeed and 360 ONE Asset with Merck Global Health Innovation Fund, Kae Capital, Novo Holdings and HealthQuad participating. NO POST-MONEY WAS DISCLOSED — not by the company, not by the leads, not in the coverage retrieved — and a valuation is never inferred from a raise size. The only valuation figure retrieved anywhere is a third-party estimate of just under $264M dated Nov 2024 (Tracxn, carried by Reuters 20 May 2025): a RELAY, carried attributed and unverified, a data-provider mark rather than a clearing price. Illustratively it is ~13x FY25 revenue of ₹175.5 crore (≈$20–21M) — arithmetic only, pairing an unconfirmed mark with a single-entity RoC filing of unconfirmed consolidation basis; for scale canon records listed HeartFlow at ~11.4x trailing sales on $191.42M of TTM revenue and ~9.4–9.6x forward on its own raised $228–232M FY2026 guide, nine times Qure.ai's revenue at a lower multiple. The nearest independent anchor is Fractal Analytics' balance sheet: per IPO analyses of its RHP it holds ~31% as an equity-accounted associate, recognised a gain on the deemed stake sale created by the Series D dilution, and absorbed a ₹45 crore share of Qure.ai losses in H1 FY26. That is a carrying value and a loss share, not a market price, and we did not open the RHP — but it is the only recurring, audited-adjacent window, and a Fractal impairment, revaluation or sale would function as a mark. Qure.ai is NOT added to canon's six-mark private aggregate (CM-COH-04, ~$33.05B) because it has no disclosed mark to add. What canon supports: Qure.ai is the smallest of the four private holders of radiology-AI software authorisations (9 against Aidoc 33, RapidAI 17, Viz.ai 12); those four hold 71 between them, more than GE (33) and Siemens (32) combined; and every one is private, which is why no US-listed vehicle for radiology-AI software exists (PB-062). Liquidity paths: another private round, acquisition by an imaging OEM, PACS vendor or imaging-services roll-up, or the India IPO the CEO frames as conditional on profitability. On acquisition value canon cuts against a clearance-led price — GE paid $2,293M for Intelerad, which holds zero FDA AI authorisations, against $357M for two AI-clearance acquisitions, and the board's words are that 'the clearance stock is the cheaper half of any such deal'. Not investment advice, not a price target, and no recommendation on any security — including the Indian and Korean listings named here, which sit outside a US-first mandate.
SWOT
Strengths
- The broadest chest X-ray clearance portfolio in radiology AI by the company's count — 26 FDA-cleared indications across 9 products, 65+ CE indications — and on canon's independent basis 9 FDA authorisations every one of which is software-code, against imaging OEMs whose lead is ~74% scanner registrations (canon REG-FDA-021)
- A go-to-market that does not queue behind a US hospital IT budget: WHO's 2021 recommendation of CAD for TB screening and triage (ages ≥15) made the category procurable for ministries and donors, and qXR was one of three products in the evidence base WHO analysed
- Distribution no US-only rival can replicate cheaply — 90+ countries and 3,000+ sites at the Series D, 105+ claimed by Jul 2026 — plus a pharma-funded programme running AI chest X-rays across 20+ countries
- Two clinically distinct franchises — chest X-ray screening (qXR/qLC) and head-CT emergency triage (qER), with different buyers, urgency and price points
- An India cost base — ₹133 crore (≈$16M) of FY25 employee cost sustains a regulatory programme that would cost multiples of that run from the US or EU
Weaknesses
- Small revenue, decelerating growth, widening loss — ₹175.5 crore (+24.5%) against a ₹90 crore loss (+87.5%) and −45.30% EBITDA, with the associate window pointing to a bigger group loss still
- Home-market failure on the record — India revenue fell ~80% to ₹1.3 crore in FY25, so the company best known for Indian public-health deployments earns almost nothing there
- Thin disclosure even for a private company — no post-money, no ARR, no consolidated audited accounts, aggregator raise totals differing ~17%, and a CEO growth claim the filed accounts do not support
- Weakest US clearance position of the four private independents (9 software-code authorisations vs Aidoc 33, RapidAI 17, Viz.ai 12) and the last of them to build a US sales organisation
- The headline metric it markets — '26 FDA clearances' — is not the unit any independent counter uses, inviting the basis error canon corrects elsewhere
Opportunities
- The structural gap canon identifies — Aidoc (33), RapidAI (17), Viz.ai (12) and Qure.ai (9) hold 71 software-code authorisations, more than GE (33) and Siemens (32) combined, and all four are private — there is no US-listed vehicle for radiology AI software (PB-062), so any liquidity event among the four is a category-defining print
- US revenue is nearly all ahead — the chest suite completed only in Feb 2026, North America sales leadership only in Jul 2026, on a base the CEO put at ~25% of revenue a year earlier
- Ultraportable X-ray plus CAD is becoming the standard screening unit in low-resource settings — a hardware cost curve doing the demand work in the exact workflow this product was designed for
- Acquisition optionality — canon's read is that for an OEM, PACS vendor or imaging roll-up 'the clearance stock is the cheaper half of any such deal', and a complete cleared chest finding set is the cheapest way to buy what the buyer lacks
- Adjacent modality expansion — a Jan 2026 grant round reported by an aggregator is directed at AI point-of-care ultrasound for low- and middle-income countries (relay, unconfirmed), which would extend the franchise beyond X-ray and CT
Threats
- The donor budget funding the flagship market is being cut — the US provided roughly 50% of international donor TB funding 2015–2024, the FY2026 US request cut TB by $228M to $178M, and by July 2025 the Global Fund's 2024–26 cycle had been cut US$1.4B, 11% of allocation (KFF; WHO Global TB Report 2025), with the 30 highest-burden countries losing health workers and seeing lab and supply disruption
- Price is being administered down in the same market — Unitaid and the Clinton Health Access Initiative announced on 17 Jul 2026 that TB-CAD annual licences fall from US$7,000–12,000 per X-ray unit per year to $2,500–2,850 for one year or $7,500 for three — up to 65% off — across 138 low-, lower-middle- and upper-middle-income countries for public and not-for-profit buyers. The named vendors are DeepTek and Delft Imaging; Qure.ai is not named, but the agreement sets the category's reference price
- No US reimbursement path retrieved for any Qure.ai product — canon's 508:1 ratio (1,524 authorisations against 3 Category I clinical-AI CPT codes, denominator secondary), the only AI NTAP precedent expired after two fiscal years at a $1,040 per-case cap, and CMS's proposed SaMS class is set at rate neutrality
- The US stroke and head-CT segment is held by better-capitalised private incumbents — Aidoc (33 software-code authorisations, $150M Series E announced 29 Apr 2026, >$500M raised) and Viz.ai (12, plus the NTAP precedent and a care-coordination network), with RapidAI (17) between them
- Squeezed from both ends — imaging OEMs bundle AI into the scanner sale while cheap capable vision models collapse the barrier to a competent chest X-ray classifier, leaving regulatory evidence and distribution as the only durable assets
Moats, dependencies & bottlenecks
Moats
Strong within its niche clearances do not expire, but they are cheap to add and every rival is adding them 9 FDA authorisations, all software-code, plus a claimed 26 indications and 65+ CE indications. Canon's constraint: at the FY2026 510(k) user fee of $26,067 the stock is 'a filing fee, not a rent', and the list added 333 authorisations in 2025 and 92 in Q1 2026. The barrier is the validation evidence, not the certificate.
the relationships are durable, the budgets are not 90+ countries and 3,000+ sites at the Series D, 105+ claimed by Jul 2026. Being written into a national screening protocol and an ultraportable X-ray workflow is far harder to copy than the model — but it makes revenue a function of donor appropriations, cut sharply in 2025–26.
The 2021 WHO recommendation created the procurable category and qXR was one of three products in the evidence base analysed, with Delft's CAD4TB and Lunit INSIGHT CXR. Shared legitimacy, not exclusivity; later WHO-aligned reviews name additional qualifying products.
a marketing budget, cancellable annually The AstraZeneca partnership funds AI chest X-rays across 20+ countries and extended into Indian public facilities in 2026. Real volume and revenue, but a therapy-franchise commercial budget rather than a clinical purchase, and no disclosure quantifies its revenue share.
₹133 crore of employee cost carries a 26-indication regulatory programme and deployment in 100+ countries. A US-based competitor cannot run that breadth at that cost — the most under-discussed structural advantage.
Tens of millions of screening scans is a genuine training and validation asset, but chest X-ray is the most public and most benchmarked modality in imaging (ChestX-ray14, CheXpert, MIMIC-CXR), so proprietary volume is worth less here than in a rare modality — and donor-programme data rights are constrained.
Dependencies
US bilateral TB programmes, the Global Fund, Unitaid, Stop TB Partnership appropriations, not purchase orders ~50% of international donor TB funding 2015–2024 was American (KFF); the FY2026 request cut TB by $228M to $178M; the Global Fund's 2024–26 cycle lost $1.4B (11%) by July 2025. Clinical need rises as the payer shrinks — demand and ability to pay move in opposite directions.
Ministries of health and national screening programmes across 100+ countries Customer concentration and procurement cycle Tender-based, price-sensitive, slow. The home market is the cautionary case: India revenue fell ~80% to ₹1.3 crore in FY25 despite Qure.ai's prominence in Indian TB screening.
Channel and revenue concentration The largest named commercial programme — multi-million scans, 20+ countries, a 2026 India extension including 20 Telangana public facilities. No disclosure quantifies the revenue share, and an unquantified concentration is itself the risk.
Every US indication is a 510(k) CLEARANCE — market entry against a predicate — not a PMA APPROVAL. Canon's list is 96.19% 510(k), 2.56% De Novo, 1.25% PMA. Clearance permits marketing; it does not establish efficacy and creates no payment.
A standalone algorithm has no channel: it must be installed into a worklist, validated by a reading group and paid from an operating budget. Every channel partner is also a competitor able to bundle a substitute at zero incremental price.
Fractal Analytics (NSE/BSE) as ~31% associate holder and incubator governance and signalling Fractal incubated Qure.ai through its Fractal Alpha venture studio, equity-accounts it as an associate, and recognised a gain on the deemed stake sale created by the Series D dilution. That gives Qure.ai a quarterly public loss window it does not control, and any Fractal impairment, revaluation or sale of the stake would be read as a mark.
Financing / liquidity ₹35 crore of cash and bank balances at 31 Mar 2025 against a ₹90 crore annual loss, no announced priced round since Sep 2024, and an IPO the CEO frames as conditional on profitability. Canon's read on the private radiology-AI names is that acquisition is the likelier route.
Advantages
- The most complete chest X-ray finding set in radiology AI by the company's count, and 100% software-code FDA authorisations on canon's independent basis
- First-mover legitimacy in the WHO-endorsed TB-CAD category, with qXR named in the evidence base behind the 2021 recommendation
- A footprint (90+ → 105+ countries, 3,000+ sites) no US-only competitor can replicate at comparable cost
- An India cost base that lets a ~$20M-revenue company sustain a 9-authorisation, 26-indication regulatory programme
- Two clinically distinct franchises — chest X-ray screening and head-CT emergency triage — with different buyers and urgency
- Pharma and medtech channels (AstraZeneca — Medtronic and J&J MedTech in India) that pay for scan volume without a health-system software budget
Weaknesses
- Small revenue, decelerating growth, widening loss, and a group loss the listed holder's associate line suggests is larger than the India filing shows
- No disclosed valuation, no ARR, no consolidated audited accounts, aggregator raise totals disagreeing by ~17%
- Home-market revenue near zero (₹1.3 crore, −80% in FY25) despite Indian public-health prominence
- Weakest US clearance position of the four private independents, and the latest to build a US sales organisation
- Revenue tied to donor and government budgets cut sharply in 2025–26, in a category whose price those same funders are deliberately reducing
- A stated break-even target and a '60–70% every year' growth claim the subsequently filed accounts did not support — a disclosure-credibility problem as much as an operating one
Bottlenecks
- No US reimbursement code retrieved for any Qure.ai product — every US sale must clear an operating budget on a productivity argument, against canon's 508:1 authorisation-to-Category-I-code ratio
- Donor price administration — the Unitaid/CHAI reference of $2,500–2,850 per X-ray unit per year caps revenue per deployed unit across 138 countries whether or not Qure.ai signs up
- The markets with the most clinical need have the least ability to pay; the markets that can pay already have a reading workflow the tool must displace
- PACS and worklist integration is bespoke per site — the friction that makes a 3,000-site footprint expensive to grow and to keep
- Evidence burden per indication — 26 claimed indications is 26 validation exercises, defended to regulators, payers and reading groups who each ask a different question
- Cash: ₹35 crore of cash and bank balances at FY25 year-end against a ₹90 crore loss and no announced equity round in the following eleven months
Top signals & trends
Top signals
Completes the chest X-ray suite for the US and grounds a complete-finding-set pitch. Watch the unit: indications inside cleared devices, not 26 device authorisations — canon counts 9.
A company builds a US enterprise sales organisation when it intends to convert clearances into contracts — and it is also an admission of how early the US base is.
Growth decelerated from +83% while spend accelerated 39%. The gap between the marketing footprint (105+ countries) and the P&L (~$20M) is the central tension in the name.
Two effects: the first recurring public window on Qure.ai's economics, and what it shows is a group loss run-rate well above the India entity's. The 31% and the ₹45 crore come from IPO analyses of Fractal's RHP; we did not open the RHP.
The most-cited deployments — Indian TB screening — are not the revenue. The clearest evidence that public-health impact and monetisation are separable in this model.
It sets a public reference price in Qure.ai's densest market and signals that donors intend to shape pricing directly. Qure.ai is not named — an absence, not a confirmed exclusion.
The largest exogenous risk to demand. Screening campaigns are the first line cut when a TB programme loses staff and drug supply.
Pharma money substituting for shrinking donor budgets in the same clinical workflow — the most plausible offset to the aid-cut risk, and evidence the channel renews.
Twenty-three months without a priced round against a widening loss. The CEO's sequencing — break even first, IPO 'in two-and-a-half years or two years' at the earliest — reads as a preference for not testing the private mark.
Canon REG-FDA-023 / PB-062. It is why radiology AI has no US-listed expression, and why any of the four listing or selling is a category-level event. A diagnostic, not a call — and Qure.ai is the smallest of the four.
Trends
Qure.ai's six-in-one clearance is the clearest instance. A reading group integrates one comprehensive tool sooner than six point tools, so breadth becomes the buying criterion — and canon's list has 528 filers holding exactly one device.
Unitaid/CHAI, 17 Jul 2026: from $7,000–12,000 to $2,500–2,850 per X-ray unit per year across 138 countries, up to 65% off. Access improves, revenue per unit falls, and the buyer's price expectation resets permanently.
~50% of international donor TB funding 2015–2024 came from the US; the FY2026 request cut TB by $228M; the Global Fund lost $1.4B (11%) of its 2024–26 cycle by July 2025.
Canon: 1,524 authorisations, 333 in 2025 and 92 in Q1 2026, 76.38% radiology, against 3 Category I clinical-AI CPT codes — 508:1. Clearance is table stakes; payment is not following.
Canon REG-CMS-004/005/006 and REG-CPT-007: 36 HCPCS codes designated SaMS in Table 61; separately, 50 codes carry proposed SI = O1 in Addendum B, and the +0.53% aggregate rate effect is measured across the 30 of those 50 that already had a CY2026 rate ($15,184.04 to $15,265.00) — two different sets, never joined. Rate neutrality is stated as the goal and the protection described as interim. Recognition without money — but recognition first.
Canon REG-FDA-021/022: ~74% of the eight OEMs' 403 authorisations are scanner registrations, and on software codes they hold 105 of 818 (12.84%) against seven AI pure-plays' 114 (13.94%). GE paid $2,293M for Intelerad (zero FDA AI entries) against $357M for two AI-clearance acquisitions.
CAD is only useful where images are produced; cheaper portable units create sites that never had a radiologist — precisely this product's market.
Building a competent chest X-ray classifier is no longer the hard part; assembling regulatory evidence, integrations and distribution is. That favours holders of clearance stock, Qure.ai included, but erodes any model-quality premium.
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.
₹18 crore of cloud cost in FY25, ~10% of revenue; no provider named in retrieved disclosure. Low-connectivity deployments also run on-premise, so the dependency is real, bounded and falling in unit cost.
including ultraportable X-ray manufacturers CAD needs an image. Portable and ultraportable units create Qure.ai's screening sites, so that hardware cost curve drives its addressable installed base.
The layer through which an algorithm reaches a radiologist's worklist — simultaneously distribution and gatekeeper, able to promote a substitute.
Clinical validation sites and annotation labour The real input to a clearance: labelled studies, reader studies, site agreements — mostly India-based, which is why ₹133 crore of employee cost supports a 26-indication portfolio.
National TB and lung-health programmes across 100+ countries The archetypal buyer — population screening at throughput pricing, warranted by WHO's 2021 CAD recommendation. Counter-evidence sits in the accounts: India revenue was ₹1.3 crore in FY25, so programme prominence is not programme revenue.
the Global Fund, Unitaid, Stop TB Partnership, the Gates Foundation They finance the campaigns that buy the software and increasingly set its price (Unitaid/CHAI, 17 Jul 2026). A Jan 2026 Gates Foundation grant round directed at AI point-of-care ultrasound for LMICs is aggregator-reported only — relay, unconfirmed.
Pharma partner and de facto customer through the Lung Ambition Alliance since 2022 — AI chest X-rays at multi-million scale across 20+ countries in Asia, the Middle East, Africa and Latin America, extended in 2026 into Indian government facilities including 20 Telangana sites.
stroke networks and imaging chains outside the US qER head-CT deployments in hospital stroke networks (Indian examples include Aster MIMS and other Kerala hospitals), where the pitch is turnaround time to intervention rather than screening yield.
emergency departments and imaging centres The market now being built for: qXR-Detect (Feb 2026) plus existing lung-nodule, pneumothorax, pleural-effusion, intracranial-haemorrhage, cranial-fracture and midline-shift clearances, sold by a North America organisation staffed in 2025–26. No US customer count or revenue figure is disclosed; the CEO put the US at ~25% of revenue in May 2025.
Medtronic and Johnson & Johnson MedTech (India) Named medtech partners/clients per Reuters (20 May 2025) and company releases — device companies using AI triage to grow a treated-patient funnel, the same channel logic as the pharma partnership.
Private. Largest independent holder: 34 authorisations, 33 software-code (canon CF-AIDOC-02); $150M Series E announced 29 Apr 2026, >$500M raised, valuation not disclosed (CF-AIDOC-01). Broadest US health-system footprint of the independents and an explicit platform strategy — the direct rival for the US enterprise contract Qure.ai is now chasing.
Private. 17 authorisations, all software-code (canon REG-FDA-018). Dominant in stroke imaging and CT perfusion, and it publishes a Medicare NTAP position for its products — head-on with qER in the segment with the most hospital urgency and budget.
Private. 12 software-code authorisations and the first AI Medicare NTAP — ContaCT, granted in the FY2021 IPPS final rule at a $1,040 per-case cap, run FY2021–FY2022, expired (canon REG-NTAP-001/002). Canon's watchlist calls it 'the cleanest case study in why a reimbursement win is not an annuity'.
Private, Netherlands. The original TB-CAD product and the third in WHO's 2021 evidence base. Named with DeepTek in the Unitaid/CHAI agreement of 17 Jul 2026 setting TB-CAD licences at $2,500–2,850 per X-ray unit per year across 138 countries — a direct rival in Qure.ai's densest market, now with a donor-blessed price.
KOSDAQ-listed — the reason 'no listed radiology-AI vehicle' is a US statement, not a global one. INSIGHT CXR was in WHO's 2021 CAD evidence base, so it meets qXR in the same screening tenders, and it also plays mammography where Qure.ai does not. No Lunit financial figure is asserted here; none was verified at 7 August 2026, and a Korean listing sits outside a US-first mandate.
Private, India. The other vendor in the Unitaid/CHAI 17 Jul 2026 agreement, identified in secondary reviews among products meeting WHO's later performance standards — an Indian competitor attacking the same public-health buyer from the same cost base at a donor-set price.
Private, Australian. Comprehensive chest X-ray and head CT finding sets — the closest product-shape analogue to qXR-Detect internationally, competing in the UK, EU, Australia and Asia rather than in donor-funded screening.
95 FDA AI authorisations (90 Siemens-named + Varian 5), 32 software-code on canon's basis. AI-Rad Companion bundles chest X-ray and CT AI into the scanner sale — the substitute that arrives with the hardware. Canon pairs SHL.DE with QURE in the exposure list for its clearance-stock shift point.
Largest single holder on the FDA AI list — 130 authorisations, only 33 software-code, ~74% scanner registrations (canon REG-FDA-010/015/021). Simultaneously channel and plausible acquirer; its $2,293M purchase of Intelerad (zero FDA AI entries) against $357M for two AI-clearance acquisitions is canon's evidence that distribution outprices clearance.
RadNet holds 29 family authorisations, 22 acquired (76%) — the listed imaging roll-up that buys clearance portfolios and runs them through its own reading pipeline. Philips holds 45. Below them Nanox AI, Gleamer, Coreline Soft, Infervision and Oxipit compete per finding on price; canon counts 781 distinct as-filed company strings and 528 names with exactly one device, the fragmentation that makes 'complete finding set' a real differentiator.