RapidAI
B2B enterprise software sold to hospitals, not payers. Per-site/per-module annual subscription for the Rapid Enterprise Platform (CTP, LVO, LMVO, ASPECTS, ICH, Aneurysm, MLS, DeltaFuse, Aortic, Lumina 3D), on-prem or via Rapid Edge Cloud into the radiology workflow, PACS and mobile alerting. Land the stroke hub, expand across its transfer network, cross-sell modules. Revenue is hospital opex, not a payer annuity.
Earnings, margins, COGS & capex
Private, capital-efficient and evidence-first, with the deepest clinical moat in its category and the thinnest public financial record on the board's watchlist. No revenue, margin or cash figure is disclosed. What is verifiable is the regulatory stock: 17 FDA AI authorisations, all 17 under software product codes, second among independents only to Aidoc's 33. The question is not whether the technology works — DEFUSE 3 and DAWN settled that using RAPID for selection — but whether guideline-embedded authority becomes pricing power when the buyer is a hospital budget and no Medicare add-on is on record for RapidAI at any point.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~0¢ is cost of goods and ~0¢ operating expense, leaving ~100¢ of operating profit.
Revenue trend
Margins
software-like; per-study inference is the variable component
regulatory + evidence spend is the fixed base, growing with every module and geography
a credit-led 2023 round implies servicing obligations; direction unknown
COGS structure
Per-study inference on CT, CTA, CT-perfusion and MR — perfusion maps, core/penumbra volumetrics, occlusion detection, ASPECTS scoring, Lumina 3D reconstruction — on an on-prem appliance or Rapid Edge Cloud. COGS mixes hosting, edge-hardware amortisation and cloud/GPU consumption with 24/7 clinical support and per-hospital integration. Perfusion is genuinely compute-heavy versus a notification-only classifier, so unit COGS sits above a triage-only rival's.
Capex
Minimal fixed capex. No owned data centers or imaging hardware; edge appliances at customer sites are the only physical footprint, typically bundled into the subscription. Capital goes to headcount and the evidence programme — the true capitalised asset, expensed.
Latest earnings
n/a
No financial guidance of any kind. Directional signals only: 2,500+ hospitals in 100+ countries, 700+ peer-reviewed studies, five FDA clearances announced Nov 2025, an AWS co-development relationship.
- FDA AI authorisations
- 17 — all 17 under software product codes (list vintage 2026-06-16, parsed 2026-08-02)
- Independent clearance pool
- 71 software-code authorisations = Aidoc 33 + RapidAI 17 + Viz.ai 12 + Qure.ai 9 — more than GE (33) and Siemens (32) combined on the same basis; all four private
- Valuation
- none disclosed, ever — last round credit-led with no post-money published
- Disclosed external capital
- ~$100M indexed ($25M Sep 2020; $75M 27 Jul 2023) — SECONDARY, incomplete
- Hospitals / countries
- 2,500+ / 100+ (company, Dec 2025), from 2,200+ (Jul 2023)
- Clinical evidence base
- 700+ peer-reviewed studies; RAPID was the imaging-selection engine in DEFUSE 3 and DAWN
- Regulatory first
- Rapid ASPECTS was the first neuroimaging device cleared in FDA's CADx (computer-aided DIAGNOSIS) category, 07 Jul 2020 — a 510(k) CLEARANCE, not a PMA approval
- Headcount
- ~258 (third-party estimator, 30 Jun 2026) — SECONDARY, not company-confirmed
Growth drivers
- Guideline embedding — RAPID selected patients in DEFUSE 3 (6-16h) and DAWN (up to 24h); AHA/ASA recommends thrombectomy at 6-24h for patients meeting those criteria
- Module cross-sell — five 510(k) clearances announced 25 Nov 2025 widen the per-site cart into medium-vessel occlusion, midline shift and aortic surveillance
- Hub-and-spoke economics — the transfer decision is made on the images, so the expand unit is the network, not the hospital
- Beyond-neuro expansion with AWS (01 Dec 2025) across orthopedic, cardiovascular and oncology
- Scale-by-decision internationally — 100+ countries plus a 20-cluster Saudi Health Holding Company agreement (11 Nov 2025)
- Procurement credibility: Lumina 3D on TIME's Best Inventions of 2025 and 28 ISC 2026 abstracts
Bull & bear
RapidAI owns the one asset in imaging AI that cannot be out-spent: the clinical evidence that changed the standard of care. The AHA/ASA thrombectomy window exists in its current form because DEFUSE 3 and DAWN selected patients with RAPID, making it the reference implementation of a guideline rather than a competitor in a bake-off — built on roughly a fifth of what Aidoc raised for twice the clearances, in a category where all four largest independent clearance-holders are private.
- Guideline embedding is the highest-durability SaMD moat: dislodging it requires new trials, not a better algorithm
- 17 authorisations with no basis gap makes it the most regulatory-efficient vendor per dollar raised
- A CADx clearance is the claim shape a quantification CPT code is built on; the Cleerly 75577 precedent shows the path
- Hub-and-spoke topology is self-reinforcing — the receiving center's choice propagates to its referring spokes
- Real velocity outside the core: five clearances announced 25 Nov 2025, Lumina 3D on TIME's Best Inventions of 2025, AWS into three new domains
- Scarcity with no listed expression: the four largest independents hold 71 software-code authorisations, beating GE and Siemens combined
RapidAI has the best evidence and the worst monetisation position in its category. No Medicare add-on is on record behind its flagship — the first AI NTAP went to Viz.ai's ContaCT and it ended after FY2022, so it sells a compute-heavy perfusion product out of a hospital budget against rivals selling a cheaper notification — while Aidoc takes the platform slot and GE and Siemens bundle the function into scanners and worklists. No equity round has been priced since at least mid-2023 and the last capital came from a credit fund. An unmarked, possibly levered, evidence-rich company with no reimbursement annuity is a strategic asset, not a compounding business.
- The reimbursement story is over, not pending: the ContaCT NTAP was extended one year explicitly for COVID-19 PHE ratesetting distortions, then appears ZERO times in the FY2023 IPPS final rule
- No priced equity round or disclosed valuation since at least Jul 2023, a credit fund as most recent lead, while a direct rival raised $150M in Apr 2026
- Perfusion is expensive to run and cheap to substitute — a lower-cost CADt rival can win the budget line while losing the accuracy argument
- Aidoc's 11-indication cleared CT-triage foundation model is what hospital IT consolidates toward; a specialist is what gets consolidated
- The OEMs can make the function effectively free — GE owns both the scanner and, via Intelerad, the worklist
- Expansion beyond stroke abandons the guideline moat while consuming the R&D that keeps it current
What it is worth
No valuation is published, and that is the finding rather than a gap. RapidAI is private and there is not even a negotiated post-money mark to caveat: the last disclosed capital was a $75M growth investment LED BY A CREDIT FUND (Vista Credit Partners, 27 Jul 2023), no valuation disclosed by any participant, and none is inferred from the raise size. Per the board, 'No valuation exists on this board and none should be added.' The listed comparator bounds the multiple only: HeartFlow closed at $25.23 on 31 Jul 2026 for a $2.18B market cap on $191.42M of TRAILING TTM revenue (to 31 Mar 2026, +40.6%) = 11.39x TRAILING price/sales, FY2026 guidance $228-232M implying ~9.4-9.6x FORWARD sales. That does NOT bound RapidAI's business — HeartFlow is priced by CMS's fee schedule through CPT 75580/75577, and RapidAI has no fee-schedule anchor and none on record at any point; the FY2021-FY2022 NTAP was Viz.ai's ContaCT.
The squeeze closes. Aidoc's platform breadth takes the enterprise radiology-AI slot, the OEMs bundle stroke triage into scanner and worklist at effectively zero incremental price, and cheaper CADt rivals reset what a hospital pays for a notification — with no Medicare add-on to defend the price, while beyond-neuro expansion burns R&D where RapidAI has no evidence advantage. Credit taken in 2023 becomes a refinancing problem, and the outcome is a down-round or a sale at the buyer's price. Not investment advice; no price target is implied.
Continued independence without a public mark. Stroke stays the profitable core, module and international expansion grow revenue at an undisclosed rate, and the company neither prices an equity round nor sells. It stays what the board calls it: a watch-list name that matters for what it demonstrates about market structure rather than for what it is worth.
Outcome, not price: RapidAI converts guideline authority into a durable payment anchor — a quantification CPT code on the Cleerly 75577 precedent applied to Rapid CTP or the CADx-cleared Rapid ASPECTS — while the Nov 2025 module wave and the AWS channel lift revenue per site beyond neuro. That makes it eligible for the 2027 IPO slate or a competitive process among OEMs, imaging roll-ups and PACS vendors, where 17 all-software-code authorisations plus a guideline citation is the category's scarcest asset.
Two disciplines apply. The denominator does not exist: third-party revenue estimates span ~$14.9M (2023) to ~$47.3M (2026), none company-sourced, so any multiple is arithmetic on an unverified base. And RapidAI is likelier priced by an acquirer's process than a round — GE HealthCare paid $2,293M in cash, net of cash acquired, for Intelerad (closed 18 Mar 2026), a workflow platform holding ZERO FDA AI authorisations, against $357M for the two AI-clearance acquisitions priced in its current 10-K and 10-Q (MIM Software $259M, icometrix ~$98M, both net of cash), which carry 12 of the 23 authorisations GE acquired; the other 11 carry nothing in that figure. NO ratio is drawn between those amounts. The board's constraint: 'the clearance stock is the cheaper half of any such deal.' Nothing here is a price target, and no position is expressible — there is no US-listed vehicle for radiology AI software.
SWOT
Strengths
- Deepest clinical-evidence moat in imaging AI — RAPID selected patients in DEFUSE 3 and DAWN, the trials behind the 6-24h AHA/ASA window
- 17 FDA AI authorisations, all 17 software-product-code; unlike Aidoc's 34/33 there is no basis gap
- First CADx clearance in neuroimaging (07 Jul 2020) — a diagnostic-aid indication, stronger than a CADt triage label
- 2,500+ hospitals in 100+ countries, and winning the stroke hub pulls in every spoke that transfers to it
- 17 clearances on ~$100M indexed capital versus Aidoc's >$500M for roughly 2x the clearances
Weaknesses
- No priced equity round since at least Jul 2023 and no valuation ever disclosed — unmarked, possibly carrying credit
- Zero financial transparency; third-party revenue estimates disagree ~3x
- No Medicare add-on is on record for RapidAI at any point — the first AI NTAP went to Viz.ai's ContaCT and ran FY2021-FY2022 before ending — it now sells purely out of hospital opex
- Narrower surface than Aidoc, whose CT-triage foundation model is cleared across 11 indications
- Its best competitive evidence is self-generated — a vendor-announced, single-centre ESOC 2025 abstract
Opportunities
- A quantification CPT code on the Cleerly precedent — Category I 75577 took effect 2026-01-01 with five of seven MACs finalising coverage, the claim shape Rapid CTP and the CADx-cleared Rapid ASPECTS already hold
- Beyond-neuro modules — Rapid Aortic (18 Nov 2025) plus AWS co-development into cardiovascular, orthopedic and oncology
- Lumina 3D as a horizontal wedge — CTA 3D reconstruction is a department-wide tool, not a stroke tool
- Sovereign deals (Saudi HHC, 20 clusters) grow revenue per procurement decision with no US-cycle equivalent
- Cleanest acquisition asset in the category — GE paid $2,293M net of cash for Intelerad (closed 18 Mar 2026), which holds ZERO FDA AI authorisations
Threats
- Aidoc: 34 authorisations of which 33 software-code, ~2,000 hospitals, 60M+ cases/yr, $150M Series E announced 29 Apr 2026, >$500M raised, valuation NOT disclosed
- OEM bundling — GE 130 and Siemens 95 authorisations; GE outguns Aidoc 3.8:1 on the same basis
- Viz.ai owns the reimbursement narrative — the first AI NTAP (ContaCT, FY2021 IPPS final rule, max $1,040/case) and $128.90 for CPT 0764T/0765T on its cardiac ECG product
- LVO/ICH triage commoditising via Brainomix (350+ hospitals — Medtronic Neurovascular tie-up Feb 2025), Avicenna.AI, Nicolab and Circle
- Aortic/cardiovascular/orthopedic/oncology expansion fights Aidoc and the OEMs where RapidAI has NO DEFUSE-3-class evidence
Moats, dependencies & bottlenecks
Moats
the strongest single moat on the board's private watchlist but decays with the guideline cycle RAPID processed the perfusion/mismatch imaging that selected patients in both trials. Honestly stated: co-founder Gregory Albers was DEFUSE principal investigator, so the evidence base is not independent of the company.
High (clearances do not expire) but non-exclusive 17 on the FDA AI-Enabled Medical Device List, all under software product codes. A barrier to a startup, not to Aidoc or an OEM.
genuinely differentiated A 510(k) clearance, NOT a PMA approval. A CADt label authorises worklist prioritisation and does not support a diagnostic claim; CADx does — the shape a quantification code needs.
High once a transfer network standardises Spans PACS, HL7/FHIR, alerting and inter-hospital transfer protocols across a hub and its spokes. Replacing it re-trains a regional stroke pathway.
10M+ scans analysed as of Jul 2023. Not exclusive — the OEMs sit on more imaging, and foundation models compress the value of a proprietary corpus here.
Dependencies
No Medicare add-on is on record for RapidAI at any point; the first AI NTAP went to Viz.ai's ContaCT and ran FY2021-FY2022 before ending. Every dollar now competes with every other software line item.
Clinical authority under the core franchise The moat IS the guideline. A revision de-emphasising perfusion-based selection would erode it without any competitor acting.
Infrastructure + co-development + channel Rapid Edge Cloud runs on AWS; the 01 Dec 2025 relationship covers foundation-model infrastructure. One hyperscaler for both COGS and distribution.
Upstream supply AND direct competitors Every study comes off a scanner made by a company that also sells competing AI — GE 130, Siemens 95, Philips 45, Canon 43.
Integration surface Reaching the radiologist requires the worklist. GE-owned Intelerad, Sectra and Pro Medicus hold ZERO FDA AI authorisations — they meter the exam instead.
The most recent disclosed capital was credit-led with no equity mark set. Refinancing risk is real and unquantifiable from outside.
Advantages
- The only imaging-AI vendor effectively embedded in an AHA/ASA guideline recommendation via the trials that used it for selection
- 17 FDA AI authorisations, all 17 under software product codes — second among independents only to Aidoc
- The first neuroimaging CADx clearance (07 Jul 2020) — diagnostic-aid rather than triage-and-notification
- 2,500+ hospitals in 100+ countries and 700+ peer-reviewed studies, the category's largest published evidence base by its own count
- Achieved on ~$100M indexed capital versus >$500M at Aidoc — the most capital-efficient clearance stock among independents
Weaknesses
- No valuation ever disclosed and no equity round priced since at least Jul 2023; the last capital was credit-led
- Zero financial disclosure — revenue, margin, retention and cash unknown, with estimates disagreeing ~3x
- No Medicare add-on is on record behind its flagship — the first AI NTAP went to Viz.ai's ContaCT and it ended after FY2022; no reimbursement annuity underwrites pricing power
- Out-capitalised and out-breadthed by Aidoc (33 software-code authorisations, >$500M raised, ~2,000 hospitals)
- Its strongest claim against Viz.ai rests on a vendor-announced, single-centre ESOC 2025 abstract
Bottlenecks
- No incremental payment mechanism and none on record at any point — the category's first AI NTAP went to Viz.ai's ContaCT and lapsed after FY2022; price is capped by an opex ROI argument, not a reimbursement pass-through
- Perfusion post-processing is compute-heavy relative to a triage classifier, raising unit COGS against notification-only rivals
- Procurement needs radiology, neurology, neurointervention, IT security and transfer-protocol owners to sign off
- Every new module needs its own 510(k) and evidence base; the regulatory machine is the growth throttle
- No public financials, so no market can price the company without a full sale or IPO process
Top signals & trends
Top signals
Three years without an equity mark while a direct rival raised $150M in Apr 2026. Readable as discipline or as a market that would not clear the prior mark; no disclosure resolves it.
A functioning regulatory machine, and the first material widening of the per-site cart beyond neuro.
Verified on one basis against the machine-readable FDA file, not a trade-press tally, with no authorisation-vs-software-code gap.
ContaCT appears zero times in the FY2023 IPPS final rule. The category's one reimbursement win was temporary and pandemic-extended.
bullish on product, neutral on evidence quality · Real if it replicates, but company-issued, single-centre and abstract-stage, comparing a rival on the company's own terms. Not an independent comparative-effectiveness finding.
Cheapest path to a larger TAM plus national-scale distribution. Contract values undisclosed; neither is a financial datapoint.
Trends
The 6-24h window created by DEFUSE 3 and DAWN makes advanced imaging selection mandatory rather than optional, and RapidAI is the reference implementation.
Hospital IT is buying one vendor, one integration, many indications — the shape Aidoc leads with, and the shape a specialist gets consolidated into.
The category's only NTAP ran FY2021-FY2022 and ended. Cleerly's CPT 75577 shows a quantification code is achievable — but HeartFlow, priced through CPT 75580/75577, is why the board treats a fee-schedule payer as the wrong side of the payer test.
GE 130 and Siemens 95 authorisations, plus GE's Intelerad purchase ($2,293M net of cash, closed 18 Mar 2026, zero AI authorisations), put incumbents on both ends of the workflow.
neutral-to-positive for RapidAI as an asset · Six disclosed private marks dated Jun 2025 to May 2026, summed, total ~$33.05B — last-round marks, not marks-to-market, and not uniformly post-money — against the whole 17-name listed pure-play cohort at ~$33.90B of market cap at the 31 Jul 2026 close; directional only. The moat-owners are the private ones: Aidoc, RapidAI, Viz.ai, Cleerly.
Compresses the value of proprietary corpora and narrow classifiers while lowering the cost of adding indications. Aidoc moved first, cleared across 11 indications.
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.
Hosts Rapid Edge Cloud; foundation-model infrastructure co-development announced 01 Dec 2025. A COGS input and a channel.
GPU compute for perfusion post-processing and inference. A cost input; no disclosed commercial relationship.
Every CT/CTA/CTP/MR study comes off a GE, Siemens, Philips, Canon or United Imaging scanner — upstream supply and most dangerous competitors at once.
Worklist, HL7/FHIR and DICOM routing are the delivery surface. GE-owned Intelerad, Sectra and Pro Medicus hold ZERO FDA AI authorisations.
US comprehensive and primary stroke centers (hub-and-spoke networks) The core buyer: 2,500+ hospitals in 100+ countries, bought from hospital budgets, with no payer add-on on record for RapidAI at any point.
Statewide stroke-network deployment announced 29 Oct 2025 — the reference shape of the land-and-expand motion.
Community-hospital deployment announced 18 Aug 2025; the spoke side of a transfer network.
Announced 11 Nov 2025 covering 20 health clusters — sovereign-scale deployment. Contract value undisclosed.
PRIVATE, largest direct threat: 34 authorisations of which 33 software-code, ~2,000 hospitals, 60M+ cases/yr, $150M Series E announced 29 Apr 2026, >$500M raised, valuation NOT disclosed; foundation model cleared across 11 indications plus a Breakthrough Device DESIGNATION (not a clearance) for report drafting.
PRIVATE, the head-to-head stroke rival. 12 software-code authorisations; won the first AI NTAP with ContaCT and holds $128.90 for CPT 0764T/0765T on its cardiac ECG product.
130 authorisations on a corporate-family basis, and owner of Intelerad, which holds ZERO FDA AI authorisations but owns the reading pipeline. Board caveat: GEHC is an imaging OEM, not an AI-in-healthcare expression.
95 authorisations (90 Siemens-named + Varian 5). Bundles AI into syngo/AI-Rad workflows across a vast scanner base; Frankfurt-listed with a US OTC ADR.
PRIVATE (UK). e-Stroke / Brainomix 360 across 350+ hospitals, with a Medtronic Neurovascular collaboration for Western Europe (Feb 2025) — a device channel that went to a rival.
45 authorisations. Advanced-visualisation incumbent; a bundling risk in Philips-standardised departments rather than a head-to-head stroke rival.
43 authorisations. Scanner-attached AI, same OEM bundling dynamic, weaker US neuro-AI presence.
PRIVATE (France). CINA-HEAD — FDA-cleared and CE-marked triage for intracranial haemorrhage and LVO plus ASPECTS scoring. Competes on price and OEM/PACS embedding, not evidence.
PRIVATE (Amsterdam, founded 2015). StrokeViewer workflow; a Carahsoft partnership announced 04 Aug 2026 reaches US public-sector buyers (VA/DoD).
PRIVATE (India). 9 software-code authorisations, fourth member of the 71-authorisation independent pool. Strongest in chest/TB but holds head-CT products.