Read this first
- The subject: Blackford Analysis runs a marketplace that lets a hospital use imaging-AI tools from dozens of vendors through one contract and one technical connection. It has had three owners since 2023: Bayer, the investment firm Sopris Capital, and now Azra AI.
- The pattern: none of the three priced Blackford as a standalone business. Each priced it by what it could do for revenue earned somewhere else. Neutrality has held only while that outside revenue stayed outside the catalog.
- What changed this week: Azra announced on October 20 that the acquisition had closed. Azra is the first owner whose core business directly overlaps a category in Blackford's catalog.
- My prediction: within 12 months, apps that detect problems in scans stay open, and follow-up care tilts toward Azra. Moderate confidence. The sale is done; the integration isn't visible yet.
- The takeaway: on any AI platform that hosts other companies' products, find where the owner's products overlap the catalog. Then find out who controls placement there.
Why neutrality is the product
Hospitals store and view medical images in a system called a PACS (picture archiving and communication system). Any AI tool a radiology department wants to use has to plug into it. Twenty AI vendors normally means twenty contracts, twenty security reviews and twenty PACS integrations. Blackford collapses that into one of each.
A lot of Blackford's distribution runs under other companies' brands. Imaging-equipment and PACS vendors, which the industry calls OEMs, resell Blackford inside their own products. Named examples include Philips, GE HealthCare, Intelerad and Merative (the former IBM imaging business).
The model depends on two kinds of trust:
- AI vendors have to believe Blackford won't favor a competitor's app.
- OEMs have to believe Blackford won't compete with them for the hospital.
All three buyers paid for that trust, and each put it at risk in its own way.
The Manus and Character.AI teardowns asked what stays coupled when a company changes hands. Blackford adds the ownership cost we first flagged with Clio in legal AI: what an owner gives up to hold a neutral intermediary.
The independent years, 2010 to February 2023
Blackford spun out of University of Edinburgh astronomy research in 2010. Its first product aligned scans taken at different times so radiologists could compare them. That got Blackford inside a lot of hospital systems, and the company used the foothold to host other vendors' AI. It moved to a platform model in 2015 and publicly introduced its multi-partner platform at RSNA, radiology's largest annual conference, in November 2017.
The decision that set the trajectory was selling primarily through PACS vendors rather than hospital by hospital. Three deals locked it in. eRAD announced a rollout of Blackford to 300+ RadNet imaging centers (2017), Intelerad began reselling through it (2018), and IBM signed for its Merge PACS (2019). By late 2020, Edinburgh's impact case study for the UK's national university research assessment (the REF) reported 27 tools from 16 vendors, 750+ sites, and revenue above £1 million for FY2019–20. Those numbers came from the university and partly from the company itself.
Scottish angels put in most of the money: £11.25 million from all shareholders by 2022, per the Archangels annual report. Gap: I couldn't reconstruct each round's size, lead investor or close date.
The second event that shaped everything after was Bayer becoming a partner before it became an owner. Blackford joined Bayer's G4A startup program in 2019 and signed a development and licensing deal in 2020, per Bayer. In June 2022, Bayer launched Calantic, its own branded AI platform, built on Blackford's technology. Once Bayer's flagship radiology AI product ran on Blackford's infrastructure, an acquisition looked like a question of timing.
The Bayer years, February 2023 to October 2026
Operating years, February 2023 to September 2025
Bayer closed the acquisition on February 13, 2023. Its 2023 annual report shows how it valued the purchase:
| Item | Amount | What it means |
|---|---|---|
| Upfront cash | ~€46M | Paid at closing |
| Contingent payments | Up to ~€54M | Paid only if research milestones were hit |
| Goodwill | ~€68M | Value Bayer assigned to "anticipated innovation potential" |
| Patents and technology | ~€10M | Value of what Blackford actually owned |
Read the goodwill-to-technology split before you read the press release. Goodwill is whatever part of a purchase price can't be tied to a specific asset. Bayer put roughly seven times more value on Blackford's expected future than on its technology. What it bought was a position in radiology departments and an option on what that position might turn into. Stefan Oelrich, then head of Bayer's pharma division, called radiology AI "the fastest growing segment" of the imaging industry.
Bayer makes its radiology money mainly from contrast agents (drugs injected so tissue shows up on a scan) and the injectors that deliver them. All of that is earned before a radiologist reads the scan, and no app in Blackford's catalog competed with it. Industry press reported that Calantic and Blackford would run as two separate platforms under common ownership.
Did Bayer damage neutrality? The cleanest public test is whether the catalog grew while Bayer owned it.
| Date | Apps | Vendors | Source |
|---|---|---|---|
| Dec 2020 (pre-Bayer) | 27 | 16 | Edinburgh REF |
| Nov 2023 | 100+ | 35+ | Philips AI Manager docs |
| Oct 2024 | 135+ | 50+ | DOTmed |
| Sept 2025 | 150+ | 60+ | Blackford site, via AuntMinnie |
It grew. GE HealthCare signed on in October 2024, well into Bayer's ownership. New vendors joined, including Agamon Health, Imbio and DeepTek. I found no vendor or OEM publicly citing Bayer as a reason to leave, and nothing suggesting Bayer's own tools got preferred placement. If you expected a pharma owner to compromise the store, the public record doesn't back you up. High confidence that nothing public shows harm. Private commercial dealings I can't check.
The economics are what failed. Blackford's Scottish operating company filed these accounts:
| Year | Revenue | Operating loss |
|---|---|---|
| 2024 | £1.06M | £14.9M |
| 2025 (spans the exit announcement) | £666,137 | £17.5M |
These cover one legal entity and probably understate the total money moving through the platform. The direction is still clear. Bayer's UK holding company wrote off its full £67.2 million investment for 2024, cancelled £18.7 million in contingent payments after a missed research milestone, and later recorded the final June 2025 milestone as missed as well.
Wind-down, September 2025 to October 2026
The write-off was booked against 2024. The public exit came in September 2025, when Bayer told AuntMinnie it would "deprioritize its digital platform business, which includes Blackford, and will discontinue offerings and services," while honoring existing contracts. In the same report, Signify Research analyst Umar Ahmed pointed to slow hospital adoption, thin insurance reimbursement for AI, and how hard hospital IT integration is. A customer letter obtained by The Imaging Wire discussed moving contracts directly to app vendors or to other platforms.
Bayer kept Blackford for another 13 months. Gap: the public record for that stretch is thin. I found no figures on how many customers migrated, which OEMs renewed, or what happened to Calantic's own customers.
Coverage of the October 2026 sale cited 130+ apps from 45+ vendors, down from September 2025. Blackford's website currently shows both 130+/50+ and 150+/60+, so I can't tell you how big any drop was (low confidence on magnitude). If the catalog did shrink, it shrank during the wind-down, after the customer letter went out, and not while Bayer was operating the store.
Finding: Bayer said it wanted its own products in the Calantic ecosystem, but its core radiology revenue (contrast and injectors) sat outside the catalog, and the public record shows no Bayer app getting preferred placement in Blackford. That is the likeliest reason neutrality held (moderate confidence). It is also why the store had no way to pay Bayer back through Bayer's own products. Bayer tied up to ~€54M of the price to R&D milestones, betting the innovation would show up. It didn't, so Bayer was spared those payments but never got the innovation they were meant to reward.
Sopris, completed in early October 2026
Sopris Capital bought Blackford from Bayer in early October. The sources disagree on the date: October 1 in UK company filings, October 2 per Azra, October 9 on Blackford's history page. The price was not disclosed. Sopris is Azra's majority owner, and Blackford's October 13 announcement of the purchase was already talking about combining the two companies.
So what was the store worth to a financial owner? Three facts point the same way. The seller had already written its investment to zero, the price was undisclosed, and the buyer's only stated plan was to fold the asset into a company it already controlled. Sopris priced Blackford as distribution for Azra, not as a business to hold and grow on its own (moderate confidence; no return target or entry multiple is public). It likely got the vendor contracts and OEM channels at a distressed price. With no price disclosed, that part is speculation.
One filing I can't explain. On September 10, a share allotment shows a single £0.01 share issued for £40 million in cash, naming neither payer nor purpose. Bayer recapitalizing the company before the sale would fit. That is also speculation.
Azra, completion announced October 20 with integration not yet visible
Azra then bought Blackford from Sopris, its own parent, which amounts to an investor moving an asset into one of its portfolio companies. Azra's software finds patients with incidental findings: unexpected problems noticed on a scan or in a report, such as a lung nodule seen on a CT ordered for something else. It then routes those patients into follow-up and cancer care. Each of Azra's 2026 moves extends that funnel. It acquired Thynk Health, a lung-screening company, in April, partnered with Blackford in August, and signed an OEM deal with 4DMedical the same month.
The October 20 announcement named John Marshall CEO of the combined companies, promised continued service to OEM and hospital customers, and described selling both products under one contract and one integration. The October 13 announcement had said Blackford would remain a separate legal entity.
The conflict shows up in Blackford's own catalog, which lists Coordinate from Agamon Health. Coordinate finds actionable findings in radiology reports, alerts the referring physician and tracks follow-up. That is squarely what Azra sells. As of today, Coordinate is still listed, and no change to its terms has been announced.
The OEMs are the most exposed. Philips' compliance documentation shows the hospital's platform agreement is with Philips, but its master service agreement and data-protection agreement are with Blackford, and Blackford is the seller of record on each app. Whoever owns Blackford sits contractually inside Philips' customer relationships.
What the three owners have in common
Neutrality has held only while the owner's core revenue sat outside the catalog. Each owner priced Blackford by what it could do for revenue earned elsewhere. Bayer paid for an option on radiology innovation, with part of the price contingent on R&D milestones it never hit. Sopris paid for distribution it could hand to Azra within weeks. Azra is paying for a stream of patients flowing into follow-up care. Bayer earned its money before the scan was read, and Sopris never ran the store as its own, so neither had much financial reason to tilt it. Azra earns from follow-up, which is one of the categories Blackford lists. That puts an owner's core revenue inside a catalog category for the first time, and gives that owner a direct financial reason to tilt the store.
What happens to the catalog next
Within 12 months of the October 20 completion announcement, Blackford will treat two kinds of apps differently.
- Detection apps stay open. Detection tools generate findings, and findings feed Azra's follow-up business. A broad detection catalog helps Azra.
- Follow-up apps close around Azra. Azra becomes the default follow-up product in the one-contract bundle. Outside tools like Coordinate get pushed down, left out of the bundle, or delisted.
I'm wrong if any of these happen:
- Coordinate or another outside follow-up tool is still bundled on equal terms in October 2027. That would mean Azra values OEM trust above winning follow-up.
- The catalog falls below 100 apps, its November 2023 level. That would mean Azra isn't paying to keep detection broad and Blackford is turning into a feeder for Azra's funnel.
- Philips, GE, Intelerad or Merative stop reselling Blackford under their own brands. That would mean the OEMs see Azra as a conflicted owner for detection too.
First checkpoint: RSNA in late November. Blackford has used it as its main launch venue since 2017. Look for whether Coordinate appears in combined-company materials and how follow-up is presented in the bundle.
Who can credibly run radiology's AI app store
Most of the independent multi-vendor marketplaces in this record ended up inside companies with some other business. Arterys went to Tempus in 2022 at an $8.3 million acquisition-date fair value. EnvoyAI, by then part of TeraRecon, went to SymphonyAI in 2020, was renamed Eureka, and moved into ConcertAI in 2021. Osimis went to deepc in February 2024. Blackford was the largest independent of that generation, and its first corporate owner wrote it to zero. Nothing in the public record shows a neutral imaging-AI marketplace paying for itself, so the most credible operator is whichever owner overlaps least with the catalog.
| Rank | Platform | Owner earns from | Overlap with catalog |
|---|---|---|---|
| 1 (conditional) | Microsoft/Nuance Precision Imaging Network | Azure cloud; Nuance's PowerScribe, the reporting software radiologists dictate into | None confirmed. Caveat: Nuance has historically marketed a follow-up tracking add-on for PowerScribe, and I could not confirm whether it is still sold. If it is, Microsoft shares Azra's conflict in follow-up. |
| 2 | Blackford (Azra) | Finding patients and routing their care | Follow-up and care coordination |
| 3 | Aidoc aiOS | Its own detection algorithms; raised a $150M Series E in April 2026 | Detection, directly, despite marketing itself as vendor-neutral |
Microsoft keeps first place only if the PowerScribe caveat resolves in its favor (moderate confidence). Azra is credible for detection and conflicted in follow-up. Aidoc competes with the detection apps it hosts.
Run this test on any AI platform role you're evaluating, in imaging or anywhere else. List the owner's products and compare them to the catalog categories. Wherever they overlap, decisions about default placement, bundling and ranking carry revenue consequences. Find out who makes those decisions. If the commercial team owns them, the design leader has less authority over the platform's most consequential screen than the title suggests.
- The £40 million allotment: Blackford's September 10 SH01 filing records one £0.01 share issued for £40 million in cash without naming the subscriber, so watch later Companies House filings for who paid and why.
- Philips' white-label contract: The Philips AI Manager compliance guide is the clearest public map of what Azra now inherits inside OEM relationships, since it makes Blackford the principal on every app and Philips the agent.
- Coordinate's catalog status: Blackford's care-coordination listing is the simplest place to check whether Agamon Health's follow-up tool keeps equal footing now that Azra owns the store.
- Azra's stated acquisition logic: In The Imaging Wire's interview with John Marshall, Azra's CEO names Blackford's OEM relationships, white-label orchestration business and health-system contracts as what made the deal attractive.
