GE HealthCare buys time

GE HealthCare announced an agreement on October 5, 2026, to acquire SOFIE Biosciences from Trilantic North America for $945 million in cash. Closing is expected in the first half of 2027, subject to conditions including regulatory approvals. The release talks about radiopharmaceuticals, precision diagnostics and growth. The more revealing number is 109.8 minutes. That is the physical half-life of fluorine-18, the isotope used in many of the PET tracers SOFIE manufactures. After roughly 110 minutes, half of the original radioactivity is gone. After 220 minutes, one quarter remains.

That property turns a molecule into an unusual industrial problem. A box of tablets can sit in a warehouse. An F-18 dose loses activity during quality control, dispensing, a car journey and any wait before the scan. In this business, the final mile has a rare characteristic: time physically destroys part of the product.

That is the useful lens for the acquisition. GE HealthCare is buying a U.S. network of 15 contract-manufacturing (CMO) sites operating 21 cyclotrons, a contract-development-and-manufacturing (CDMO) site focused on theranostics, an approach linking diagnosis with targeted treatment, manufacturing relationships with several drug companies and the U.S. rights to FAPI-74, an F-18 tracer currently in Phase III. The network already exists. FAPI-74’s eventual commercial value remains uncertain. ¹

A drug that decays on the road

PET, or positron emission tomography, uses a radioactive tracer to observe biological activity. For F-18 products, calibration time provides a reference for calculating decay that continues throughout the process. The FDA gives fluorine-18 a 109.8-minute half-life. The decline is exponential: about 68% of activity remains after 60 minutes, 50% after 110 minutes and 25% after 220 minutes. ¹⁰

That does not mean every dose suddenly becomes unusable after two hours. Each drug has its own concentration, calibration and expiration rules. It does mean the manufacturer has to design physical decay into the entire operating process. To deliver 100 activity units at injection, roughly 146 units are needed one hour earlier and 213 units two hours earlier, before allowing for synthesis, dispensing and handling losses. ¹⁰

The economic implication is straightforward: production capacity and distance to the patient cannot be separated. Delivery reach depends on the drug, starting activity, stability and authorised conditions. SOFIE says FAPI-74 can support delivery within an eight-hour radius and an expected twelve-hour shelf life after synthesis. These manufacturer statements show why half-life alone does not set a universal distance limit. A local network, by contrast, can turn a nuclear constraint into a logistics advantage. ⁵

A cyclotron is only the first machine

The word cyclotron naturally attracts attention. It is the accelerator used to produce certain radionuclides. Commercial value is built in the steps that follow. The radionuclide must be incorporated into the target molecule, the required controls must support final release for administration, and the dose has to be dispensed, shielded and transported while enough activity remains.

The FDA subjects PET drug production to dedicated current-good-manufacturing-practice requirements. The NRC and Agreement States separately regulate the possession, production and distribution of radioactive materials. The supply chain therefore combines industrial equipment, chemistry, quality assurance, nuclear licensing and medical logistics. Buying a cyclotron is possible. Re-creating an operating, compliant network around it takes longer. ^{11 12}

SOFIE already had meaningful scale before the transaction. When Trilantic invested in 2024, the company said it operated 15 locations and manufactured more than 567,000 doses annually for partners and customers. Those were 2024 operating figures, not 2026 revenue or current volume. They do show that GE is buying a live industrial system rather than a laboratory project. ⁸

FAPI-74 gives the network a second source of value

There is another layer to the deal. In 2023, GE HealthCare and SOFIE signed a licensing agreement around tracers targeting fibroblast activation protein, or FAP. GE took global rights to Ga-68 FAPI-46 and ex-U.S. rights to F-18 FAPI-74. SOFIE retained the U.S. rights to the latter. The acquisition would bring those rights into the same group once the transaction closes. ^{4 1}

FAPI-74 remains a clinical-stage asset. SOFIE launched two Phase III studies, FAPI-GO in gastroesophageal cancers and FAPI-PRO in pancreatic ductal adenocarcinoma. By August 2026, the company reported 15 activated clinical sites and ongoing enrollment. Early in 2026, seven of its radiopharmacies were already able to manufacture FAPI-74 for trials and collaborations. Both ClinicalTrials.gov records, last updated on September 17 and checked on October 7, still show Phase III recruitment and no posted results. ^{17 18 6 7 5}

This is where infrastructure and intellectual property meet. If the product eventually wins the required approvals, manufacturing methods still have to be transferred and qualified across sites, batches released and doses brought close to imaging centers. A network can shorten the path from approval to geographic access. The value remains conditional: Phase III is neither approval nor a guarantee of commercial success.

What $945 million can actually buy

The $945 million purchase price is public. The documents reviewed do not disclose SOFIE’s current financials. GE HealthCare did not disclose current target revenue, EBITDA, net debt or a purchase-price allocation across the manufacturing network, intellectual property, customer contracts and other assets. A defensible acquisition multiple therefore cannot be calculated without inventing a denominator. ¹

That matters because it prevents a common mistake: dividing the deal value by an old revenue figure found online. A public disclosure from a former shareholder reported SOFIE turnover of about $67.7 million in 2019. Seven years, a change in ownership and substantial network investment later, that figure is far too stale to value the 2026 transaction. ¹⁶

The clean approach is to separate observed assets from GE’s investment thesis. The visible assets are the network, cyclotrons, CDMO capabilities, manufacturing relationships and U.S. FAPI-74 rights. GE expects SOFIE to grow in the low double digits and says the acquisition should be accretive to revenue growth, adjusted EBIT margin and adjusted earnings per share in the first full year of ownership. Those are buyer forecasts, not achieved results. ¹

Why GE can plug this into an existing engine

At closing, SOFIE would join Pharmaceutical Diagnostics, GE HealthCare’s segment spanning contrast media and molecular-imaging agents. The segment generated $2.9 billion of revenue in 2025. In the second quarter of 2026, revenue was $843 million, up 15.6% on a reported year-on-year basis, with $250 million of segment EBIT and a 29.6% EBIT margin. ^{3 2}

The $250 million is segment EBIT, an indicator before interest and taxes defined by GE that excludes items including certain acquisition charges and amortisation of acquired intangibles. ¹⁹ Those numbers do not provide a SOFIE multiple. They show the industrial context of the buyer. GE already has commercial teams, F-18 products, imaging hardware, digital tools and hospital relationships. SOFIE adds local manufacturing and distribution. The intended integration is therefore broader than a product portfolio: it links tracer production to the equipment that images it.

There is a commercial tension, however. SOFIE also manufactures products for other radiopharma companies. GE says that after closing SOFIE will continue to act as an independent manufacturing partner to existing customers, including products from other providers, while GE will continue to use other CMOs for its own portfolio. The value of the network will partly depend on that perceived neutrality. A competitor may dislike having a rival own its manufacturer, yet still stay if the network’s quality, capacity and geography are hard to replace. ¹

U.S. reimbursement has changed the equation too

Radiotracer economics do not end with manufacturing cost. A hospital also needs to know how the drug will be paid for. Medicare changed the treatment of high-cost diagnostic radiopharmaceuticals under the hospital outpatient OPPS system beginning in 2025. For 2026, the threshold is $655 per day. CMS compares an estimated per-day cost for each HCPCS product code with this threshold, rather than a dose’s invoice price. Above the threshold, the product is separately paid; at or below it, payment is packaged with the scan. For qualifying products with claims data, separate payment is based on arithmetic mean unit cost (MUC). New products without those data follow a specific method, while transitional pass-through payment is a separate regime. The rule therefore does not automatically apply to every tracer at every provider. ^{13 14}

That policy says nothing about a future FAPI-74 price and certainly does not guarantee reimbursement. It does change the framework hospitals use when evaluating high-cost new diagnostic agents. A clinically useful drug can still diffuse slowly if its cost disappears inside an inadequate bundled payment. Separate payment reduces that problem for qualifying products while leaving pricing, coding, coverage and actual use unresolved.

The economic value of SOFIE’s network therefore runs on three clocks: physical decay, the regulatory path of the product and the reimbursement path. The three eventually have to line up.

The real product is reliability

A radiopharmacy operates less like a warehouse than a tightly scheduled transport network. Demand arrives as patient appointment slots. Production has to be sized before injection, quality controls must finish on time and transport must preserve enough activity for the scan. A lost minute therefore carries a physical cost while idle capacity carries an economic one. Profitability sits between the two.

That tension explains the potential value of redundancy. SOFIE says it intends to equip sites with dual cyclotrons and dual dispensing hot cells. A second machine can look excessive when everything works. It becomes valuable when equipment fails or a batch has to be repeated. In a business where the product cannot simply wait until tomorrow, spare capacity can operate both as operational insurance and as a commercial selling point. The company does not disclose the cost of that redundancy or its effect on margins. ⁹

Geographic density creates another effect. A broader network can move production closer to imaging centers and may allow some development, method-transfer and quality functions to be shared. The economics are not automatic: each additional site needs qualified staff, licensing, procedures and enough volume. The useful metrics would therefore go beyond cyclotron count to utilisation by site, qualified products, delivery reliability and the share of batches lost or remade. The documents reviewed do not provide those data.

For GE HealthCare, the combination may create portfolio effects. A site already producing FDG or other tracers can support additional products after the necessary transfer and validation work. Regulation also limits shortcuts: an approved molecule, a validated process and experienced staff do not instantly turn every cyclotron into a universal factory. That is the distinction between owning a machine and owning a platform.

Time also reaches working capital. Materials, consumables, labour and transport are committed before a dose is administered and billed. A late cancellation can leave a product whose activity keeps falling. A denser network may improve the ability to reallocate doses or smooth production, but the size of that benefit depends on the drug, rules and customer contracts. l0g found no public data that would allow that recovery rate to be measured for SOFIE.

What can break the thesis

The deal is compelling because it joins physics, manufacturing and healthcare. It also carries several material risks.

The first is FAPI-74. The product remains in Phase III. A study can fail, produce a narrower diagnostic benefit than hoped, or lead to limited clinical use. Part of the $945 million therefore buys an option whose future value is uncertain.

The second is network utilisation. Cyclotrons, dispensing cells, qualified staff and compliance infrastructure still cost money when the schedule is not full. Economics depend on the number of batches, doses and contracts actually processed. GE does not disclose the site-level utilisation data needed to measure that today.

The third is reliability. Short half-life rewards proximity but punishes failure. A production incident, delayed batch release, broken-down courier vehicle or cancelled patient appointment can destroy part of the economic value of a batch. SOFIE says it is building redundancy with multiple cyclotrons and dispensing hot cells. The cost and effectiveness of that redundancy will need to be observed. ⁹

The fourth involves third-party customers. GE acquires infrastructure that competitors may use. That can improve utilisation, but it requires credible governance around customer data, capacity and production priorities.

Finally, the deal still requires approvals and is expected to close only in the first half of 2027. Until closing, the industrial combination described here remains a plan. ¹

GE HealthCare is buying a geography of time

From a distance, the SOFIE acquisition looks like a conventional growth deal: $945 million for a radiopharmacy platform and a clinical-stage asset. Fluorine-18 physics changes the reading. When a product loses half its activity in less than two hours, factory location becomes part of the product.

That is the most useful way to understand the transaction. GE HealthCare already owns imaging agents, scanners, hospital relationships and part of the FAPI rights. SOFIE adds a layer that cannot be downloaded or moved overnight: licensed production sites, teams, cyclotrons, quality systems and short routes to patients.

The return on the $945 million cannot be calculated cleanly from the public data available today. The network, however, explains why the price is buying more than a molecule. In radiopharma, a few dozen minutes can be worth more than a warehouse.

For further analysis of industrial constraints, read Drug shortages: the price of reliability and GPU-backed debt: the fourth year. The 10-K reading guide explains how to separate reported figures, forecasts and missing information.

Sources and documents

1. 01 · GE HealthCare · 2026-10-05Acquisition release
2. 02 · GE HealthCare / SEC · 2026-07-29Q2 2026 results
3. 03 · GE HealthCare · 2026-02-04Full-year 2025 financial results
4. 04 · GE HealthCare · 2023-10-11FAPI licensing agreement
5. 05 · SOFIE Biosciences · 2026-01-27FAPI manufacturing expansion
6. 06 · SOFIE Biosciences · 2026-08-21Q3 FAPI update
7. 07 · SOFIE Biosciences · 2026-02-11FAPI-PRO Phase III
8. 08 · SOFIE / Trilantic · 2024-06-04Growth investment
9. 09 · SOFIE Biosciences · 2026Network page
10. 10 · FDA · 2026Fluorine-18 label
11. 11 · FDA · 2009PET drug CGMP guidance
12. 12 · NRC · currentNuclear material licensing
13. 13 · CMS · 2025-11-25CMS-1834-FC · Final OPPS rule · II.A.3.c(3)
14. 14 · CMS · 2026Medicare payment systems
15. 15 · Reuters · 2026-10-05Deal report
16. 16 · SGX/Jubilant disclosure · 2020Historical SOFIE financial disclosure
17. 17 · ClinicalTrials.gov / SOFIE · 2026-09-17NCT07217704 · Phase III · recruiting
18. 18 · ClinicalTrials.gov / SOFIE · 2026-09-17NCT07217717 · Phase III · recruiting
19. 19 · GE HealthCare / SEC · 2026 Q210-Q · Note 3 · Segment EBIT
20. 20 · SOFIE Biosciences · 2025-01-15FAPI-74 · Five manufacturing sites

Method and limitations. Sources checked on October 7, 2026. Decay follows A(t) = A(0) × 2^(−t/109.8), with t in minutes; required starting activity scales by 2^(t/109.8), excluding synthesis and handling losses. Fractions measure radioactivity, not diagnostic efficacy or a universal expiration time. The sensitivity figure divides the announced purchase price by hypothetical annual dollar revenue: these inputs are not SOFIE estimates, and the ratio is not an enterprise-value multiple. Network figures from 2024, FAPI manufacturing sites from January and clinical sites from August 2026 have distinct dates and scopes. GE’s forecasts, expected closing date and SOFIE’s operating statements remain attributed to their authors.