The Radiopharma Rental Economy: Why Big Pharma Stopped Buying

In 2023, big pharma bought radiopharma biotechs like software companies. Bristol Myers Squibb paid $4.1B for RayzeBio. Eli Lilly paid $1.4B for Point Biopharma. Novartis paid $1B for Mariana Oncology. The combined headline value of those three deals exceeded $6.5B in cash, all of it spent inside an eighteen-month window.

In April 2026, Regeneron and Telix announced a different model entirely. $40M upfront for four programmes. 50/50 cost and profit share. Up to $2.1B in back-end milestones. Optionality for four more programmes. No equity changing hands.

That is not an acquisition. That is a long-term lease.

The shift from one deal architecture to the other is the most important structural development in radiopharma since the modality became investable. It tells you what the operating moat actually is, where the margin is going to sit for the rest of this decade, and which leadership profiles are about to become uncopyable.

This article unpacks what changed, why, and what it means for the next 36 months of capacity, capital, and talent.

What big pharma actually bought in 2023

The acquisition wave of late 2023 and early 2024 was framed as biotech entering radiopharma. In hindsight, it was big pharma underestimating an industrial process.

BMS / RayzeBio, $4.1B (December 2023). RayzeBio's lead asset was an Ac-225 conjugate targeting somatostatin receptor-positive neuroendocrine tumours, with a Phase 3 trial called ACTION-1. Within twelve months of the deal closing, BMS quietly halted enrolment on ACTION-1 because global Ac-225 supply could not meet trial demand. The acquired molecule was not the constraint. The radioactive payload was.

Lilly / Point Biopharma, $1.4B (October 2023). Lilly paid for Point's lead Lu-177 PSMA asset, PNT2002, plus a 180,000 square foot manufacturing campus in Indianapolis. The campus was real. The Phase 3 readout in 2024 was not what they paid for. PNT2002 failed to demonstrate a mature overall survival benefit, leaving Lilly with significant captive infrastructure and a programme that needed reframing. A reminder that infrastructure does not save a clinical readout.

Novartis / Mariana, $1B (May 2024). Different shape. Novartis was already the only fully vertically integrated radiopharma operator in big pharma, with Pluvicto and Lutathera as commercial assets. Mariana brought additional Ac-225 capability, and Novartis subsequently committed $23B to captive U.S. manufacturing across multiple sites in Indianapolis, Florida and Texas. Novartis is the exception, and we will return to why that matters.

The combined message of the wave was: write the cheque, take the assets, scale them in-house. Capital was the lever.

By mid-2025, the model had cracked. Capital, it turned out, could not bend physics.

The four constraints capital cannot solve

The reason the rental model emerged in 2026 and not 2023 is that four hard physical constraints stopped being abstract risks on a deal memo and started showing up in P&Ls.

The isotope isn't there

Global supply of Ac-225 is not yet a market. It is a small set of producers running constrained processes. As of early 2026, total available Ac-225 ran at single-digit Curies per year. Real-world clinical consumption is currently sub-Curie because supply has been rationing enrolment, with allocation caps inside trial protocols mirroring what producers can ship rather than what investigators want to dose. Industry projections put unconstrained demand at full Phase 3 and early commercial pace at multiples of current supply. Public expansion announcements from Cardinal Health (Indianapolis CTA, April 2026), TerraPower Isotopes ($450M Philadelphia plant, online 2029), Niowave (long-term Novartis supply agreement, February 2026), NorthStar (FDA DMF accepted Q2 2026), and Eckert & Ziegler (first GMP batches, 2026) collectively promise meaningful capacity growth. None of it solves the next 36 months.

The constraint is sharper than a market shortage. Ac-225 has a 9.9-day half-life. You cannot stockpile it. Capital cannot pause decay. When BMS halted ACTION-1 enrolment, no amount of money could have produced more isotope on the timetable that mattered.

Lu-177 is more abundant but still under structural strain. SHINE Technologies, Eckert & Ziegler, ITM and Curium are the major producers. Demand from approved Pluvicto, expanding Lutathera, and a growing pipeline of Lu-177 conjugates is straining enrichment, irradiation and processing capacity simultaneously.

The steel isn't there

Building a GMP radiopharmaceutical manufacturing site is not a fast process even with unlimited capital. A single shielded hot-cell, the radiation-isolated workspace where conjugation and dispensing happen, requires 12 to 18 months for steel fabrication alone. The cells are typically lead-shielded to 100mm or thicker, with proprietary glove-port and transfer-system geometries that vary by isotope and process.

Validating an ISO-5 sterile environment inside a negative-pressure radiation containment zone is harder. The two requirements pull in opposite directions: sterile environments want positive pressure to keep contamination out, radiation containment wants negative pressure to keep contamination in. Engineering through that paradox plus full FDA qualification routinely runs four years from groundbreak.

You cannot retrofit a monoclonal antibody plant for radiation work. The shielding alone makes the structural retrofit uneconomic against a greenfield build.

The people aren't there

Radiopharma manufacturing is one of very few drug modalities where the rate-limiting input is human radiation exposure. Operator dose limits in most jurisdictions cap a radiochemist at 500 mSv to the extremities per year. A skilled operator dispensing a high-activity Ac-225 batch absorbs measurable dose with each handling event. Once that operator hits their annual limit, they are off the floor regardless of throughput need.

To scale a manufacturing operation toward the 200,000 patient doses per year that commercial Ac-225 and Lu-177 programmes will require by 2030, the staffing model is not 1x or 2x. It is approximately 3x rotation: enough trained operators to keep the cells productive while individuals cycle off-floor on dose limits.

The pipeline does not exist to fill that. There are perhaps a few thousand qualified radiopharmaceutical manufacturing professionals globally. Time-to-competence for a radiochemist or hot-cell QP is 24 to 36 months. Universities are not graduating cohorts at a rate that comes close to demand. This is one of the most acute talent constraints anywhere in pharma manufacturing.

The time isn't there

The fourth constraint is the most underappreciated. Standard parenteral sterility testing takes 14 days. Most therapeutic radioisotopes used in conjugates have half-lives shorter than that. Lu-177 is 6.6 days. Ac-225 is 9.9 days. Pb-212 is 11 hours. At-211 is 7.2 hours.

The drug decays before the test completes. Doses are released and administered under "at-risk" or rapid-release protocols, with full sterility confirmation arriving after the patient has already been dosed. FDA 483 citations against commercial radiopharma scale-ups are now appearing on this exact failure mode: failures in rapid-release sterility procedures, environmental monitoring excursions, and inadequate documentation of at-risk release rationale.

This is not a problem that more capital fixes. It requires manufacturing maturity, regulatory dialogue, and operator-grade quality systems. The operators with that maturity are concentrated in a small number of specialist firms.

The Regeneron / Telix deal as blueprint

Set against those four constraints, the April 2026 deal reads differently than a press release.

Regeneron is one of the most capable biologics engines on the planet. Their VelocImmune platform produces fully human antibodies and bispecifics at industrial scale. They have $14.3B in 2025 revenue, an in-house oncology portfolio, and the financial firepower to acquire any radiopharma specialist on the market.

They chose not to. They chose to pay $40M upfront for access to four programmes, signed a 50/50 global cost and profit share, and structured up to $2.1B in back-end milestones with low double-digit royalties available as an alternative if they opt out of co-funding.

What Regeneron is paying for, on a deal-structure basis, is not molecules. It is execution capacity. They are paying for Telix's radiochemistry platform, manufacturing footprint, isotope supply contracts, RLS radiopharmacy network, and the regulatory and operational maturity to deliver short-half-life products at scale across multiple geographies.

The deal architecture is closer to a build-operate-transfer infrastructure agreement than to a traditional pharma collaboration. The contracted party brings the buildings, the people, the supply contracts, and the operating systems. The sponsoring party brings the molecule and a share of the cost.

This is the rental economy, made explicit.

The grid is the moat

In radiopharma, the molecule is patentable. The isotope is rentable. The third thing, the operating layer that gets a 9.9-day half-life product from a producer's reactor or accelerator into a patient's bloodstream within a defined dose window, is what we call the grid.

The grid is the network of GMP manufacturing sites, the radiopharmacy nodes that prepare doses for individual patients, the temperature-controlled and shielded logistics that move material on commercial flights and dedicated couriers, the customs and import infrastructure that handles radioactive cargo, and the trained operators at every step.

You cannot buy the grid in the sense of acquiring it as a finished asset. You can buy companies that own pieces of the grid, which is what Lilly did with Point and BMS did with RayzeBio. But integrating those pieces into a coherent operating system is a multi-year exercise that runs against the same physical constraints that capital cannot bend.

This is why specialist operators command the margin they do. Telix posted a 64% gross margin in 2025. That is not a software margin. That is a physical-infrastructure margin that exists because the asset is hard to replicate, regulated, geographically distributed, and underwritten by a workforce nobody else can hire from.

PE and infrastructure capital have noticed. Specialist radiopharma CDMOs are now being underwritten the way data centres were underwritten a decade ago: long-duration physical assets with capacity-reservation revenue models, anchored by multi-year offtake commitments from sponsors who cannot reasonably insource.

That margin, in deal architecture terms, is the rent.

The counter-thesis

The rental thesis is not permanent. Anyone arguing that big pharma will surrender 50% of radiopharma profit to specialists forever is not paying attention to what Novartis is doing.

Novartis has committed approximately $23B to captive U.S. radiopharma manufacturing. They are breaking ground on multiple sites, expanding Indianapolis, building in Florida and Texas, and continuing to acquire targeted capability such as Mariana. Their thesis is that the rental economics of 2026 are temporary, and that the apex player in any modality eventually integrates. Once Ac-225 supply scales materially post-2029, once accelerator-based production matures, once the workforce pipeline catches up, the deal architecture will shift again.

That is a credible read. The current configuration is best understood as a 60-month bridge: the period during which specialist operators capture surplus because demand has run ahead of integratable capacity. After that bridge closes, the architecture changes shape.

Two qualifications matter. First, the bridge is real and substantial. Three years at 64% gross margin compounds into significant capital that specialists can use to deepen their moats, vertically integrate into isotope supply, and acquire complementary capabilities of their own. Second, the post-2029 picture is not predetermined. Specialists who use this window to lock in long-duration sponsor commitments, regulatory firsts, and geographic coverage may make themselves harder to displace than the analogy to historical pharma cycles suggests.

In other words, the rental economy is structural for now. What it transitions into depends on what specialists build with the rent.

What this means for hiring

The deal architecture shift has direct implications for talent strategy across the modality, and these are the conversations our team is having most frequently with operators in 2026.

Specialist sponsors and CDMOs hold the leverage. Telix-side, Curium-side, Lantheus-side, Nucleus-side, ITM-side roles in operations, manufacturing, quality and tech transfer are now structurally premium. Compensation for these positions has run ahead of equivalent roles in conventional biologics or small molecule operations. The spread is widening, not narrowing.

Big pharma captive builds need a different profile. Novartis-style internal manufacturing build-outs require leadership that has actually run a radiopharma operation, not generalist pharma operations leaders translating across. The talent pool capable of building a captive radiopharma site from green-field is small enough that we can name it. Searches at this level run long and often require relocation packages, dual-location working, or assembled leadership teams rather than single hires.

The most uncopyable profile is the bridge operator. The rare leader who has scaled a specialist radiopharma manufacturer through commercial inflection, dealt with FDA on rapid-release sterility, and managed dose-limited workforce planning is the person every captive build wants to hire and every specialist wants to retain. This is the constraint funnel at its sharpest. Ten or fifteen people globally, most of them currently committed.

Quality leadership is the single most contested function. Sponsor-side and CDMO-side both need VPs of Quality with radiopharma GMP experience, FDA inspection track records, and clinical-to-commercial transition history. The pool is small, and the candidates who exist are not in motion. Mandates here run six to nine months and require active outreach to passive candidates.

For founders, investors and operating leaders building or scaling a radiopharma platform in 2026 and 2027, the leadership question is no longer "can we afford this hire." It is "can we identify this person before our competitor does, and can we structure terms that move them."

If you are working through any of these problems, we are happy to compare notes. Our radiopharma practice runs retained mandates at VP and C-suite level across manufacturing, quality, MSAT, CMC, and commercial functions, alongside the published ProGen Radar intelligence series and the State of Radiopharmaceuticals 2026 report.

The deal architecture has changed. The talent architecture is catching up. Both will define which radiopharma platforms get to scale, and which run out of operating capacity before they run out of capital.

Byron Fitzgerald is the Founder of ProGen Search, a retained executive search and market intelligence firm serving life sciences, radiopharma, CDMO, ADC, and cell and gene therapy. ProGen publishes intelligence reports including The State of Radiopharmaceuticals 2026, The Isotope Production Map Q2 2026, and Weaponised Capacity: The CDMO Intelligence Report.