Europe ­rethinks the CRO-CDMO model

A drug application is judged on the medicine and, separately, on the manufacturing site that makes it. Xspray Pharma learned that the hard way when it signed a manufacturing and supply agreement with NerPharMa in December 2017. The Milan contractor would make both the drug substance and finished product for HyNap-Dasa, a reformulated version of the leukemia drug dasatinib, from clinical supply through to worldwide commercial sale. The Swedish company Xspray Pharma aimed to launch in the United States in 2021.

Nine years on, the product is still not on the market. The U.S. Food and Drug Administration (FDA) turned the application down in 2023 over dosing information and again in 2024 over label comprehension and concerns at the contract manufacturer’s site.

In October 2025, the FDA issued another complete response letter, this time citing good manufacturing practice observations at the contractor’s site. Xspray said none of the observations concerned the production line used for its product. “It is unfortunate that manufacturing-related issues beyond our control are delaying our launch,” then chief executive Per Andersson said. On Aug. 19, 2026, Xspray disclosed that the FDA had rejected the application again. The agency pointed to unresolved observations at NerPharMa and requested data from additional consecutive commercial-scale batches. Xspray says responsibility for closing them now rests with the site’s new owner, the Benta Group.

Despite those setbacks, Xspray says the FDA has raised no questions about the drug’s clinical data, bioequivalence or stability. What it does not control is the factory. Few biotechs do. They own a molecule, a data package and a set of contracts, but whether the drug reaches the market can depend on how well those external partners perform.

Clinical trials stop for the same reason. In January 2025, the FDA placed Atara Biotherapeutics’ active investigational new drug applications on clinical hold after a pre-license inspection found GMP compliance failures at one of the company’s third-party manufacturing sites. The impact extended beyond products made there directly. One affected program used drug product manufactured at a separate, compliant facility, but its starting material came from the site cited by the FDA. The holds were lifted in May.

The problem is not outsourcing itself. It is that regulatory responsibility cannot be outsourced with it. Sponsors remain accountable for the quality and reliability of their manufacturing network even when they do not own the facilities, employ the operators or control how quickly deficiencies are corrected. Xspray’s experience shows how that dependence can delay a launch; Atara’s shows that the same risk can reach further upstream and stop clinical development altogether. That dependence has also changed in nature: sponsors are no longer outsourcing capacity alone.

Why outsourcing became strategic

What sponsors buy through those contracts has changed. Outsourcing began as a way to get capacity more cheaply than building it. Today, increasingly, it is a way to access capabilities that a biotech cannot realistically maintain under one roof.

EUCROF, the European CRO Federation, whose member associations span twelve European countries, says CROs “were originally largely a solution to cost; today they are increasingly a solution to complexity and part of Europe’s operational infrastructure for clinical research, not just execution partners.” By its account, trials have gone global, patient populations have narrowed and data and technology requirements have grown, so sponsors need judgment: where to place a trial, where the patients are, which sites can deliver, and which technologies earn their place.

The same logic runs through manufacturing. Few developers can justify owning every capability required across the lifetime of a drug, particularly as modalities become more specialized. Outsourcing therefore gives sponsors access to expertise and infrastructure they may need only at particular stages of development.

Where outsourcing breaks down

The risk sits at the handoffs. Drug development does not divide neatly into independent work packages, and decisions made by one provider can shape what the next one is able to do.

“Cell line development affects upstream productivity; upstream conditions influence downstream purification; process decisions affect analytical strategy, formulation, comparability and ultimately cost of goods,” says Cornelius Klöck, vice president for strategy and technology at Rezon Bio, a Polish CDMO. Divide that chain among vendors and a decision can be right for one work package and wrong for the program. There is an organizational argument too. Each new team reconstructs a process it did not develop, context is lost as data changes hands, and accountability blurs when several companies contribute to one outcome.

Clinical development creates the same tension. “Too many specialist vendors create complexity and interfaces, while too much reliance on one large provider can reduce flexibility and create dependency,” says Martin Krauss, co-founder of FGK Clinical Research and president of Germany’s CRO association BVMA.

Advanced modalities make those trade-offs harder still. An antibody-drug conjugate needs three largely separate industrial disciplines before sterile fill-finish: mammalian cell culture, high-potency payload and linker chemistry and conjugation under containment. Very few sponsors own all three. Similarly, an autologous cell therapy ties one patient’s material to a manufacturing slot and an appointment; a radiopharmaceutical decays while it ships; and for RNA medicines, delivery chemistry is part of the therapeutic itself.

The more tightly those steps depend on one another, the more consequential every handoff becomes. A formulation decision can reshape a clinical protocol, while missing process knowledge or incomplete data transfer may not become visible until a regulatory review years later.

Providers try to close the gaps

Providers are responding to those risky handoffs by bringing more capabilities together – either under one roof or through tighter partnerships.

Samsung Biologics is pursuing the first route. The South Korean CDMO is bidding CHF 1.46 billion (€1.6 billion) in cash for PolyPeptide, which would add peptide APIs to a business built around antibodies and ADCs. The move would give Samsung another manufacturing capability in a field where demand, particularly for GLP-1 drugs, has strained peptide capacity.

Lonza has arrived at greater integration from the opposite direction. In March 2026, it agreed to sell control of its Capsules & Health Ingredients division to Lone Star Funds at a CHF 2.3 billion (€2.5 billion) enterprise value while keeping 40%. The deal completes Lonza’s shift to a pure-play CDMO and releases CHF 1.7 billion (€1.8 billion) for growth, acquisitions and share buyback. Samsung is broadening its manufacturing capabilities; Lonza is narrowing its focus. They are arriving at the same destination – a greater emphasis on contract manufacturing – but from opposite directions.

Alignment can also be contracted. In December 2025, the Czech cell-therapy CDMO SCTbio and the global CRO Fortrea formed an alliance intended to match GMP manufacturing readiness to clinical logistics from the start of development, addressing the kind of manufacturing-to-clinical interface that Atara’s hold exposed.

Specialists choose their limits

Smaller providers are often moving in the opposite direction: narrowing their focus rather than trying to offer every capability themselves. Rentschler Biopharma announced in January 2025 that it would leave cell and gene therapy and close its Stevenage site, saying growth in that market had been slower and demand lower than it expected, less than 18 months after the site received its cGMP certificate. The exit came alongside a renewed concentration on cell-culture biologics at Laupheim and Milford.

Rezon Bio arose from a different narrowing. Polpharma Biologics split in September 2025 into a Swiss biosimilars developer that kept the original name and a standalone Polish CDMO trading as Rezon Bio. Its Warsaw-Duchnice facility won FDA approval for commercial biosimilar manufacturing in February 2026. Rezon says it separated the CDMO from the proprietary pipeline specifically so that it would not develop products in competition with its clients.

Neither company presents that focus as a retreat from integration. “The future is not integration versus specialization,” Klöck says. “It is intelligent integration,” by which he means keeping tightly coupled work together where continuity improves speed, cost of goods and accountability, while using specialist partners where they add capabilities that would be inefficient to build internally. Asked the same question, Krauss answers it almost identically: the key, he says, is to combine specialization with integration.

His company, FGK, keeps the core of clinical development in-house but uses specialist technology providers where they offer stronger capabilities or expertise, allowing it to stay “flexible in selecting the most suitable partner for each project, especially in fast-moving technology areas.”

Both CDMOs make drug substance in-house and reach outside for the GMP drug-product step, Rezon developing formulations itself but coordinating manufacture through qualified partners; Rentschler using Coriolis Pharma for formulation and its Xpert Alliance with Vetter for aseptic filling. Patrick Meyer, Rentschler’s global head of business development, treats that as a feature. Requirements increasingly “demand a higher level of specialization which cannot always be provided by integrated partners,” he says.

What the client still carries

Every configuration returns the risk to the developer. A vendor network produces coordination cost and fragmentation; a single integrated partner replaces that with concentration. Regulators have already settled where the exposure sits. ICH’s good clinical practice guideline lets a sponsor transfer trial activities but not responsibility for participant safety or data reliability, and requires oversight of work vendors subcontract onward. EU GMP rules put the parallel duty on the contract giver: assess the contractor, supply the knowledge it needs, keep access to the records.

Klöck’s answer to the concentration problem is knowledge retention. Integration can reduce complexity, he says, but if designed poorly, it swaps many vendors for dependence on one. “Outsourcing should never result in a sponsor becoming disconnected from its own molecule. The moment outsourcing leads to a loss of knowledge, the model is not working as it should.” In practice, that means holding the development data, analytical methods, batch and validation history, reference standards and the reasoning behind critical decisions, and audit trails held in the contractor’s validated systems.

Krauss puts the same requirement in contract terms. That is where a sponsor can act on it in advance. Sponsors should keep control of their data, know-how and strategic decisions, he says, and make sure systems and contracts allow activities and data to be moved if necessary. “Outsourcing should extend internal capabilities, not create lock-in.”

The same record now has to serve models as well as people. Process data can feed predictive tools and digital twins, but only where provenance is traceable and definitions stay consistent across sites, and if the process moves, the question is whether the structured dataset moves with it.

Klöck treats a move as routine. Transferability is “part of responsible process development and an important way of managing vendor dependence risk,” he says, and the usual trigger is growth, since a process built for early supply may later need more capacity than its site has. Keeping the same CDMO does not eliminate the work. Rentschler began supporting ­VarmX’s VMX-C001 at Laupheim in 2022, and in January 2026 the companies said Phase 3 supply and process validation would move to Milford. A transfer runs through equipment, training, analytical procedures, validation batches and comparability. And if the transfer happens after approval, the sponsor may also need to submit the appropriate post-approval regulatory filing.

Geography stops being procurement

Then there is politics. “Geopolitics has moved from being primarily a procurement consideration to becoming an important part of CMC and manufacturing strategy,” Klöck says.

U.S. trade policy is adding another variable. On April 2, 2026, President Donald Trump imposed new pharmaceutical tariffs under Section 232 of the Trade Expansion Act, after a Commerce Department investigation concluded that dependence on imported drugs and active ingredients posed a national security risk. Covered patented pharmaceuticals and APIs face a headline tariff of 100%, while qualifying EU products are capped at 15%. Certain categories, including cell and gene therapies and antibody-drug conjugates, can qualify for zero tariffs under specified conditions. That can change the economics of a manufacturing route without necessarily changing which route is technically viable. Scientific fit, available capacity and familiarity with the target regulator still determine where a product can realistically be made.

U.S. policy is also starting to determine which suppliers companies can safely rely on. The BIOSECURE Act, enacted in December 2025, will restrict federal agencies from procuring biotechnology equipment or services from companies designated as national security concerns, and will also affect entities that use those suppliers in federally funded work. The Office of Management and Budget has until December 2026 to publish the initial list of covered companies.

WuXi AppTec shows how the commercial impact can begin before those restrictions formally bite. The Pentagon added the Chinese life sciences services group to its separate Section 1260H list of Chinese military companies in June 2026, although a court temporarily blocked the designation in August. Even without an immediate BIOSECURE prohibition, the uncertainty around a supplier’s future eligibility can be enough to make sponsors reconsider where they place work.

The clinical side faces a different set of pressures. “In the CRO business, this is less about manufacturing and more about the resilience of a development program,” Krauss says. Patient access, quality, timelines and cost still determine where a trial goes, but sponsors now weigh political stability, sanctions, data sovereignty and dependence on any single region alongside them. He stops short of advocating regional trials. Geographic concentration has become a risk factor in its own right, and a multi-country strategy gives sponsors room to shift if conditions change. It also supports the case for a CRO structured like his own.

“The disruptions of recent years have demonstrated that a technically strong supply chain is not necessarily a resilient one,” Klöck says. Rezon says some companies manufacturing outside Europe are now looking to qualify a second European manufacturing site while their products are still in development, rather than waiting until a disruption forces them to find an alternative. Building that redundancy is costly. It can require technology and analytical method transfer, qualification of the new site and, depending on the stage of development, comparability studies and regulatory filings. For a small biotech, that work can absorb capital that might otherwise have funded the next clinical or development milestone. “Regionalization should not be confused with complete localization,” Klöck says. “Biopharmaceutical supply chains are global and will remain so.”

Where the seams belong

Europe is trying to make that network easier to operate. The Critical Medicines Act covers shortages rather than every innovative biologic, and the broader European Biotech Act remains a proposal. EUCROF, which works with the European Medicines Agency on the Act, backs its ambition and says faster authorization and greater harmonization would reduce fragmentation. But regulation alone will not be enough, the federation argues: Europe also needs investment in clinical research infrastructure, digital capability, sites and people. “Europe’s diversity is an asset; fragmentation is not.”

For sponsors, however, there is no regulatory fix for the central design question. Neither an integrated partner nor a network of specialists solves the design problem alone, and providers on both sides of the industry say so. Integration creates value where continuity across tightly coupled work reduces delays, and a specialist justifies the added handoff when its expertise outweighs the coordination it requires.

The sponsor ultimately decides where those seams belong. It also has to retain the process knowledge, data, financing runway and transfer options that keep the network workable. Even a very small biotech therefore needs someone internally who owns the integrated schedule, CMC and quality, the link between clinical operations and product release, and access to the underlying data.

Xspray shows what happens when that control reaches its limits. The company says it plans to resubmit in 2026 and is prioritizing the additional commercial batches requested by the FDA. Whether its drug reaches U.S. patients next year now depends in part on how quickly an Italian manufacturing site under new ownership can satisfy an American regulator. That dependence traces back to a manufacturing agreement signed in 2017.