Somewhere between a promising result in a university laboratory and a therapy that reliably reaches patients, something often goes wrong. Not in the science. Not in the clinic.
In the space between them, a space that has no owner, no defined process, and no margin for the kind of improvisation that research environments thrive on.
For advanced therapy medicinal products (ATMPs), this gap is not primarily a funding problem. It is a translation problem. And in an era when the entire cell and gene therapy field is under pressure to prove that its science can become medicine at scale, the organizations that know how to navigate that gap are becoming the ones that matter most.
The Phrase That Contains Everything
There is a principle that runs through every serious conversation about ATMP manufacturing: the process is the product. It is repeated often enough to risk becoming background noise.
But for those working at the boundary between science and GMP manufacturing, it is not a slogan. It is a structural fact with consequences that compound across every decision made in development.
Sweden's ATMP Ecosystem: Three Things Worth Knowing
1. A manufacturing history that predates the field. Sweden's GMP infrastructure for cell and gene therapies has been operational since the late 1990s, years before ATMP became a defined regulatory category in Europe. The country was building the manufacturing foundations for living medicines while most of the world was still debating whether they were feasible.
2. Europe's first CAR-T clinical milestone. In 2014, Sweden manufactured the first CAR-T cells to enter a clinical trial anywhere in Europe. CAR-T therapy, a form of cell and gene therapy in which a patient's T cells are genetically reprogrammed to target cancer, is classified as an ATMP under European regulatory frameworks. The milestone reflected not just scientific ambition but the presence of serious, functional GMP capability at an early stage of the field's development.
3. Integration as a design principle. Sweden has been among the early movers in establishing ATMP infrastructure that deliberately integrates research, pre-GMP process development, GMP manufacturing, and clinical delivery within coherent institutional frameworks. Reducing the organizational distance between these stages is not an administrative preference. It is where therapies are most likely to survive the crossing.
In conventional biologics, a molecule retains an identity that is at least partially separable from how it was made. In ATMPs, that separation collapses. The living cells that constitute the therapy are not merely shaped by the manufacturing process.
They are, in a meaningful sense, its product. Alter the expansion parameters, change the media formulation, modify the cryopreservation protocol, and you do not have the same therapy in different packaging.
You may have a different therapy entirely. This is not a manufacturing problem. It is a scientific reality that manufacturing must be built around from the very beginning.
A Valley Misdiagnosed
The "valley of death" in drug development is typically framed as a capital problem. For ATMPs, the valley is real, and the capital consumed crossing it, or lost when the crossing fails, has been considerable.
The 2024-2025 period alone saw a significant number of programs discontinued or companies wound down, and funding pressure was a genuine factor in many of those outcomes. But capital alone does not explain the pattern. The most treacherous terrain in this valley is not financial, it is structural.
Research environments and GMP manufacturing environments operate on fundamentally different logics. In research, variability can be generative: an unexpected result is data.
In GMP, variability is a deviation, and deviation carries regulatory and patient safety consequences that cannot be reasoned away. Nowhere is this tension more visible than in the analytical challenge of release testing for autologous therapies, where every batch originates from a different patient's cells.
Designing a potency assay that is scientifically meaningful and regulatorily defensible when the starting material is inherently variable, and when the process is inseparable from the product, is not a bureaucratic problem. It is one of the field's most consequential unsolved scientific questions.
A therapy developed in the research environment must eventually live in the GMP environment. The distance between them is not bridged by funding alone.
It is bridged by the depth of understanding required to move a process from one logic to the other without losing what made it work. In a field already navigating regulatory complexity, supply chain fragility, and the heightened scrutiny that comes with allogeneic platforms—where a single master cell bank must perform consistently across years of production runs—that translation challenge is not a phase to be managed. It is the central risk of the entire program.
The consequences of getting this wrong are not abstract. A late-stage process failure in an ATMP program does not produce a setback that can be quietly corrected.
It produces a regulatory notification, a manufacturing hold, a clinical pause, and an investor conversation that nobody in the room wants to have. For allogeneic platforms in particular, where the commercial promise rests on the ability to produce identical product at scale across years, a process that was not adequately understood before it was locked is not a minor liability.
It is a structural vulnerability that will find expression at the worst possible moment. The translation gap and the funding gap are not separate phenomena.
They are connected: programs that cannot demonstrate process understanding and analytical rigor cannot attract the capital to continue, and the ones that run out of money often do so precisely at the moment the translation challenge becomes visible.
The valley of death claims most of its victims not through dramatic scientific failure but through the slow accumulation of decisions that seemed reasonable at the time and proved, collectively, to be wrong.
What Translational Intelligence Actually Means
This is where the CDMO's role is undergoing a fundamental shift: from service provider to active development partner. Not as a rhetorical repositioning, but as a practical response to where the real risk in ATMP development now sits.
Translational intelligence is the organizational capacity to read a therapy developed in the logic of research and reconstruct it faithfully in the logic of GMP manufacturing, without losing its identity in the crossing. It does not live primarily in equipment or infrastructure.
It lives in people who have spent careers at precisely that boundary, and in institutions that have accumulated enough history across enough programs to know which variables matter, which risks are predictable, and where the surprises reliably hide.
That kind of institutional depth does not appear quickly. It is built through successive generations of increasingly complex biologics, each one adding a layer of understanding that the next can draw on.
The scientists and process experts navigating ATMP complexity today are doing so with the accumulated pattern recognition of an organization that has been developing and manufacturing advanced biologics under GMP conditions since 1992, through the early years of recombinant protein production, through the scaling of monoclonal antibody platforms, and through each successive wave of complexity the field has generated since. That history is ours.
ATMPs are not the first time this field has confronted a process-product entanglement. They are its most demanding expression yet.
Regulatory fluency is an integral part of this capability, not a separate discipline brought in at submission time. The EMA has developed one of the most rigorous and structured frameworks in the world for ATMP oversight, with defined checkpoints across the development journey that require manufacturing evidence to be built progressively and deliberately.
For therapy developers thinking beyond Europe, that discipline has tangible value: a product developed through the EMA pathway—with its layered documentation requirements and its expectation of process understanding at each stage—arrives at any subsequent regulatory conversation, including with the FDA, with a depth of manufacturing evidence that programs developed less systematically often lack.
Developing in Europe is not a geographical preference. For many programs, it is an investment protection strategy.
The Stopover by Design
When a therapy departs from a research environment heading toward GMP manufacturing, the most consequential decisions about its future are often still unmade. Research-grade workflows are, in the majority of cases, structurally incompatible with the GMP regulatory framework, designed as they are for iteration and flexibility rather than the control and reproducibility that GMP demands.
Quality attributes defined by researchers may not map cleanly onto regulatory expectations. Cryopreservation protocols that performed reliably in a university setting may fail to deliver consistency under controlled-rate industrial conditions.
These are not exotic failure modes. They are the ordinary challenges of translation, and they are expensive precisely because they tend to surface late, after commitments have been made.
Our Innovation Hub was designed as a deliberate answer to this problem. It exists at the inflection point between research origin and GMP destination: a pre-GMP environment, co-located within the same facility as the GMP manufacturing suite, where process development happens in genuine proximity to production reality.
Therapy developers do not hand their work over at this stage. They work alongside the scientific team, co-designing the process before it is locked, making the decisions that will determine what becomes possible at scale while those decisions are still open.
The co-location is not incidental. A development environment that shares walls with a production line is categorically different from one that does not.
Process learnings transfer in real time. The people who develop the process are in direct and continuous contact with the people who will execute it under GMP conditions.
In a field where process and product are inseparable, that continuity is not a convenience. It is a risk mitigation strategy, and in the current climate of regulatory scrutiny and investor caution, it is one of the most defensible arguments for engaging a manufacturing partner earlier than the industry has traditionally expected.
The practical consequences are measurable. In a recent program we supported, developing an allogeneic cancer vaccine classified as an ATMP, the team moved from empty rooms to a first technical batch in six months.
The first GMP batch was released within eighteen months of program initiation. These timelines were not achieved by moving fast and hoping. They were the result of making the right process decisions early, in an environment specifically designed to support that kind of front-loaded rigor, with a therapy developer present and engaged throughout.
The Infrastructure Beneath the Breakthrough
Sweden's position in the ATMP landscape is not accidental, and it is not recent. It reflects decades of sustained investment in exactly the kind of connective tissue between research, manufacturing, clinical delivery, and regulatory expertise that most ecosystems treat as secondary until they discover they need it.
The country's GMP infrastructure for cell and gene therapies has been operational since the late 1990s. By 2014, Sweden had already manufactured the first CAR-T cells to enter a clinical trial anywhere in Europe.
This milestone reflected not just scientific capability, but the presence of serious, functional GMP infrastructure at a time when most of the world was still treating cell therapy manufacturing as a future problem. That early institutional investment is a significant part of why the ecosystem is credible now.
Within that context, the work being done to redefine what a CDMO can be is a continuation of a longer tradition: the conviction that manufacturing is not downstream of innovation but is, in the most literal sense, part of it. The organizations that will matter most in the next decade of ATMP development are not necessarily those with the largest production footprint.
They are those with the deepest translational capability, the ones that can meet a therapy developer at the earliest stage of process thinking and remain a scientifically engaged, operationally accountable partner through to commercial supply.
Final Reflections
The therapies that will define the next generation of medicine are already in development.
Some will reach patients, while others will not, and when they fail, the failure will rarely be traced to insufficient science. It will be traced to translation: to decisions made too late, in environments not designed to make them well, by partners who had not seen the problem before.
The gap between laboratory promise and clinical reality has always existed. What changes is whether the infrastructure to cross it exists alongside the science, and whether therapy developers engage with that infrastructure early enough to let it do its work.
In the geography of ATMP development, knowing where that infrastructure lives, and what it is built from, is increasingly part of the science itself.
About the Author
Luisa Lundin, PhD, Alliance Manager at NorthX Biologics.
