CGT Industrialization Starts With CDMO Alignment

By Arnaud Deladeriere, Ph.D., Cell&Gene Consulting Inc.

A foundational source of dysfunction in the CDMO landscape stems from the nature of the processes entering the system. Many early-stage companies arrive with academic protocols that were fit for purpose in a Phase 1 proof-of-concept study, but bear no resemblance to processes that can be industrialized.

These processes often depend on research-grade reagents, fetal calf serum, fixed feeding schedules incompatible with biological variability, or legacy steps that reflect scientific habit rather than manufacturing logic. They arrive with informal tacit knowledge that assumes the operator knows what “good cells look like” but offers no written equivalent. They are rarely assessed for scalability, operator burden, or automation compatibility. And yet, companies frequently insist these processes must be executed without modification, often on the assumption that adjustments can be “made later,” usually in Phase 2, something that almost never occurs.

CDMOs, for their part, struggle to balance respect for client ownership with the reality that commercialization requires standardization, validated assays, reliable supply chains, and systems that break down if every client demands a bespoke approach.

A Low-Trust Relationship

This misalignment produces a consistently low-trust dynamic. A CDMO may execute dozens of flawless batches, only to have one deviation erase years of confidence. Early-stage biotechs, often inexperienced in manufacturing, may not fully understand the content or implications of quality agreements. When deviations occur, CDMOs follow the communication pathways contractually defined, only to be met with confusion or frustration because the biotech did not fully grasp what it had previously agreed to.

In this environment, every failure becomes existential. Rather than collaborative problem-solving, the relationship often defaults to blame and defensiveness. Small companies look to the CDMO as both scapegoat and savior; CDMOs attempt to protect themselves from reputational fallout.

Two Worlds Speaking Different Languages

Part of this tension stems from the cultural divide between academia and industry. Academic centers (even those capable of producing GMP-compliant material) are not built to think about commercial scalability, industrial robustness, or life cycle management. Their mission is patient access and scientific discovery, not process optimization or the ability to deploy globally. Early stage biotechs are also hyperfocused on surviving past Phase 1 rather than the long game, drifting away from the product development questions they should answer before moving toward manufacturing.

CDMOs, by contrast, must build processes with the inherent discipline of a commercial manufa.cturer: validated analytical methods, rigorous change control, supply chain redundancy, equipment qualification, and stability programs. These expectations often clash disastrously with the academic origins of many CGT programs.

Specialization As A Path Forward

The Foundry discussion repeatedly circled back to the same conclusion: No CDMO can be everything to everyone.

Early-stage, highly variable, academically derived programs require a very different operating model than a late-stage commercialization-ready product. CDMOs attempting to straddle both worlds risk failing at both.

Specialization, in this case early-stage technical translation vs. late-stage industrial execution, is one of the most immediately actionable improvements available to the field.

Data Sharing: The Industry’s Biggest Bottleneck

Across the discussion, one frustration surfaced repeatedly: the industry’s collective refusal (or inability) to share process knowledge.

Every company, every CDMO, every academic group repeats the same experiments, steps on the same rakes, and rediscovers the same pitfalls — all at enormous cost. Tens of millions of dollars are wasted each year solving problems that others solved years prior. Yet these insights are treated as “trade secrets,” even when they pertain to basic, non-proprietary practices such as mycoplasma testing approaches.

This hoarding of trivial knowledge has stunted the maturation of the field.

The Black Box Of Redacted BLAs

A striking example of this problem can be found in publicly available BLA filings. The CQA tables for commercial CAR-T products are so heavily redacted that even basic information (such as the testing method and acceptance criterion for mycoplasma) is blacked out.

What should be shared – foundational industry standards – is instead treated as proprietary. Regulators accept these redactions, companies protect them, and the field is left without a reliable benchmark for what commercial CMC should look like.

The absence of transparent success models leaves early-stage companies guessing — and CDMOs forced to correct those guesses under time pressure.

Proposed Pathways To Transparency

The participants surfaced several ideas that could materially shift this dynamic:

• Crowdsourced CQA/COA Tables
  Publishing anonymized, consensus-based CQA tables for CAR-T, NK, CD34+, and other cellular products in venues like Cytotherapy could give the field baseline expectations without compromising company-specific IP.
• Greater Industry Participation in Standards-Setting Forums
  Academic and society-driven groups often develop guidelines without the insight or rigor of commercial manufacturing experts. Bringing experienced CDMO leaders into these conversations can restore realism and ensure that emerging standards are commercially executable.
• FDA/Regulatory Agencies as a Data Repository
  The most radical idea proposed: leveraging the FDA (or other regulatory agencies), which already receive full data sets, as an anonymized, precompetitive data hub. While politically complex, this would immediately eliminate the knowledge bottleneck that slows the entire field. Artificial intelligence (AI) could be a great tool to help the industry execute on this extremely challenging task.

Without such mechanisms, the CGT landscape will continue to mature at a fraction of its potential speed.

Industrialization: The Missing Evolution

Despite significant scientific innovation, CGT manufacturing remains structured like a cottage industry (operator dependent, variable, and highly manual). Every product is a bespoke recipe; every process is a one-off; every CDMO is expected to adapt around client variability.

The participants argued that for the field to progress, the industry must shift away from this artisanal model toward true industrialization: standardized unit operations, digital traceability, predictive analytics, and manufacturing platforms that automate 70%–80% of the process while allowing 20%–30% variation.

This mirrors the evolution of biologics: standard cell lines, standard media compositions, predictable unit operations, and a supply chain that can support global commercial demand.

The Real Barrier: Capacity, Not Cost

A recurring claim in CGT is that high COGS is the barrier to patient access. Participants disagreed sharply.

The evidence shows that where therapies provide clear clinical value, demand is overwhelming and growing rapidly. The number of patients treated by commercial CAR-Ts has grown 115% – 160% year-over-year, despite list prices unchanged from initial launch.

The bottleneck is not reimbursement; it is manufacturing supply. Only ~5% of eligible patients globally receive CAR-T today.

Improving COGS is important, but primarily for margin expansion, not accessibility.

The Facility Fallacy

Another myth hampering progress is the belief that biotechs must own their own manufacturing to be acquisition-ready. This misconception traces back to the earliest CAR-T acquisitions (e.g., Juno, Kite, Spark), where companies happened to have built in-house facilities.

The result has been billions of dollars sunk into facilities that create fixed cost structures and slow the field’s ability to adapt.

In contrast, CDMOs demonstrated their value during the COVID-19 pandemic, rapidly scaling manufacturing capacity in a way no single company could have achieved alone. The ability to scale up and down according to pipeline needs is a strength, not a vulnerability.

The field would advance faster if capital were directed toward process innovation, digitalization, and data infrastructure rather than redundant company-specific bricks and mortar.

The Digital Backbone: Enabling The Next Step Change

Participants repeatedly emphasized that the technologies needed for industrialization already exist: bioreactors, automated separation platforms, closed systems, predictive tools. None of these are the limiting factor anymore.

What is missing is the digital layer that makes these elements cohere into a modern manufacturing ecosystem.

A digital backbone would:

• enable real-time visibility across unit operations
• allow data reuse across runs rather than resetting learning each time
• power digital twins for scenario testing
• support autonomous documentation and compliance
• orchestrate supply chain and scheduling
• dramatically reduce operator burden and manual error.

CDMOs and biotech programs that invest in digitalization will scale faster, fail less often, and become attractive partners in a future where distributed manufacturing becomes a reality.

The First 70%

The foundational premise: 70% – 80% of CGT manufacturing steps are common across programs. Standardizing and digitizing these steps would provide enormous stability. The remaining 20% – 30% (product-specific biology, vector design, cytokine concentrations, or activation strategies) become manageable and transparent variations rather than architectural reinventions.

This creates an environment in which optimization accelerates instead of repeatedly restarting from zero.

Practical Steps Toward An Industrialized CGT Ecosystem

Industrialization won’t come from a single breakthrough. It will come from a series of coordinated, pragmatic shifts across sponsors, CDMOs, academics, regulators, and technology providers. Based on the Foundry discussion, five tangible steps emerged, each achievable today and each capable of moving the field closer to real scale.

---

Specialize CDMO business models: stop trying to serve everyone

One-size-fits-all CDMOs create misalignment, inefficiency, and frustration on all sides.
The panel was unanimous: the sector needs CDMOs to choose a lane.

• Early-stage CDMOs should focus on academic spinouts and seed-stage biotechs, with high-touch tech transfer, process maturation, and iterative development support.
• Late-stage/commercial CDMOs should be optimized for reproducibility, throughput, industrial discipline, and global regulatory expectations.

Those that remain generalists (working with both early-stage science projects and commercialization-ready programs) will continue to fall through the cracks. Specialization is the precondition for scale.

---

Build a culture of precompetitive data sharing

Every participant agreed on one uncomfortable truth: the industry wastes staggering amounts of capital repeating the same mistakes.

Practical steps include:

• crowdsourcing anonymized CQA/COA tables for CAR-T, HSCs, NKs, etc., led by journals such as Cytotherapy
• encouraging companies to publish standard practices (e.g., mycoplasma modalities, cytokine ranges, viability expectations) that are not true trade secrets
• increasing industry representation in academic/regulatory working groups, ensuring guidance reflects commercial realities
• long-term exploration of an FDA-hosted, non-identifiable data repository for in-process and release data.

Without shared knowledge, industrialization is impossible. With it, the learning curve collapses.

---

Rebuild trust through radical transparency

The biotech–CDMO relationship is notoriously low-trust. One deviation, one failed batch, and relationships collapse.

To industrialize, the field needs:

• clear communication paths defined in quality agreements (QAAs), not improvised under pressure
• early expectation setting around process changes, scalability, and regulatory requirements
• explicit discussions about goals:
• Is the biotech trying to get acquired? Reach a BLA? Generate Phase 1 data?
• Each requires a different CMC strategy.
• CDMOs that proactively challenge academic or unscalable processes rather than simply “executing instructions.”

Industrialization requires predictability. Predictability requires trust built on transparency, not assumptions.

---

Invest in the digital backbone the industry has lacked

The group was clear: the next decade will be defined not by new bioreactors but by digital infrastructure.

A mature CGT industry needs:

• manufacturing execution systems (MES) that eliminate paper, track in-process signals, and make deviations visible in real time
• knowledge reuse across runs and products: the ability to apply learnings from run 1 to runs 2 – 200
• digital twins to test process changes without risking patient material
• supply chain orchestration technologies that can coordinate materials, schedules, apheresis slots, and logistics across multiple sites.

Digitalization is no longer optional; it is the only way to scale complex biological systems safely and efficiently.

---

Reframe the cost conversation: fix supply, not COGS

A key revelation in the discussion: COGS is not the barrier to patient access.

Demand for commercial therapies is growing at 80% - 120% per year, far outpacing the industry’s capacity to supply them. Cost matters primarily for investor confidence and margin expansion, not as a determinant of whether patients receive therapy.

The real priorities are:

• increasing manufacturing throughput
• reducing failure rates and variability
• shifting processes toward industrial design rather than artisanal execution
• avoiding unnecessary capital spending (e.g., every biotech building its own plant)
• simplifying clinical pharmacy handling through nominal dosing.

Fix the supply bottleneck, and access expands. Continue obsessing over COGS in isolation, and the field remains stuck.

---

Conclusion

The cell and gene therapy field is at a critical inflection point. Scientific innovation has matured, but manufacturing and operational maturity lag far behind. CDMOs absorb blame for failures rooted in problems they did not create; biotechs are frustrated by constraints and regulatory expectations they did not expect; regulators continue to receive inconsistent filings that reflect a fragmented, opaque industry.

But progress is achievable. The Foundry discussion underscored that with candor and alignment, the industry can move decisively toward scale.

Scaling cell and gene therapies will require:

• a shift in mindset from bespoke science to industrialized manufacturing
• a commitment to transparency and data sharing
• greater specialization and division of labor
• a modern digital infrastructure as the new backbone
• and, above all, a realignment between biotechs and CDMOs grounded in trust, shared goals, and realistic expectations.

If these changes take hold, the industry can move from serving 5% of eligible patients to serving 50%, and eventually far more.

The science already works. The challenge ahead is to build the system that allows it to scale.

About The Author:

Arnaud Deladeriere, Ph.D., is principal consultant at Cell&Gene Consulting Inc. Previously, he was head of MSAT and Manufacturing at Triumvira Immunologics, and before that, manufacturing manager at C3i. He received his Ph.D. in biochemistry from the University of Cambridge.