How CGT Developers Should Think About Manufacturing Models, Hybrid Strategies, And Scale

By Erin Harris, Editor-In-Chief, Cell & Gene
Follow Me On Twitter @ErinHarris_1

During our recent Cell & Gene Live, How CGT Developers Can Choose the Right Manufacturing Model for Their Pipeline, Omkar Kawalekar, Ph.D., Manufacturing & Supply Chain Lead at Deloitte Consulting, Zoe Zheng, Executive Director, Pipeline Technical and Strategic Lead at Legend Biotech, and I had a wide-ranging discussion on how CGT developers should think about manufacturing as they move from early development to commercialization.

Our conversation opened with the key reminder that manufacturing strategy should not be treated as a simple build-versus-buy or facilities decision. “The most expensive mistake is treating your manufacturing model decision as a facilities decision rather than an operating model decision,” said Dr. Kawalekar. He argued that companies often move too quickly to lock in a manufacturing approach before they have fully considered process requirements, biological variability, patient logistics, and long-term scalability.

How Early Clinical Choices Can Limit Long-Term Commercial Scalability

A recurring theme throughout our event was that many CGT companies optimize too heavily for near-term clinical needs. That can be understandable in Phase 1, when the priority is speed, patient dosing, and cash conservation, but it can create major problems later if the process, supplier base, or digital infrastructure cannot support comparability or multi-site execution. When describing the hidden cost of choosing a strategy that works for the clinic but not for the market, Dr. Kawalekar shared, “They may save three months at the beginning, but they lose on 18 months or more later.”

He also stressed that process changes in CGT are rarely isolated. A single adjustment can ripple into analytics, comparability work, regulatory engagement, tech transfer, digital system updates, and even clinical bridging studies. In his view, the best manufacturing models are designed with the product life cycle in mind, not just the next milestone. “Design the manufacturing model like you’re planning a city, not pitching a tent.”

Why Operational Heroics and Manual Workarounds Signal an Unready CGT Manufacturing Model

The panel then explored how to tell when a company is still thinking too clinically and not commercially enough. Dr. Kawalekar pointed to operational heroics as one of the biggest warning signs. If senior scientists are constantly troubleshooting batches, QA is processing exceptions one by one, and supply chain teams are managing patient slots through spreadsheets, the company may be getting through Phase 1, but it is not ready for launch-scale complexity.

He also said that treating process instability as normal can become a dangerous habit. In CGT, some variability is expected, but persistent deviations, frequently changing batch records, and weak process controls suggest a company is still operating in survival mode. He offered a simple but critically important diagnostic question: if demand doubled tomorrow, or if a site went down, would the team know what to do in the first hour? If not, the model is still clinical, not commercial.

How Modality Changes the Manufacturing Decision Framework for Autologous and Allogeneic Therapies

Zheng brought a technical and practical lens to the discussion, noting that modality fundamentally affects the manufacturing decision framework. Autologous therapies demand patient-specific manufacturing and logistics, while allogeneic and in vivo approaches may offer greater flexibility for scale and inventory management. “Modality is a significant factor in the decision making,” she said. “The patient experience, access, and product consistency remain critical regardless of platform.”

She also emphasized that developers need to consider process complexity, process stability, chain of identity, scalability, and quality oversight early. For autologous programs in particular, Zheng said that fully closed, highly automated systems can help reduce variability across sites. At the same time, she warned that QA oversight becomes especially challenging in decentralized models unless companies invest in the right digital systems and controls.

What Hybrid Manufacturing Really Means in Practice Across CGT Programs

Our discussion then shifted to hybrid manufacturing, a term Dr. Kawalekar described as one of the most overused and underdefined in CGT. He outlined four hybrid archetypes, including asset-level hybrid, process-step hybrid, network hybrid, and lifecycle hybrid. Each of those can be legitimate, depending on the product, the stage of development, and the operating context.

He said what does not work is a fragmented hybrid model with multiple vendors, inconsistent processes, different data standards, and no meaningful digital integration. “Hybrid done right is a jazz ensemble; different instruments, but one conductor, one score,” he said. “Hybrid done wrong, looks like a traffic jam.” Hybrid can be powerful, but only if the model is designed intentionally and governed tightly.

Why CGT Developers Should Retain Process Knowledge, Analytics, and Supply Chain Orchestration

When asked what capabilities should never be outsourced, both experts agreed that companies must retain deep process and product knowledge internally. Zheng said developers should always keep deep knowledge and retain key analytics, proprietary technologies, and testing capabilities. Dr. Kawalekar expanded on that idea, saying that knowledge, analytical strategy, and real-time decision making should stay close to the sponsor.

He also identified supply chain orchestration as something that should never be left entirely to a CDMO. Knowing where patient material is, what state it is in, and what happens if something goes wrong must remain with the sponsor. “Everything else negotiable, but the control of critical information and decisions cannot be outsourced away,” Dr. Kawalekar said.

How Supply Chain Orchestration Becomes a Control Tower Problem in CGT

Dr. Kawalekar described supply chain orchestration as one of the hardest challenges in manufacturing because the biggest vulnerabilities usually occur at the intersections. Patient scheduling may shift, raw materials may be delayed, QC release may become a bottleneck, and each issue alone may seem manageable. Together, they can cascade into a serious operational risk.

He argued that CGT supply chains need to be viewed as a control tower problem, not a shipment problem. That means building end-to-end visibility into patient slots, material status, batch progress, and release testing. For smaller CGT biotechs with limited resources, he recommended prioritizing investments by patient impact, regulatory risk, and reversibility. The first move should be visibility; the second should be chain of identity and chain of custody controls; the third should be risk management for critical materials.

How CGT Developers Can Evaluate Whether a CDMO Is Truly Transparent About Capacity

The panel also tackled the CGT developer-CDMO relationship. When asked how CDMOs could better support developers, Dr. Kawalekar urged the industry to be more transparent. He said CDMOs should stop selling capacity and start selling capability. Developers need partners who understand process complexity, share data openly, and invest in people and digital infrastructure.

He also offered a practical way for sponsors to evaluate whether CDMOs are overstating their capabilities by asking about the last three batch failures, what changed as a result, and for actual slot utilization over the past 12 months. Those questions reveal more than a glossy deck ever will. In Dr. Kawalekar’s view, the right CDMO welcomes those questions; the wrong one is uncomfortable answering them.

Why Phase 1 Teams Should Build for Optionality While Preparing for Scale

Dr. Kawalekar outlined the innovations he believes will most shape commercialization in the next several years, including closed automated manufacturing platforms, AI-driven analytics and release support, and digital orchestration across clinical scheduling, manufacturing, logistics, and quality. Still, he warned against overhyping AI. “AI doesn’t create clean data,” he said. “It consumes that data.”

Zheng added that companies that scale globally will need strong portfolio planning, capacity management, standardized quality systems, data infrastructure, and cross-functional coordination. She also stressed the importance of culture, saying that operational excellence requires teams that can collaborate effectively as the organization grows. Her final advice for Phase 1 companies was to align early and build with the future in mind. Dr. Kawalekar echoed that point, urging developers to design for optionality and start building commercial discipline from day one.

In CGT, scale is not achieved by science alone. It is achieved by pairing strong science with manufacturing models, data systems, and operational discipline that can support the journey from clinic to commercial reality.

Watching the full Cell & Gene Live is the best way to capture the nuance, context, and practical detail behind the discussion. Click here to view the full-length presentation.