For much of the past decade, success in cell and gene therapy was measured by scientific milestones. Could researchers engineer a cell to attack cancer? Deliver a gene to the right tissue? Turn a promising concept into a clinical response?
Those questions haven’t disappeared. But as the field matures, they are no longer the only ones that matter.
Today, the therapies most likely to reach patients are those that can be manufactured reliably, transferred efficiently between sites, scaled economically, and, ultimately, commercialized.
This reshaping decisions across the development process. Manufacturability is being considered much earlier. Technology transfer is becoming a strategic capability rather than an operational handover. Platform choices are increasingly influenced by commercial realities, while AI is finding its place as a practical tool.
To understand how this transition is changing the industry, PharmaSource spoke to four experts working across the cell and gene therapy ecosystem: Michael Mercaldi, who leads technical operations at gene therapy startup Torque Bio; Gilad Beck, who has overseen technology transfers across cell therapy manufacturing sites in the US and Europe; Shashi Murthy, a three-time founder who has spent his career developing manufacturing technologies for advanced therapies; and drug delivery scientist Mruganka Parasnis, whose work focuses on translating lipid nanoparticle formulations from the laboratory to industrial manufacture. Together, the lessons they shared offer a roadmap for sponsors, developers, and CDMOs navigating the next phase of the industry’s evolution.
Lesson 1: Manufacturability Starts in Product Design
For years, manufacturing challenges in cell and gene therapy were largely viewed as downstream problems. Low yields, high cost of goods, inconsistent product quality, and difficult scale-up were issues to solve once a therapy had been shown to work.
The four experts interviewed for this article suggest that assumption no longer holds.
Instead, they argue the biggest manufacturing decisions are increasingly being made much earlier, during product design itself. Rather than asking how to manufacture a promising therapy, developers are beginning to ask whether the therapy has been designed to be manufactured in the first place.
After more than 15 years in process development and manufacturing, Michael Mercaldi believes improvements in manufacturing platforms have fundamentally changed where the industry’s biggest bottleneck lies.
“The biggest bottleneck, in my opinion, starts further upstream from the manufacturing floor all the way to the product design.”
Modern manufacturing platforms, he argues, have become increasingly robust and standardized. The limiting factor is often no longer the process, but the construct itself. Some molecules simply carry “inherent liabilities” that result in poor yield or product quality, regardless of how sophisticated the manufacturing platform becomes.
At Torque Bio, that changed the development strategy. Rather than building a bespoke manufacturing process around each candidate, the company deliberately selected constructs that could work within its CDMO partner’s existing platform. According to Michael, the result was “minimal process and analytical method development and rapid entry into GMP manufacturing.”
Mruganka Parasnis reaches the same conclusion from a different perspective. Working on lipid nanoparticle formulations, she has spent years optimizing delivery systems in the laboratory, only to see how quickly those formulations encounter manufacturing realities.
“A highly effective formulation that cannot be stored, transported, or manufactured consistently may never reach patients,” she says. “Developability should be considered alongside efficacy from the earliest stages.”
That distinction between discovering something that works and discovering something that can become a medicine runs throughout her work. Academic research naturally prioritizes innovation, but industrial development must also account for stability, reproducibility, regulatory expectations, and commercial manufacturing. As she puts it, “The biggest challenge is often not discovering something that works. It is understanding what will continue to work when manufacturing, regulatory, clinical, and commercial realities are introduced.”
Shashi Murthy, who has founded companies developing manufacturing technologies rather than therapeutics, arrives at much the same conclusion from the opposite direction. In his view, manufacturing strategy should never begin with automation, equipment selection, or production capacity. It should begin with the commercial case.
A therapy intended for a handful of patients requires a very different manufacturing strategy from a platform expected to support multiple products and large patient populations. Companies that chase sophisticated automation too early, he argues, often solve the wrong problem. For small programs, “simplicity and robustness rather than scale” may be the better choice, while broader pipelines justify investment in integrated manufacturing platforms and long-term technology partnerships.
His advice is: “Know what you’re building and why.”
Taken together, these perspectives point to a broader shift in how manufacturing is being approached across the industry. Success is becoming less about rescuing difficult products with increasingly complex manufacturing processes and more about designing therapies that fit proven manufacturing platforms from the outset. Increasingly, manufacturability is no longer a downstream CMC consideration; it has become a design principle.
Lesson 2: Tech Transfer Is Mostly a People Problem
If manufacturability begins in product design, technology transfer is where those decisions are put to the test. It is also where promising programs can quickly come unstuck.
The instinct is often to see tech transfer as an engineering exercise: transferring a process from one site to another, validating equipment, reproducing analytical methods, and demonstrating consistent quality. Those tasks are undeniably complex, but the interviewees repeatedly pointed to a different source of failure. More often than not, they argued, technology transfer breaks down because knowledge, communication, and expectations fail to move as effectively as the process itself.
Gilad Beck has overseen technology transfers across more than a dozen cell therapy manufacturing sites in Europe and the US. Looking back, he believes the industry’s biggest mistake is surprisingly simple.
“People tend to underestimate timelines. That’s the biggest piece.”
Successful transfers, he argues, depend less on technical brilliance than disciplined project management. Maintaining “constant communication between all parties, and specifically the project managers” on both the sponsor and CDMO side is what keeps programs moving when inevitable problems arise. Equipment can be qualified and protocols rewritten, but without trust between teams, even relatively small deviations can become major delays.
Mruganka sees the same challenge from the formulation laboratory. Academic researchers often develop highly effective formulations, but much of the knowledge behind them never appears in a protocol. Instead, it exists as experience: subtle decisions, observations, and practical adjustments made over months of experimentation.
That becomes a problem when manufacturing moves elsewhere.
“Successful technology transfer is about transferring scientific understanding, not just protocols. If the knowledge transfer focuses only on the protocol and not the reasoning behind it, important details can get lost.”
She argues that one of the biggest risks during scale-up is assuming that reproducibility comes automatically. In reality, seemingly minor process parameters (flow rate, lipid concentration, or buffer exchange conditions) can have significant effects on product quality. Without a shared understanding of why those parameters matter, transferring a process becomes far more difficult.
It is also why Mruganka believes the strongest programs bring formulation scientists, analytical scientists, process engineers, and manufacturing teams together long before formal technology transfer begins. Working sequentially, she argues, simply allows knowledge gaps to accumulate.
Shashi approaches the same problem from the perspective of manufacturing technology. Having spent years developing automated platforms, he believes the easiest way to reduce tech transfer risk is to avoid unnecessary process changes altogether.
Where possible, companies should choose manufacturing technologies that scale using the same underlying principles from laboratory development through clinical manufacture. If preclinical, process development, and GMP production can all use variations of the same platform, the transition becomes significantly less disruptive.
“You’re not changing the process.”
It is a deceptively simple observation, but one that reflects a broader lesson emerging across the industry. Successful technology transfer is rarely about copying a manufacturing process exactly. It is about preserving understanding of why a process works, which variables matter, and how different teams interpret the same information.
Lesson 3: The Industry Is Redesigning Therapies Around Manufacturing
If manufacturability has become the defining challenge, the industry’s response is increasingly to design around it.
Rather than relying solely on better manufacturing technologies, developers are beginning to rethink the therapies themselves, choosing modalities, delivery systems, and product designs that reduce manufacturing complexity from the outset. The goal is no longer simply to improve manufacturing, but to make therapies inherently easier to manufacture.
Where the interviewees differ is in what that future will look like.
For Gilad, the direction of travel is clear. After more than a decade in cell therapy manufacturing, he sees the industry steadily moving beyond bespoke, patient-specific manufacturing towards platforms that promise greater standardization and scalability.
“A lot of companies are trying to make allogeneic cell and gene therapy work,” he says, while also pointing to the growing interest in in vivo approaches that generate or modify therapeutic cells directly inside the patient.
The appeal is obvious. Autologous therapies require manufacturing a unique batch for every patient, using starting material that varies from one individual to the next.
“There’s zero tolerance for errors with autologous therapies.”
Allogeneic manufacturing offers greater consistency by starting with a standardized cell source and producing therapies at a larger scale. But Gilad is careful not to overstate its maturity. Despite years of investment, every approved cell therapy today remains autologous, with allogeneic programmes continuing to face significant safety and regulatory challenges.
Shashi is even more cautious about framing the industry’s future as a simple choice between autologous and allogeneic manufacturing.
“I don’t think it’s an auto versus allo question.”
Instead, he argues that companies should work backwards from their pipeline rather than forwards from the technology.
A single therapy for a small patient population demands a very different manufacturing strategy from a platform supporting multiple products. Four related allogeneic programs may justify investment in shared manufacturing infrastructure. A mixed pipeline spanning autologous and allogeneic products, each serving different markets, presents a far more difficult planning challenge.
For Shashi, manufacturing decisions should be driven less by scientific fashion than by commercial reality.
Michael takes the argument one step further. In his view, the biggest opportunity is not choosing a different manufacturing platform but designing therapies that fit existing ones with commercialization in mind.
This means selecting constructs whose process productivity can satisfy projected commercial demands right at the start of development and that its CDMO partner has the scale to satisfy these demands. By taking this approach, a clear path toward commercial supply can be realized which streamlines the entire development pathway of the drug and brings life changing therapies to patients sooner.
Historically, manufacturing was expected to adapt to the biology. Increasingly, the biology is being adapted to manufacturing.
The same philosophy is shaping delivery technologies across the sector. Non-viral delivery systems, improved viral vectors and in vivo approaches all promise to remove some of today’s most complex manufacturing steps, whether by simplifying production, reducing dose requirements or eliminating ex vivo cell manipulation altogether.
No one interviewed suggested that a single modality would dominate. In fact, Shashi cautions against becoming too captivated by the latest platform, noting that biologics, small molecules and conventional cell therapies continue to advance alongside newer approaches.
What the interviewees do agree on is that manufacturability is becoming a selection pressure in its own right. Therapies are increasingly being judged not only by their biological potential, but by their ability to be manufactured consistently, scaled economically, and delivered to patients.
The industry’s next breakthroughs, in other words, may come as much from simplifying the product as from improving the process.
Lesson 4: AI Is an Enabler, Not the Solution
Across the industry, AI is increasingly being viewed as another tool for improving decision-making, reducing manual work, and supporting manufacturing teams. Useful, certainly, but only if the underlying manufacturing process is already robust.
Shashi has watched successive waves of manufacturing analytics arrive over the past decade.
“It used to be called advanced analytics… then machine learning. And now it’s AI. It’s kind of all the same thing.”
For him, the technology is valuable precisely because it solves specific problems rather than trying to reinvent manufacturing. Whether optimizing workflows, supporting patient-specific manufacturing decisions, or accelerating data analysis, AI should ultimately be judged by the same metric as any other manufacturing investment: does it reduce cost, improve quality, or shorten timelines?
“AI shouldn’t be the main story. I think that’s where sometimes things get overhyped.”
Gilad sees the same distinction between aspiration and reality. While some manufacturers are beginning to apply machine learning directly to bioprocesses using AI-enabled bioreactors to optimize cell culture conditions in real time, he believes many of today’s most successful applications sit alongside manufacturing rather than inside it.
At Galapagos, for example, his team built an AI-powered troubleshooting tool using years of manufacturing experience. Operators could query previous batch deviations during production, receiving recommendations based on historical cases before escalating more complex issues to human experts.
“It wasn’t replacing people,” Gilad explains. It was making existing expertise easier to access.
For formulation scientist Mruganka, AI’s greatest limitation is not the algorithms themselves but the data available to train them.
“AI is only as good as the data behind it.”
Generating high-quality experimental datasets remains one of formulation science’s biggest challenges, meaning traditional experimental design still has a central role to play.
Rather than replacing Design of Experiments (DoE), she sees the two approaches as complementary. Well-designed experiments generate the structured data that AI models need, while AI can then identify patterns that might otherwise go unnoticed. “Robust experimental design, such as Design of Experiments, provides the high-quality datasets needed for AI to be effective in formulation development,” she says.
Michael is perhaps the most optimistic about AI’s near-term impact, particularly beyond the manufacturing floor. At Torque Bio, his focus includes using enterprise AI to accelerate CMC timelines and automate technical documentation, areas where repetitive, knowledge-intensive work can often consume significant resources without directly improving the product itself.
Lesson 5: Every Manufacturing Decision Starts with the Commercial Case
By the end of each interview, the conversation inevitably returned to the same question: not whether a therapy could be manufactured, but whether it could be manufactured sustainably.
That may sound obvious, yet it represents a shift in how many cell and gene therapy programs have traditionally been developed. Scientific feasibility has long been the primary objective. Increasingly, however, manufacturing strategy is being shaped by economics from the very beginning.
For Shashi, that should be the starting point for every program.
“Consider the commercial case right off the bat and work backwards from there.”
That means understanding not only the size of the patient population, but also the broader pipeline, expected product lifecycle, and the investments a company is realistically prepared to make.
A therapy intended for a few hundred patients may never justify highly automated manufacturing. In those cases, Shashi argues, companies should resist the temptation to pursue the most sophisticated technology.
“Don’t chase automation. Just go for simplicity and robustness rather than scale.”
Conversely, organizations developing multiple therapies based on similar manufacturing processes should think strategically about platform investments, equipment choices, and long-term partnerships with technology providers. The manufacturing strategy should reflect the commercial opportunity, not the other way around.
Michael reaches a similar conclusion from the perspective of gene therapy manufacturing. Although manufacturing platforms have improved dramatically over the past decade, cost remains the factor that will determine whether advanced therapies move beyond rare diseases into larger patient populations.
He expects continued reductions in the cost of viral vector manufacturing, driven by improvements in production platforms and technologies such as stably transfected producer cell lines. At the same time, he believes CDMOs face a growing challenge as the market diversifies.
Some gene therapies may serve fewer than ten patients. Others could eventually treat tens of thousands.
“Can the CDMO run a 2,000-litre reactor for one client and a 10-litre reactor for another client in the same facility?”
That widening gap, he argues, will place increasing pressure on manufacturing networks and may accelerate consolidation as CDMOs seek to offer expertise across multiple genetic medicine platforms rather than specialising in individual modalities.
Mruganka brings the discussion back to patients. Manufacturing delays are often discussed in terms of timelines, budgets, and operational complexity, but she argues they ultimately have a human cost.
“Every delay in scale-up, tech transfer, or process optimization can delay access to therapies for patients.”
That is why she believes manufacturability should never be viewed as separate from scientific innovation. A formulation that cannot be manufactured reproducibly, characterized robustly, or transferred successfully may never become a medicine, regardless of how promising it appears in the laboratory.
Gilad adds another dimension: regulation.
After years of working across cell therapy manufacturing, he has become convinced that engaging regulators early is not simply good practice, but it is good business. Too often, companies treat regulatory engagement as a milestone rather than an ongoing conversation, only discovering major issues after significant time and capital have already been invested.
Early dialogue, he argues, helps companies identify problems while they are still relatively inexpensive to fix.
A manufacturing or regulatory setback that delays a program by one or two years may postpone approval or determine whether a company survives at all.
Across all four interviews, the same pattern emerges. Product design, technology transfer, automation, AI, and manufacturing platforms all matter. But none of those decisions exist in isolation.
Each is ultimately judged against the same question: can this therapy be manufactured reliably, affordably, and at the scale needed to reach the patients it was designed to help?
Increasingly, that commercial reality is shaping manufacturing strategy just as much as scientific possibility.
Michael Mercaldi will be speaking on the cell and gene therapy panel at CDMO Live Americas 2026, taking place in Boston, October 20th-21st. Download the agenda or book your ticket now.
