“The capacity globally is more constrained on drug product than on drug substance right now. Everything is under construction, but a lot of that build-out is still ongoing.”
Scott Hilderbrand leads the CMC function for antibody-drug conjugate (ADC) programs at Crescent Biopharma. Over the last couple of decades, he has moved from early click-chemistry research at Massachusetts General Hospital to leading early-phase conjugation development at ImmunoGen and CMC for antibody-oligonucleotide conjugates (AOCs) at Dyne Therapeutics.
The ADC pipeline is surging, and manufacturing capacity is struggling to keep pace. With hundreds of antibody-drug conjugates in Phase 1 and Phase 2, alongside a fresh wave of dealmaking, from Novartis’ acquisition of Myricx Bio and GSK’s $10.6 billion Nuvalent deal to the Lonza–Engitix ADC license, demand for conjugation and fill-finish slots is rising faster than qualified capacity. For small and mid-sized biotechs, choosing the wrong CDMO can therefore have severe consequences. As Scott argues, the critical question has shifted from “Can this be made?” to “Who can make it reliably, and will the drug-product capacity be there when I need it?”
Design ADCs for Differentiation and Efficacy
The instinct at a small biotech, Scott explains, is to worry that a novel linker or payload will prove impossible to manufacture, but he disagrees. The ADC field is old enough, and crowded enough with approved products, that a truly undevelopable molecule is now a rare event.
“In CMC and tech ops, I try to stay agnostic on the molecule and the chemistry,” he says. “If a molecule can be made at research scale and it shows superiority to what gets screened out, it’s our responsibility to figure out how to manufacture that specific molecule and scale it up effectively.”
He points to the arc of the field for support. “ADCs have been around since ImmunoGen was founded in the late 1980s. There are 15 or 16 commercially approved ADCs globally and growing. The likelihood of a research team coming up with an ADC that’s completely not manufacturable is very low at this point.”
Most programs still run on auristatin or maytansine payloads, or the newer topoisomerase-inhibitor class, using well-understood Val-Cit chemistry. When companies innovate on a linker or payload, as Novartis is betting on with Myricx’s payload technology, Scott sees two motives, and neither is “proving it can be built.” One is carving out freedom to operate and patent space in a competitive field. The other is moving the needle on the payload–linker–target combination itself. Manufacturability is largely a solved problem; differentiation and efficacy are the open ones.
Match the CDMO to the Molecule
Because manufacturability is rarely the binding constraint, Scott treats CDMO selection as a sliding scale set by how novel the chemistry is and how fast the company wants to move.
At one end sits standard chemistry. “If it’s a Val-Cit MMAE, DAR4 interchain cysteine conjugate, process development should be fairly straightforward, so you may want a larger, established CDMO that’s ready to transition into late-phase and commercial,” he says. Off-patent building blocks make this route more common: trastuzumab already anchors several ADCs, and Val-Cit-MMAE chemistry is coming off patent soon, lowering the barrier to a fast, conventional program.
At the other end sits novelty. “If it’s new chemistry where you don’t yet understand how the molecule will behave, you need a partner with a well-developed process development organization, or a smaller CDMO that gives you the flexibility to change paths as you generate data.” The mistake is defaulting to a big, commercial-ready name for a molecule that still needs process-development experimentation or forcing an exotic molecule through a partner with no research bench to interrogate the chemistry.
The same logic extends to adjacent modalities. Scott’s AOC work at Dyne taught him that the conjugation toolbox is shared, but molecules behave differently in CMC: ADC development is largely about masking payload hydrophobicity, while a highly charged oligo payload creates a different set of conjugatability problems. The point for a sponsor is the same: pick the partner whose technical depth matches your molecule’s actual failure modes.
The Quality System Protects Clinical Supply
Once the shortlist fits the molecule, Scott’s selection criteria for a small-to-mid biotech are less about capacity slides and more about partnership and quality systems.
“From a smaller company’s perspective, one of the key things is finding a partner that’s willing to work with you and has expertise that’s complementary or adjacent to your own, because at a small biotech, you don’t have an expert in everything,” he says. A sponsor team of a handful of people needs a CDMO with the bandwidth to proactively spot process risks and opportunities the sponsor can’t staff for.
Scott also explains that how the partner behaves when something goes wrong is key to his selection criteria. He wants to see the track record on GMP manufacturing, how deviations are documented, and how investigations are run.
“By the time a major deviation happens, you’re often already in the clinic, and you don’t necessarily have an unlimited supply of drug substance and drug product,” he says. “You want to know your partner can troubleshoot, solve the problem quickly, and get the program back on track, so you don’t stock out of drug for the patients.”
Recent headlines underline the stakes of a fragile program: ADC Therapeutics lost more than half its market value after a Phase 3 safety readout, and researchers continue to surface tolerability signals such as severe neutropenia. A supply interruption on top of a clinical setback is the kind of compounding risk a mature quality system exists to prevent.
The Crunch is on Drug Product
Scott has watched the CDMO landscape build from nothing. ImmunoGen had to construct its own GMP capacity because no conjugation network existed. The Adcetris and Kadcyla approvals validated the modality, and Lonza, BSP Pharmaceuticals, and Piramal built up commercial ADC capability over the following decade.
The problem now is that demand has outrun that seasoned tier. “There’s still only a handful of CDMOs that are seasoned and able to do commercial manufacturing,” he says. “There are a lot of up-and-comers now, but they’re at earlier stages of proving themselves at the same level as the established partners.” Larger biologics players (Samsung Biologics, WuXi Biologics, MilliporeSigma) are moving in, and deals like Lonza’s new ADC license with Engitix and Samsung’s investment in next-generation ADC work show the established names deepening their positions. But capability is still under construction, and the tighter constraint is downstream: drug product fill-finish, not drug substance conjugation.
Scott Hilderbrand will be speaking on the ADC Development Strategy panel at CDMO Live Americas 2026, taking place in Boston, October 20th-21st. Download the agenda or book your ticket now.
