I spent the week at NIIMBL. What follows is my read on where biomanufacturing is actually heading, five signals I heard on the floor and in the breakouts, and the move we are making at Alphinity in response.
1. ADCs stole the show.
Antibody-drug conjugates ran away with the agenda. Not as a niche modality any more, but as large-scale manufacturing, with capacity targets in the range of 20 to 30 kg per week.
2. The downstream tsunami.
Upstream titers keep climbing, and the room repeatedly named downstream as the bottleneck about to break. Concentration, buffer exchange, and polish steps are where the next round of capacity gets found or lost.
3. Standardization was a dominant theme.
Strong, shared appetite across the floor for standardized unit operations and platform approaches that take cost and variability out of a build. I hear the argument. I also have a caveat, below.
4. Continuous processing, and controlling it.
Downstream continuous manufacturing was a major focus, with breakout groups working through how to run it and how to control it. The hard part is the recurring one: getting components from different vendors to behave as one connected process.
5. The software question.
Much of the conversation circled the platform that has become the industry's default for process software, and what teams now want from it: cloud-based, connected across the process, and AI brought in on their own terms, controlled and secure rather than an open frontier model.
Gene therapy (AAVs) and continuous mAb processing rounded out the major platform themes. Different molecules, same underlying question: how do you get a fragile, high-value product through a downstream train without paying for it in yield and potency?
The one take I would push back on.
Standardization is a genuinely good idea, and I support it in principle. The caveat is what it optimizes for. A single standard system is optimized for the average molecule, and the processes that matter most to the people at NIIMBL are precisely the ones the average system fails. Enveloped viral vectors, LNPs, concentrated ADC intermediates, cell therapies: these are not average molecules, and their failure modes are not the ones a platform system was tuned to handle. Standardize the interfaces and the data. Do not standardize away the fragility of the product.
"Everyone wants one standard system. But standardization optimizes for the average molecule, and no one is trying to make the average molecule. The processes that matter are the ones a standard system fails."
On standardization
The throughline.
The industry is scaling its most fragile, valuable modalities to production volumes, and the bottleneck is shifting downstream. That is the work gentle, single-use processing is built for.
Our response at Alphinity is more of what we are already doing. Our Prove-It campaign puts our single-use TFF system on your bench, at no cost, and runs your process, unchanged, on your material, head to head against your existing equipment. You see what it does to your yield and potency on your own product, and decide from there. We are opening more slots to more teams over the second half of the year.
The other half of our response is public. Our Fundamentals Hub is where we publish the engineering behind why pumps damage product, how pulsation scales from bench to production, what pressure-control accuracy actually costs in filtration, and why some architectures cavitate and others do not. No gate. Just the physics, the tests that confirm it, and how to design it out.
If one of these five signals is loud in your own work right now, I would like to hear which. What you tell us shapes what we build and what we write next.