Complexity is inherent to cell and gene therapy and that includes its manufacturing, where moving towards producing products at a commercial scale continues to challenge companies in the sector.
To improve their operations, many firms are looking to artificial intelligence (AI) and several of them took to the stage at the BIO International Convention (BIO 2026) on 22 June, 2026 in San Diego, California, for a session on ‘Manufacturing the Future’.
Go deeper with GlobalData
One of those on the panel was Fraser Wright. The co-Founder and chief technology officer of Spark Therapeutics, whose Luxturna (voretigene neparvovec) was the first US Food and Drug Administration (FDA)-approved gene therapy, Wright is currently the scientific co-Founder and chief gene therapy officer at Kriya Therapeutics.
The North Carolina-based biopharmaceutical company is seeing several tangible benefits from its use of AI in manufacturing. These include design identification and harnessing in-house work to create their own data lake, as Wright told the BIO 2026 audience.
“In vector design, while these techniques have been in existence before, they’re much easier to implement now that we have such a powerful tool in AI.
“In one of our programs, we actually went through I think it was 80 different constructs to look at their manufacturability. It’s [still] a little bit manual, but using AI helped us identify the best constructs.”
As part of its work with AI Kriya made “a big investment” in a fully integrated manufacturing platform, one that covers cells all the way to a viable product. This in turn generated a huge resource of data.
“Based on many engineering and GMP-like runs, as well as GMP runs, we’ve accumulated a very, very large data lake. I think it’s approaching 100 million data points. These numbers will build, of course, and we feel this is a very powerful tool.
“We have hundreds and hundreds of points at a single process step. We can understand what correlates with better yield, for example, and so those are very, very important considerations for us, and we’re continuing to use advanced enterprise level AI to analyse our data and feed back into optimised, efficient processes.”
Another of the companies on the BIO 2026 stage was Opus Genetics, which is working to develop first-in-class gene therapies for inherited retinal diseases. The North Carolina clinical-stage biopharmaceutical company’s lead candidate is OPGx-LCA5, which is in Phase III clinical trials for Leber Congenital Amaurosis Type 5 (LCA5).
A key feature of the company’s work requires very small batch sizes, consequently producing a very limited batch history.
“For us, the challenges always are analysing your data in a meaningful way when you have such limited data available. So, we’ve used AI at this point to analyse data and also prepare for PPQ [process performance qualification] … [and] pulling together our CQA [critical quality attributes] data … so that we can do that more efficiently.
“We have a very small team at Opus, we’re about 38 or 39 people right now. At this point, we have five in CMC, and so for me, AI is a way to streamline my operation and make the staff more efficient, and then you get a better data output.”
He added: “We look at it from a cost savings perspective. I mean, if we can streamline the PPQ process and come away with meaningful risk assessments that allow us to do reduced testing for some part of our platform, then we can save costs.”
A similar sized company to Opus that was also represented on the BIO 2026 panel was the 35-person Epicrispr Biotechnologies, a clinical-stage company working on gene-modulating therapies. Epicrispr’s lead candidate EPI-321 is in Phase I/II trials for the neuromuscular disease facioscapulohumeral muscular dystrophy (FSHD).
As a startup, Epicrispr leans on a contract manufacturing and development organization (CDMO) for its bulk manufacturing, so its own application of AI is geared towards optimising its own processes, as vice president of Technical Operations Dipali Patel outlined.
“We implemented QbD (quality by design) principles and [using] advanced tools like AI really helped us generate smarter experiments, smarter [experiment] designs, and we’re able to loop that and feed that data back to really predict the next layer of studies to do in that same time and really cut the number of experiments needed to get there.
“We were able to, one: identify our critical process parameters, and two: understand synergistic relationships between the key process parameters; and then three: help set thresholds. This is in a much more concise time frame than I’m used to in my past life.”
However, Patel cautioned, “the stakes matter” and she urged particular caution around using AI to generate regulatory documents.
Companies such as Kriya, Opus, and Epicrispr operate in a space where bringing new products to market is more than usually challenging. Indeed, last year saw a decline in US approvals after four consecutive years of record approval numbers, with four approvals, compared to eight in 2024, according to GlobalData’s latest annual New Drug Approvals and Their Contract Manufacture trend report.
Against that background, careful use of AI, where both risks and benefits are understood, has begun to power a technology-enhanced future for cell and gene therapy manufacturing, one that could free up time and resources to focus even more firmly on improving product success rates.
