“Process chemistry alone is no longer the whole story—it needs to be connected with engineering solutions, digital tools, and data utilization”
Stefan Randl, Chief Scientific Officer at Siegfried, joined the Swiss-based CDMO in January 2025 after 16 years at Evonik, where he led innovation management for healthcare, ran drug substance operations in the US, and headed sales across Asia.
Scientific Leadership at Scale
With the global small molecule CDMO market leader facing increasing time pressure from sponsors, emerging sustainability mandates, and the challenge of building strategic rather than transactional relationships, his scientific vision will shape how Siegfried competes in an evolving landscape.
In this conversation with PharmaSource, Stefan reveals how fundamental process chemistry remains central to competitive advantage, why integrated manufacturing accelerates development timelines, and where AI will actually create value versus where it remains overhyped.
The Integration Advantage
While Chinese CDMOs can undercut API pricing, Stefan argues integrated manufacturing, API through finished dosage form under one roof, creates distinct value that justifies premium positioning for specific projects.
“The majority of our business is not integrated,” Stefan acknowledges. “Drug substance and drug product are separated for many reasons—cost, technology, capacity, risk diversification. Pharma companies may also produce one themselves.”
However, Siegfried continues securing integrated projects for compelling reasons. Speed tops the list, particularly during development. “When you combine drug substance and drug product out of one CDMO, you minimize loss of time,” Stefan explains. “When you work with two different CDMOs, you have to build a buffer between campaigns to account for delays. There’s the risk of penalties if the drug substance provider delivers late to the drug product CDMO.”
Beyond scheduling efficiency, the integrated model simplifies supply chain coordination. “The CDMO takes full control of drug substance logistics throughout the entire life cycle,” Stefan notes. Quality management becomes more straightforward with unified documentation, deviation management, and a single negotiation process for commercial agreements.
Technical interdependencies provide another driver. “Drug substances are becoming more and more complex, and process parameters can impact the formulation and drug product production more directly,” Stefan explains. “Enabling technologies such as micronization and spray drying sit at the interface between drug substance and drug product. In these cases, it’s easier when one CDMO takes care of both processes.”
Sustainability as Competitive Differentiator
Siegfried considers sustainability as a business advantage. Scope 3 accounts for a majority of pharma industry emissions, making a CDMO’s manufacturing processes central to sponsors’ sustainability commitments.
“The good thing with sustainability is that it often goes hand in hand with cost savings,” Stefan observes. “The more efficient, the shorter, the higher yielding, the cheaper the processes are, the more sustainable it is as well.”
However, exceptions exist. Some highly effective solvents like dimethylformamide (DMF) create sustainability challenges despite their chemical utility. Similarly, certain toxic reagents drive efficient processes while raising environmental concerns.
Stefan emphasizes the importance of addressing sustainability early in development. “Excellence in process chemistry and understanding scale-up challenges is a differentiator from a sustainability perspective,” he notes. “Ideally, this starts in preclinical and Phase 1, because fundamental changes to the process later will be difficult due to timeline pressure during development and regulatory hurdles after commercialization.”
AI Applications Beyond the Hype
Stefan takes a measured view of artificial intelligence in manufacturing, identifying where computational tools create genuine value versus areas where human expertise remains irreplaceable.
“The human will always have an important role and will not become obsolete,” he states. “When it comes to chemistry—designing the right route, choosing the right reagents and technologies—this will always require great science, great chemists, great engineers.”
Chemical reaction databases have aided chemists for decades. AI enhances these tools, but fundamental decision-making remains human-driven. “What will change is the speed and efficiency with which we’ll be able to understand reactions and processes,” Stefan explains.
Quality by Design and Process Optimization
Quality by design (QBD) provides the foundation for process development, with design of experiments (DOE), process analytical technology (PAT), and continuous reaction monitoring at its core. “These tools help you run as few experiments as possible, ideally data-rich experiments, then map and understand the design space quickly and build your silico models,” Stefan notes.
AI agents will accelerate this process. “We’re not too far away from AI virtually running or planning experiments autonomously,” he predicts. “AI will help build the digital twin out of silico models and real-world data, making scale-up more reliable.”
Once scale-up occurs, the iterative process between real-world production and models enables faster optimization toward the “golden batch.” This requires hardware investment in production facilities to record and utilize data effectively.
Eventually, AI could enable preventative maintenance. “Once you have defined your golden batch, you can use AI to find out quickly when your reaction isn’t running according to norm, indicating something must be wrong in your plant.”
Current Implementation Status
Siegfried has made significant progress in development. “We’re fairly advanced—very quick to build our models,” Stefan reports. “All our chemists and engineers are trained in digital tools and DOE.”
The production side requires more work. “Some work remains to be done, and this is pretty much across the industry, in being able to generate and utilize data in production to have that perfect iteration between the lab model and your scale-up model for continuous improvement.”
Industry’s Underestimated Challenge
When asked about underestimated scientific challenges, Stefan points to timeline compression.
“The pharmaceutical industry is under more financial pressure to reduce prices. Accelerated development timelines and extended patent lifetime once drugs reach market will be key.”
This challenge intensifies as AI revolutionizes drug discovery. “Where AI may have an even larger impact is in drug discovery,” Stefan observes. “AI can design molecules much faster and more reliably. You’ll see faster output, higher quality output, and more output in general of drug candidates.”
The manufacturing bottleneck looms. “You cannot build up manufacturing capacity as fast as the number of molecules increases coming out of drug discovery,” Stefan warns. “What will be key is making use of existing manufacturing capacity as much as possible.”
AI can help here too—through continuous improvement of existing processes, compressing cycle times, reducing setup and clean-down times, and optimizing product sequencing in multipurpose plants. “I’m convinced this capacity will be needed down the road,” Stefan concludes.
