S02.3 · Healthcare & Life Sciences
A $170-200B outsourced-manufacturing layer growing on biologics and cell-and-gene outsourcing, with durable take-or-pay revenue.
Contract development and manufacturing organizations — the outsourced production layer for biopharma — are a $170-200B market in 2025 (Grand View Research / Fortune Business Insights; small molecule, biologics and fill-finish combined), growing 8-10% forward as big pharma goes asset-light and outsources the biologics, ADC and cell-and-gene therapy production it does not want to build in-house. What makes the segment unusual for a services business is the quality of the revenue: multi-year take-or-pay contracts, plus FDA/EMA facility approval that makes a sponsor expensive to move once a site is qualified. The switching cost is regulatory, and regulatory switching costs are the durable kind.
The historical run rate was 7-9%; forward projections sit at 8-10%. Revenue concentrates in the US and EU while capacity grows fastest in China and India, with Western reshoring pressure building in response to prior supply-chain disruptions. Concentration is moderate: an estimated 25-30% combined share sits with the largest players — Lonza, Catalent/Novo, Samsung Biologics, WuXi and Thermo Fisher's PPD unit — over a long tail of niche, modality-specific CDMOs.
Upstream sit API and raw-material suppliers; downstream, biopharma sponsors and regulatory bodies. This is a capital-intensive business — bioreactor and cleanroom build-out — running 20-30% EBITDA margins. The take-or-pay contract base delivers recurring revenue few services models can claim, and FDA/EMA facility qualification means a sponsor that has validated a site with a given CDMO faces high switching costs to leave it.
Agentic AI is working through batch-record review, deviation investigation and tech-transfer documentation — the same quality and regulatory paperwork that dominates cost across pharma manufacturing generally. Here the compression lands on QA and regulatory cost inside a fee structure sponsors are already negotiating down, so part of the saving accrues to the customer rather than the operator. The deeper effect is on the tech-transfer moat. Process expertise — the accumulated know-how of moving a sponsor's process from development to commercial scale — has separated CDMOs from one another, and AI is starting to standardize tech transfer across sites, narrowing that gap.
Expansion runs on two tracks that should not be read the same way. Modality expansion — biologics manufacturers building cell-and-gene therapy capability — chases a genuinely faster-growing outsourcing category and adds new revenue. Geographic reshoring capacity answers tariff and geopolitical risk; it protects the book rather than growing it. Clinical-trial supply and logistics is a further adjacency some CDMOs are building into. The AI horizon runs longer than in pure administrative functions — 2-5 years — because validated-process regulatory approval cycles gate any process change, and those cycles do not move faster because the underlying software does.