Pharma services has been a consensus-bullish sector for the last three years, and much of that consensus is well-founded. Demand for sterile fill-finish, peptide synthesis, ADC manufacturing, and specialty CRO services is real and durable. Capital remains available. Multiples have held up better than most healthcare sub-sectors through the most recent rate cycle.
Several developments within that bullish picture, however, warrant closer attention, particularly as they begin to differentiate winners from average performers. None of these themes overturns the broader sector thesis. All are increasingly relevant for founders preparing for a transaction and sponsors building positions in the space, because they are beginning to show up materially in valuations and process outcomes.
Here are the three themes likely to matter most over the next 24 months.
Theme one: Specialization is outperforming integration
The integrated CRDMO model has been a popular exit narrative for five years. WuXi was the archetype. Thermo Fisher built scale across both clinical (PPD) and manufacturing (Patheon). Labcorp built Fortrea, and ICON acquired PRA Health Sciences to position for end-to-end clinical-to-commercial coverage.
What is emerging in 2025 and 2026 is a more nuanced picture, one in which specialization appears to be outperforming broad integration strategies. WuXi divested its CRO and SMO business to Hillhouse in October 2025 and is concentrating on the CRDMO core, effectively moving in the opposite direction of integration. Labcorp spun Fortrea, which has reported significant goodwill impairments through 2025. The Thermo PPD-Patheon integration has reportedly underperformed initial synergy expectations, and in practice, pharma sponsors continue to source CDMO and CRO services from separate vendors regardless of single-vendor availability. Meanwhile, Medpace, a pure-play biotech-focused mid-market CRO with no CDMO ambitions, has been one of the best-performing names in the sector, with revenue up roughly 24 percent in the third quarter of 2025 and 81 percent of revenue coming from emerging biopharma.
The pattern emerging is that specialization is increasingly being rewarded over integration. The likely explanation is structural. Pharma R&D leaders source services based on capability fit, not vendor convenience. A Phase 1 oncology program requires the strongest Phase 1 oncology CRO, not necessarily the same vendor managing the company’s commercial fill-finish operations. Buyer behavior continues to indicate that single-vendor convenience matters less than capability depth.
What this means in practice
For founders building specialty platforms, depth of expertise should increasingly be positioned as a competitive advantage rather than a gap in scope. Focused platforms in oncology CRO, rare disease clinical operations, peptide CDMO, ADC conjugation, sterile fill-finish, and behavioral health research have been outperforming integrated peers in recent processes on both growth and multiples.
For PE sponsors, the diligence implication is to carefully test platform-plus-bolt-on theses for the true source of incremental value. The bolt-ons most consistently adding value are those that deepen capability within a specialization rather than broaden scope across the value chain.
Theme two: CDMO capacity timing is becoming a critical valuation variable
PharmaSource tracked roughly $25 billion in announced CDMO capacity expansions globally in 2025, with the majority flowing into the U.S. Lonza, CordenPharma, Vetter, PCI, Catalent, Samsung Biologics, and a dozen mid-market players are all building. GLP-1 demand justifies a meaningful portion of this expansion, and there appears to be little scenario in which demand meaningfully softens before 2028.
The question is not current demand, but future capacity
The more important question is what happens when 2029 to 2032 capacity comes online. By that point, the first GLP-1 biosimilars will be approaching patent cliffs, with semaglutide composition-of-matter protection expected to expire in major markets around 2032. Pharma sponsors will likely have established second-source resilience, and the capacity announced in 2024 and 2025 will be fully operational.
The cell and gene therapy CDMO sector offers a useful reference point. Capacity was built ahead of demand on a similarly consensus-driven thesis between 2020 and 2022, and the sector spent much of 2024 and 2025 working through painful overcapacity, multiple compression, and impairments at Resilience, Catalent’s CGT business, and others. The takeaway is not that GLP-1 capacity faces the same outcome, but that consensus growth narratives can eventually create oversupply if expansion outpaces realized demand.
This does not necessarily point to a downturn. The picture for 2026 to 2028 remains strong. The more relevant question is how operators are positioning for the back half of the decade.
The strongest CDMOs are using the current scarcity premium to extend contracted demand further into the future while pharma sponsors still feel supply constrained. The specific contracting structures vary by customer base. Top-10 pharma companies working with the largest CDMOs (Samsung, Lonza, Catalent, Thermo Fisher) can credibly commit to long-dated take-or-pay supply agreements with capacity reservation deposits, and these structures are increasingly being disclosed in connection with large biologics and GLP-1 capacity buildouts.
Mid-market CDMOs serving emerging biotech and specialty pharma customers face a different reality, because biotech customers often cannot underwrite multi-year volume commitments when their own demand depends on clinical outcomes and funding. In that segment, the more relevant structures are multi-year minimum volume floors, exclusivity arrangements, capability reservation fees, and inflation escalators on extended terms. The form of protection varies. The strategic objective does not: convert spot-market exposure into contracted demand before negotiating leverage shifts.
What this means in practice
For PE sponsors, contracted backlog quality through 2028 and beyond is becoming a more important valuation lens for CDMO platforms than headline EBITDA. A CDMO with 70% of 2028 revenue under multi-year contracted arrangements underwrites very differently than a CDMO with 70% of 2026 revenue supported by one-year purchase orders, even if 2026 EBITDA appears identical.
The diligence question increasingly worth pressing on is straightforward: what does contracted backlog look like in 2028 through 2030, what structural protections exist, and how do those protections vary across top customers? The answer increasingly reveals more about durable value than current run-rate EBITDA alone.
Theme three: AI is changing the services revenue equation, and leading operators are evolving business models in response
AI integration in CDMOs and CROs is real and accelerating, and much of the industry conversation continues to focus on the productivity gains AI will deliver. A question deserving more attention, however, is who ultimately captures the value created by those productivity gains.
Why productivity gains do not automatically translate to profit
In services businesses, greater efficiency can unintentionally pressure revenue if pricing structures remain unchanged. Services businesses sell time, and AI compresses time. If site selection tools meaningfully reduce recruitment cycles, that translates to fewer billable site-management hours. If process analytical technology and digital twins compress process development timelines, that means fewer billable scientist hours. Unless the commercial model evolves alongside these efficiencies, the productivity gains accrue primarily to the pharma customer rather than the services vendor.
This dynamic is not unique to pharma services. It is how the productivity equation has historically played out across consulting, legal services, and IT services over the last two decades, and there is little reason to believe pharma services will be fundamentally different.
The commercial model must evolve alongside AI
The implication is not that AI is negative for services businesses. AI is increasingly necessary, and operators that fall behind on integration risk losing competitive position. The more important point is that AI integration alone is unlikely to protect revenue and margin. The commercial model must evolve alongside it.
The early indicators worth watching are movements away from straight time-and-materials and unit-rate billing toward structures that capture a portion of the productivity gain. These include milestone-based engagements, outcome-linked fees, risk-sharing arrangements, yield-share components in manufacturing contracts, and tiered pricing tied to delivery performance.
Adoption of these structures has been discussed for years and has remained relatively limited in practice, but the AI productivity question is likely to place renewed pressure on the conversation over the next 24 to 36 months.
This is where the diligence question increasingly appears to be heading, and one increasingly worth incorporating into founder positioning conversations: as AI delivers the productivity gains operators are investing to enable, where do the revenue dollars come from?
If the answer is the same hourly rates with fewer hours, the financial math becomes unfavorable. If the answer involves a credible evolution of the commercial model, the business is positioned to participate in the productivity gains rather than simply absorb them.
What this means in practice
For founders, the takeaway is to begin thinking about how the commercial model may evolve over the next three to five years, even if the immediate operational priority remains AI integration itself.
For PE sponsors, AI capability and AI commercial-model evolution should increasingly be treated as two separate diligence questions. The second is becoming just as important as the first.
What this means for the next 24 months…
For founders preparing for a process
These three themes increasingly shape what buyers and investors expect to see in a credible operational story. Lean into specialization rather than apologizing for the lack of integrated scope. Lock in long-dated contracted backlog while leverage still favors you. Build a credible business model evolution story around AI, not simply an AI marketing layer.
Three concrete actions over the next two quarters stand out:
- Tighten therapeutic-area or modality focus and exit segments where the business is subscale
- Convert the top five customer relationships from year-by-year purchase orders to multi-year contracted backlog with structural protections appropriate to the customer base
- Pilot at least one outcome-based or yield-share engagement to begin building a credible business model evolution narrative for a future process
For PE sponsors evaluating platforms
The diligence questions increasingly worth prioritizing over the next 24 months are equally clear.
- What percentage of 2028 revenue is contracted with multi-year structural protections, and what specific forms do those protections take?
- What is the gross-margin trajectory under realistic AI productivity scenarios?
- When a portfolio company has both CDMO and CRO capabilities, what is the actual customer sourcing pattern?
- Are customers using both, or sourcing one side of the relationship from a specialist competitor?
The pharma services sector remains a fundamentally attractive place to invest and operate. Demand is real, capital remains available, and the underlying drivers, including biologic complexity, biotech innovation, and modality diversification, appear durable through the next decade. The themes above do not change that picture. They sharpen it.
Operators and sponsors that engage directly with these developments in their 2026 strategy are likely to be better positioned for premium outcomes than those still relying on the consensus playbook alone.