What AI and Advanced Therapies Demand of Biopharma Leaders

The Rise of More Complex Science and Digital Tools

AI Is Easy to Pilot but Hard to Operationalize

Biopharma has always involved scientific and organizational complexity. What has changed is the form that complexity now takes. The scientific frontier increasingly includes therapies that are individualized, operationally demanding, and difficult to assess through traditional development methods. At the same time, digital technologies offer opportunities to improve discovery, development, and manufacturing, but their value depends on the quality of the underlying data and governance.

Artificial intelligence has moved from a research topic to an operating question. Biopharma companies nevertheless continue to face fragmented data, inconsistent terminology, and weaknesses in model governance. These conditions make enterprise use more difficult than small-scale pilots and complicate the validation of models used in regulated settings.¹˒²

A model that summarizes research or improves clinical-site selection in one department is not yet an organizational capability. Broader use requires common data definitions, controlled procedures for changing models, clear responsibility for results, and documentation that explains how decisions were made. Many organizations have not yet established all of these foundations.

This helps explain why AI initiatives often remain in the pilot stage. A pilot may demonstrate a useful technical result without changing routine work. Unless organizations redesign roles, incentives, review procedures, and accountability, automation may simply be added to existing practices rather than improve them. The result can be a visible demonstration with limited economic or operational impact.

Speed must also be distinguished from evidence. In drug discovery, faster hypothesis generation is valuable only when scientists can determine why a recommendation was produced and whether its source data were complete, current, and free of important bias. Manufacturing and quality applications require still greater control. A prediction that cannot be adequately explained during validation, a deviation review, or an inspection may weaken rather than strengthen confidence. The central governance issue is therefore whether an organization can defend AI-supported decisions when the consequences are clinical, regulatory, or financial.

Advanced Therapies Do Not Fit the Old Industrial Model

Advanced therapies have become a substantial part of the development pipeline. A 2026 review of cell, gene, and RNA therapy reported more than 3,200 active clinical trials in these fields worldwide by late 2025.³

Cell, gene, and precision medicines challenge familiar assumptions about development, manufacturing, and payment. Production batches may be small or specific to an individual patient. Clinical populations may be rare, making robust comparisons difficult. Long-term benefit may be scientifically plausible but uncertain when a therapy first reaches the market. These features complicate both manufacturing economics and payer assessment.

Traditional pharmaceutical operations were designed for large-scale, repeatable production and evidence from sizeable patient populations. Advanced therapies often involve small populations, individualized handling, and little tolerance for operational error. Leadership teams therefore cannot manage these products as if they were conventional additions to an established portfolio. They may require different expertise, supply systems, evidence plans, and financial expectations.

Payment presents an additional challenge. Health systems and insurers may face a high initial cost for a therapy whose benefits are expected to develop over many years. This has encouraged interest in installment payments, outcomes-based agreements, and other financing approaches. These arrangements also require long-term evidence about treatment durability and patient outcomes. A scientifically important therapy may consequently experience a slow launch if plans for follow-up evidence are not developed early.

Clinical Evidence Is Moving Closer to Care Delivery

Clinical development is becoming more closely connected with routine healthcare delivery. EY noted that more than 550,000 active clinical studies were registered across 228 countries as of mid-2025, while advanced therapies, decentralized trial designs, and expanding regulatory requirements placed additional pressure on trial supply chains.⁴

Decentralized trials, remote monitoring, and community-based participation can improve access and reduce burdens on patients. They can also introduce new variation into trial operations. Local partners, digital platforms, and remote data collection create additional requirements for validation, training, and oversight. A study may therefore become more convenient for participants while also becoming more difficult to standardize.

Representative participation remains an important concern. When recruitment depends too heavily on a narrow group of locations or familiar academic centers, the resulting evidence may be less applicable to the wider patient population. Real-world evidence may help, but only when electronic health records and longitudinal information can be linked, standardized, and governed well enough to support credible analysis. Differences in data ownership, format, and incentives remain significant barriers.

This is not solely a technical problem. It requires coordination among healthcare providers, research sites, sponsors, regulators, and technology companies. Each transfer of information can weaken consistency, traceability, or patient follow-up. As evidence generation becomes more integrated with clinical care, success will depend increasingly on whether healthcare organizations can participate as active research partners rather than serve only as sources of data.

The Frontier Is Now Scientific and Infrastructural

The future of biopharmaceutical innovation will not be determined by discovery alone. It will also depend on whether organizations can build the infrastructure needed to use new science at scale. This infrastructure includes manufacturing systems capable of handling individualized therapies, evidence methods suitable for small patient populations, and digital governance that makes AI-supported decisions understandable and defensible.

Scientific complexity can create both advantage and cost. It may distinguish a product and support strong clinical value. It can also increase the likelihood of manufacturing failure, payer uncertainty, and regulatory scrutiny. The organizations best positioned to lead will not simply pursue the most advanced science. They will develop the practical systems needed to produce, evaluate, finance, and deliver it without losing the characteristics that make it valuable.

Key Issues at a Glance

AI Promises Scale, but Data and Governance Lag

The productivity prize is substantial, yet uncontrolled automation can magnify compliance, bias and trust risk.

• Fragmented Enterprise Data: Inconsistent data and ontologies limit model reliability across discovery, clinical and manufacturing workflows.
• Validation and Control Risk: Generated outputs and model changes can introduce unsupported claims or untraceable decisions.
• Value Realization: Pilots proliferate without redesigned workflows, adoption measures or accountable business ownership.

Advanced Therapies Strain Conventional Operating Models

Cell, gene and precision medicines challenge traditional development, manufacturing, evidence and payment systems.

• Variable, Patient-specific Production: Small or individualized batches undermine economies of scale and complicate consistent quality.
• Ultra-small Evidence Populations: Rare populations strain recruitment, statistical power and conventional endpoint strategies.
• High Upfront Costs and Value Risks: Payers face large one-time costs before long-term benefit is fully known.

Clinical Trials and Real-World Evidence Move Closer to Care

Representative recruitment, remote data and real-world evidence increasingly depend on provider and community workflows.

• Representative Recruitment: Limited geographic and genetic diversity weakens relevance, fairness and recruitment credibility.
• Decentralized Trial Reliability: Remote data, local partners and digital health tools add operational and validation complexity.
• Interoperable Real-World Data: EHR and longitudinal data remain fragmented, rights-constrained and difficult to standardize.

Frequently Asked Questions

References

1. “Digital Transformation in Pharmaceuticals: The Impact of AI on Supply Chain Management.” Exploratory Research and Hypothesis in Medicine.
2. EY. “How Digital, AI-Enabled Supply Chains Can Strengthen Clinical Trials.”
3. Auclair, J. R., Joung, J., Singh, M. A., et al. “Artificial Intelligence and the Transformation of Cell and Gene Therapy Development.” Pharmaceutics, 2026.
4. EY. “How Digital, AI-Enabled Supply Chains Can Strengthen Clinical Trials.”