Stratenity POV, Pharmaceuticals & Life Sciences Industry Scan

The economics of failure at scale

Pharmaceutical R&D is a business defined by the mathematics of failure. Bringing a new molecule from discovery to approval takes 10 to 15 years and, on a capitalized basis including the cost of failed programs, an estimated 2 to 2.8 billion dollars. The clinical success rate from Phase 1 to approval sits around 10 to 12 percent across therapeutic areas, meaning the industry funds roughly nine failures for every drug that reaches a patient. The value of the entire enterprise therefore hinges less on the winners and more on how fast and how cheaply the failures are identified.

Layered on top of this is a revenue crisis of timing. An estimated 200 billion dollars or more in annual branded revenue is exposed to patent expiration and biosimilar or generic entry through the end of this decade, as blockbuster biologics and small molecules lose exclusivity. A company that does not refill the pipeline before its anchor product goes off patent faces a revenue cliff that no amount of commercial execution can offset.

Risk-adjusted portfolio value, not headline R&D spend

The naive view treats R&D as a cost line to be minimized. The correct view treats it as a portfolio of risk-adjusted options, where value is created by allocating capital toward programs with the highest risk-adjusted net present value and, crucially, killing weak programs before they consume Phase 3 budgets. A single Phase 3 trial can cost 100 to 300 million dollars, so a no-go decision made in Phase 2 rather than Phase 3 can preserve hundreds of millions in capital for redeployment.

The worked example: Phase 2 is where value is truly made or destroyed, because it is the last cheap point to kill a program before Phase 3's nine-figure commitment. A rigorous Phase 2 no-go on a program with a weak efficacy signal, made with conviction rather than optimism, can free 200 million dollars for a higher-probability asset. The discipline to stop is the most underrated source of portfolio value in the industry.

Bench scientists, clinical operations, and the translational gap

The talent equation spans deep scientific specialization and industrial-scale clinical operations, and the scarcest capability sits at the translational bridge between them.

• Translational scientists who connect biological mechanism to clinical endpoint are decisive, because a wrong translational bet is the most expensive error in the industry.
• Clinical operations and biostatistics talent run the trials that consume the largest budgets, and their design choices determine both cost and the probability of a clean readout.
• Regulatory affairs professionals who can navigate FDA and EMA in parallel are a strategic asset, since divergent filing strategies can add years.
• Computational biology and AI-in-drug-discovery talent is in intense demand, but its value is capped by whether the organization can operationalize model outputs into wet-lab and clinical decisions.

AI at the frontier, but validation is the gate

Drug discovery is where AI has produced its most tangible scientific results, from protein structure prediction compressing work that once took years into hours, to generative chemistry proposing novel candidate molecules, to machine-learning models flagging trial sites and patient cohorts likely to enrich a signal. These are not hypothetical. Several AI-originated or AI-optimized molecules have entered clinical trials, and the technology is genuinely reshaping the earliest, cheapest phases.

But the sector's discipline is unforgiving: an in silico prediction is a hypothesis, not evidence, and it must survive wet-lab validation and then human clinical data. The governance requirement is therefore severe. Every model-generated candidate, biomarker, or trial-design choice must carry a traceable evidence chain, because the FDA and EMA will ask which data supported the decision, and a black-box answer is not admissible. Data integrity under regulations like 21 CFR Part 11 makes provenance a legal obligation, not a best practice.

FDA, EMA, and the evidentiary standard that governs everything

No sector is more thoroughly governed. The FDA and EMA control every gate from investigational new drug application through marketing authorization, and the standards touch trial conduct, manufacturing, and post-market surveillance simultaneously.

• FDA IND and NDA/BLA pathways, and the EMA centralized procedure, define the evidence required to test in humans and to market.
• Good Clinical Practice and Good Manufacturing Practice govern trial conduct and production, with 21 CFR Part 11 governing electronic records and data integrity.
• Pharmacovigilance obligations require ongoing safety monitoring and adverse-event reporting for the life of the product.
• The Inflation Reduction Act empowers Medicare to negotiate prices on selected high-spend drugs, a structural change to the US revenue model with the first negotiated prices taking effect in 2026.

The IRA deserves special emphasis because it compresses the earn-back window. Small molecules become eligible for negotiation 9 years after approval and biologics after 13, which shortens the effective period of pricing freedom and raises the strategic premium on choosing the right indications and sequencing launches deliberately.

Efficacy, safety, and the widening evidence bar for access

The ultimate customer outcome in pharma is a therapy that is both efficacious and safe, proven to a standard no other industry matches. But the definition of a successful outcome is widening beyond the regulatory approval itself, because payers and health-technology-assessment bodies increasingly demand evidence of real-world value, not just trial efficacy.

This means the evidence strategy can no longer stop at approval. A drug that clears the FDA but cannot demonstrate cost-effectiveness to payers, or comparative benefit against standard of care, faces access restrictions that strand its commercial value. Real-world evidence, health-economics modeling, and outcomes data are now integral to the product, not a post-launch afterthought, and building that evidence chain from the trial design forward is what separates an approved drug from a reimbursed one.

Open innovation as the operating model

No single company owns the full value chain anymore, and the most productive pipelines are assembled from a network of specialized partners.

• Biotech in-licensing and acquisition is now a primary source of pipeline, with large pharma effectively outsourcing early discovery risk to venture-funded biotechs.
• Contract research and manufacturing organizations run a large share of trials and production, making partner governance central to quality and timeline control.
• Academic and translational partnerships supply the earliest mechanistic science and target validation.
• Payer, HTA, and real-world-data partnerships increasingly shape which indications and evidence packages are worth pursuing before a trial is even designed.

The portfolio and every gate decision as a governed artifact

The defining decisions in pharma are the go/no-go gates, and the industry's persistent failure is optimism bias, carrying weak programs into expensive stages because stopping feels like defeat. Stratenity's position is that the R&D portfolio and each stage-gate decision should be a typed, versioned decision artifact, with defined inputs, the full evidence chain, risk-adjusted valuation, and the human governance approval attached to every consequential call.

This reframes AI's role. AI can accelerate discovery and sharpen the probability estimates that feed a gate, but its highest value is enabling a faster, more honest, and fully traceable kill decision. Every recommendation carries its reasoning: which preclinical and clinical data, which model version, which assumptions, so that a no-go is defensible to the board and a go is defensible to the regulator. In a business where nine of ten candidates fail, the governed decision to stop early is the most valuable artifact the organization produces.

Where to focus the next twelve months

• Institutionalize evidence-based stage gates with pre-committed kill criteria so weak Phase 2 programs are stopped before they consume Phase 3 budgets.
• Build IRA-aware lifecycle strategy into launch planning, sequencing indications against the 9- and 13-year negotiation clocks rather than after them.
• Deploy AI in discovery and trial design where it compresses the cheapest phases, with mandatory wet-lab and clinical validation gates on every model output.
• Embed real-world-evidence and health-economics strategy from trial design forward so the drug is reimbursable, not merely approvable.
• Formalize partner and CRO governance so outsourced trials and manufacturing meet the same data-integrity and quality standard as internal work.

Sector-specific levers with a metric on each

• Enforce disciplined Phase 2 go/no-go gates, measured as R&D capital preserved by earlier kill decisions per program stopped.
• Apply AI-driven target and molecule generation in discovery, measured as reduction in discovery-to-candidate cycle time.
• Optimize trial site and patient selection with predictive enrichment, measured as trial recruitment time and screen-failure rate reduction.
• Sequence launches against IRA negotiation timelines, measured as protected-revenue years captured before price-negotiation eligibility.
• Build real-world-evidence packages from trial design, measured as payer access and reimbursement rate at launch versus the therapeutic-area baseline.

Related reading

Put this sector view to work with the cross-cutting Stratenity frameworks.