Skip to main content

Pipeline & R&D

Updated Jun 24, 2026 at 8:22pm

Research Draft High 1,141 words

A pharmaceutical company's pipeline is its portfolio of drug candidates in development, organized by clinical phase, and R&D is the spending and scientific process that moves them toward market. For pharma, the pipeline is the future earnings stream: today's revenue comes from drugs whose patents are eroding, and the central question for any investor is whether what's in the lab can replace what's coming off-patent. The core tension is that drug development is enormously expensive, slow, and mostly fails — yet the rare success is so valuable that the whole industry economics rest on it. Analyzing pharma means pricing low-probability, high-payoff bets under regulatory uncertainty.

How a pipeline is structured

Candidates progress through a regulated sequence, each stage a kill-or-continue gate:

  • Discovery / preclinical — target identification, lead optimization, animal toxicology. The largest funnel; most candidates die here.
  • Phase 1 — first-in-human, typically dozens of healthy volunteers (or patients in oncology). Goal: safety, dosing, pharmacokinetics.
  • Phase 2 — hundreds of patients with the target disease. Goal: efficacy signal and dose-finding. This is where most programs die.
  • Phase 3 — hundreds to thousands of patients, randomized and controlled. The pivotal, most expensive stage; generates the data that supports approval.
  • Regulatory submission — NDA (small molecules) or BLA (biologics) to the FDA; MAA to the EMA.
  • Approval / Phase 4 — post-marketing surveillance and label-expansion studies.

The total process commonly runs 10-plus years from first-in-human to approval (Tufts CSDD).

How it's used in practice

Reading the pipeline table. Every pharma/biotech publishes a pipeline chart: each row a candidate, columns showing phase, indication, and mechanism. Analysts assess (1) stage — later is de-risked and more valuable; (2) catalyst calendar — upcoming Phase 2/3 readouts, FDA decision dates (PDUFA dates), advisory-committee meetings, all of which move stock prices sharply; (3) diversification — is value concentrated in one molecule or spread across many shots on goal; and (4) mechanism novelty — first-in-class (higher risk, higher reward) vs. me-too.

Valuation: risk-adjusted NPV (rNPV). The standard tool for development-stage assets. Each candidate's projected lifetime cash flows (driven by peak sales, market penetration, pricing, and patent life) are multiplied by the cumulative probability of success from its current phase, then discounted to present value (commonly 10-15%, per multiple valuation guides). Company value = sum of product rNPVs + net cash − corporate costs. rNPV systematically yields lower values than naïve NPV because it bakes in attrition — which is the entire point.

The patent cliff frames everything. A drug earns monopoly economics only during its exclusivity window. At loss of exclusivity (LOE), small-molecule blockbusters typically lose the large majority of revenue within roughly a year of multi-source generic entry (industry estimates cite ~80-90% price/volume erosion); biologics erode more slowly because biosimilar substitution is harder. Industry analyses estimate $200-400 billion in branded revenue faces LOE between 2025 and 2030. Pipeline strength is judged against this hole.

Adoption, debate & evidence

Success rates are low and well-measured. The most-cited benchmark is the BIO / Informa Pharma Intelligence / QLS Advisors study Clinical Development Success Rates 2011–2020, analyzing 12,728 phase transitions across 9,704 programs. Its headline: the overall likelihood of approval (LOA) from Phase 1 was about 7.9%, and Phase 2 is the single biggest hurdle (its phase-2-to-3 transition is the lowest of any stage — roughly 28–30%, in line with the ~30.7% the prior 2006–2015 BIO dataset reported). Earlier 2006–2015 BIO data put overall LOA near 9.6%, and various estimates land in a roughly 5-12% range depending on period, dataset, and whether oncology dominates the sample — recent analyses suggest the rate has drifted toward the lower end. The precise number is dataset-dependent; treat any single figure as an estimate, not a constant.

Therapeutic area matters enormously. Oncology has historically shown the lowest LOA (commonly cited around 3-6% from Phase 1), while vaccines and some rare-disease/hematology programs run far higher. Programs that use a patient-selection biomarker (e.g. a genetic test that enriches for likely responders) have shown materially higher success in the BIO data — a finding widely cited to explain the shift toward precision medicine.

Cost estimates are contested. The Tufts Center for the Study of Drug Development (DiMasi et al., 2014–2016) estimated a capitalized cost of ~$2.6 billion per approved drug (~$2.87 billion including post-approval). That figure is industry-foundational but criticized: it is heavily driven by the cost of failures allocated to each success and by an imputed cost-of-capital "time cost" (~half the total) rather than cash out the door. Lower independent estimates exist (e.g. studies finding median costs well under $1 billion for some cancer drugs). The honest read: out-of-pocket cost per attempted program is far lower; the multi-billion figure reflects portfolio attrition and capitalization, and is genuinely disputed.

Strengths & limitations

Pipeline analysis works because the development process is structured, disclosed, and statistically characterized — there are real base rates to anchor probabilities, and catalysts are scheduled. It fails in predictable ways:

  • Binary risk is brutal and idiosyncratic. A single Phase 3 failure can erase a small-cap's value overnight; base rates don't tell you about this molecule.
  • rNPV is precise-looking but assumption-fragile. Outputs swing wildly on peak-sales and probability inputs; it offers false comfort. Garbage in, garbage out.
  • The #1 misuse: applying generic phase-success probabilities without adjusting for therapeutic area, mechanism risk, trial design, and whether a biomarker is used — the differences are larger than the average.
  • Disclosure is promotional. Companies frame pipelines optimistically; failed programs quietly disappear from the chart.

Sources

Disputes flagged: the Tufts $2.6B figure is contested (failure-allocation + capitalized time cost); overall LOA varies 5-12% across datasets and is not a fixed constant.