Drug discovery statistics [2026]
Approvals, odds, time, cost and new kinds of medicine, with dates
Matic Broz, PhD11 min
Drug discovery turns a large, expensive pipeline into a small number of new medicines. The FDA's drug center approved 46 novel drugs in 2025 and 47 in 2026 through October 20, while only 6.7% to 13.8% of drug programs that enter Phase 1 trials go on to approval. A drug that succeeds spends a median of about 8 years in human testing and review, and published estimates of its cost run from $879 million to $2.67 billion.
Newer kinds of medicine are counted on separate lists. The FDA's September 2026 Purple Book listed 95 licensed biosimilars, and its list of cellular and gene therapy products had 54 entries. By our count 11 of those are AAV gene therapies, and one, Casgevy, is a CRISPR-edited therapy.
Drug discovery statistics at a glance
The headline figures below each come from a single dated source, so they can be quoted with their period attached. The sections that follow explain what each number counts and link to the guides that cover it in depth.
| Statistic | Value | Period or date |
|---|---|---|
| FDA novel drug approvals | 46 | 2025 |
| FDA novel drug approvals so far in 2026 | 47 | Through Oct 20, 2026 |
| Average FDA novel drug approvals per year | 46.6 | 2016 to 2025 |
| Chance that a program entering Phase 1 is approved | 6.7% to 13.8% | Records from 2000 to 2023 |
| Median time from first human study to FDA approval | 8.3 years | Drugs approved 2010 to 2020 |
| Cost per approved drug, including failures and capital | $879 million to $2.67 billion | 2018 and 2025 US dollars |
| Licensed US biosimilars | 95 | Sep 2026 |
| Biosimilars approved for use in Europe | 150 | Mar 2026 |
| Monoclonal antibody products still licensed by the FDA | 150 | Sep 2026 |
| FDA-licensed cellular and gene therapy products | 54 | Sep 17, 2026 |
| FDA-licensed AAV gene therapies | 11 | Sep 17, 2026 |
| FDA-approved CRISPR therapies | 1 | Sep 17, 2026 |
| AI-enabled drugs that had entered clinical trials | 117 | Jul 1, 2025 |
How many novel drugs does the FDA approve each year?
The FDA's Center for Drug Evaluation and Research (CDER) approved 46 novel drugs in 2025, made up of 34 new molecular entities and 12 new therapeutic biologics.[1] By October 20, 2026 it had approved 47, already more than in all of 2025.[2] Over the ten years from 2016 to 2025 the annual count ranged from 22 to 59 and averaged 46.6.[1]
A novel drug contains an active ingredient never before approved in the United States. The count leaves out generics, biosimilars, new uses of existing medicines, and products licensed by the FDA's biologics center, which reviews vaccines and cell and gene therapies.[1] That is why yearly novel approvals look tiny next to the total of FDA-approved drugs. What the faster pace of 2026 means for the industry is the subject of drug discovery trends.
What percentage of drug candidates reach FDA approval?
Large benchmarks estimate that 6.7% to 13.8% of drug programs entering Phase 1 reach approval. Citeline's 2024 analysis of phase transitions recorded from 2014 to 2023 gives the low end.[3] The BIO benchmark for 2011 to 2020 put the figure at 7.9%.[12] Wong, Siah and Lo reconstructed development paths for more than 21,000 compounds in trial records from 2000 to 2015 and arrived at 13.8%.[4]
The spread comes from the years studied, the databases used, and whether a success is counted per compound or per drug and disease pair. Every benchmark agrees on where most programs stop: only 28.0% of programs moved past Phase 2 in Citeline's data and 28.9% in BIO's, the first point at which a drug has to show it works in patients.[3][12] The clinical trial failure rate guide sets the estimates side by side and separates program failure from trials that stop early.
How long does it take to develop a drug?
A drug that reaches approval spends a median of 8.3 years between its first human study and FDA approval, according to Brown and colleagues' analysis of 405 innovative drugs approved from 2010 to 2020.[5] The BIO benchmark gives a longer average of 10.5 years from Phase 1 to approval, with Phase 2 the longest stage at 3.6 years.[12]
Neither clock includes target discovery, chemistry and animal studies, which come first and add several more years. Time also feeds directly into cost, because every extra year ties up capital. The phases are broken down by sponsor and disease in how long clinical trials take.
How much does it cost to develop a drug?
Estimates of the cost per approved drug range from about $879 million to $2.67 billion, because each method counts different things. Sertkaya and colleagues modeled spending from clinical trial data and estimated a mean of $879.3 million per approved drug in 2018 dollars, once failed candidates and an 11% cost of capital are included. The approved drug's own direct spending was only $172.7 million.[6]
Deloitte works from the other end, dividing the R&D budgets of 20 of the largest biopharma companies by the assets they launch. That figure was $2.67 billion in 2025, up from $2.23 billion in 2024.[7] A systematic review of 45 published estimates found capitalized values from $161 million to $4.54 billion, so no single figure deserves to be quoted alone.[13]
Most of that money pays for clinical trials and the candidates that fail in them, and the drug development cost guide reconciles the main estimates. Early screens such as molecular docking and ADMET prediction help teams drop weak molecules before expensive experiments, but they cannot remove the cost of failing in patients.
How many biosimilars are approved in the US and Europe?
The FDA's Purple Book listed 95 licensed biosimilars in its September 2026 data release, each a separate license under section 351(k) of the Public Health Service Act, with none revoked.[8] Europe has more. A March 2026 review of European Medicines Agency (EMA) records counted 150 biosimilars approved for use, out of 168 that EMA had recommended.[9]
A biosimilar is a copy of a biologic medicine, such as an antibody or insulin, that has been shown to be highly similar to the original. US approvals have accelerated sharply. By first approval date, we count 44 licenses from 2015 through 2023, then 21 in 2024, 21 in 2025 and 9 in 2026 through September 25.[8] Most copy antibodies: 67 of the 95 licenses reference a monoclonal antibody, led by copies of denosumab, adalimumab and ustekinumab, and 51 have also been designated interchangeable with their reference product.[8] The original antibodies are counted in how many monoclonal antibodies are FDA approved.
The FDA's own table of biosimilar products, current as of October 5, 2026, shows 89 entries because it puts some products with two licenses, such as Tyenne and Wezlana, on a single row. The same convention gives 18 biosimilar approvals in 2025 rather than 21.[14] Both figures are correct; they count different units.
Europe approved its first biosimilar in 2006, nine years before the United States. EMA runs a centralized scientific review, after which the European Commission issues the marketing authorization.[15] The 168 figure counts positive EMA recommendations, while the 150 figure removes 18 products withdrawn after approval. EMA recommended 41 new biosimilars in 2025 alone.[9][16]
How many cell and gene therapies are FDA approved?
The FDA's list of licensed cellular and gene therapy products had 54 entries at its September 17, 2026 update.[10] The FDA does not publish separate gene therapy and cell therapy subtotals on that page, and some products belong to both categories.
The list mixes gene-modified cell therapies such as CAR T-cell products, gene therapies given directly to patients, cord-blood products, and other cellular therapies. Calling all 54 gene therapies, or all 54 cell therapies, would overstate either category. These products are licensed by the FDA's biologics center, so none of them appears in the novel drug counts above.
This classification problem matters when comparing sources. A tally that leaves out cord-blood products, for example, reports a lower total from the same list on the same date.
How many AAV gene therapies are FDA approved?
We count 11 FDA-licensed AAV gene therapies as of September 2026. Adeno-associated virus (AAV) vectors are modified, non-infectious viruses that deliver a working gene to the patient's cells. Seven AAV products were approved through 2024, Itvisma followed in November 2025, and Otarmeni, Genglycos and Fayuvi were approved in April, August and September 2026.[10][17][18][19][20][21]
| Approval year | AAV gene therapy |
|---|---|
| 2017 | Luxturna |
| 2019 | Zolgensma |
| 2022 | Hemgenix |
| 2023 | Elevidys, Roctavian |
| 2024 | Beqvez, Kebilidi |
| 2025 | Itvisma |
| 2026 | Otarmeni, Genglycos, Fayuvi |
The two newest approvals both treat rare inherited metabolic diseases. Genglycos uses an AAV8 vector to deliver a working G6PC gene to the liver in glycogen storage disease type Ia, and Fayuvi uses AAV9 in Sanfilippo syndrome type A.[20][21]
This is a product count, not a count of distinct AAV capsids or diseases; Zolgensma and Itvisma deliver the same gene. It also counts licensed products even if a manufacturer later limits or stops commercial sales.
How many CRISPR therapies are approved or in clinical trials?
Casgevy is still the only FDA-approved CRISPR therapy, on the FDA's list as updated on September 17, 2026. The FDA approved it in December 2023 for sickle cell disease, as the first treatment to use CRISPR/Cas9 genome editing.[10][22]
Counts of CRISPR clinical trials depend on how the search is done. A 2025 review by Yang and colleagues reported 69 active CRISPR clinical trials and 6 completed studies.[23] A 2026 survey of ClinicalTrials.gov and the EU Clinical Trials Register by Şahin and colleagues counted 14 registered CRISPR-related trials completed since 2017, and found that only the Casgevy program had completed Phase 3.[24]
Broader gene-editing trackers report higher numbers because they include base and prime editing, more trial statuses, or preclinical programs. Any count should state its search date, registries, statuses and technologies.
How many AI-discovered drugs are in clinical trials?
No drug discovered or designed with AI had been approved by October 2026. Haque and colleagues counted 117 AI-enabled drugs from 63 companies that had entered clinical trials by July 1, 2025; by December 2025, 60 had completed Phase 1 and 8 had completed Phase 2.[11] Whether these molecules succeed more often than others, and what the counts include, is covered in the AI drug discovery statistics.
What the numbers say together
The figures describe one funnel seen from different points. Roughly 47 novel drugs a year come out of a process in which most programs fail in Phase 2, each success takes close to a decade in the clinic, and the cost of the failures is carried by the drugs that make it. Biosimilars, cell and gene therapies, and CRISPR medicines are growing quickly, but they are counted on separate FDA lists, so the most common error with drug discovery statistics is adding numbers that count different things.
Methodology
FDA lists were read on October 24, 2026. The biosimilar counts are distinct 351(k) license numbers in the Purple Book file purplebook-search-September-data-download.csv, dated by the earliest approval date of each license; antibody biosimilars are those whose proper name carries the -mab stem, and interchangeable ones carry an interchangeability approval date. The same file supports the analysis script behind our FDA-approved drug counts. The AAV count is our classification of products on the FDA cellular and gene therapy list, checked against each product's FDA announcement or vector description. The 46.6 average divides 466 CDER novel approvals by the ten years from 2016 to 2025.
ProteinIQ. "Drug discovery statistics [2026]." Updated October 24, 2026. Accessed [your access date]. https://proteiniq.io/guides/drug-discovery-statistics
