FDA approves 500 novel drugs in a year
Will FDA's Center for Drug Evaluation and Research approve 500 or more novel drugs in a single calendar year?
Chance by Q4 2031
<1%
Cumulative probability
- Latest forecast
| Year | Q1 | Q2 | Q3 | Q4 |
|---|---|---|---|---|
| 2026 | <1% | |||
| 2027 | <1% | <1% | <1% | <1% |
| 2028 | <1% | <1% | <1% | <1% |
| 2029 | <1% | <1% | <1% | <1% |
| 2030 | <1% | <1% | <1% | <1% |
| 2031 | <1% | <1% | <1% | <1% |
Model reasoning
FDA's drug center has never approved more than 59 new drugs in a year. It has approved 46 so far in 2026, so 500 would take about eight to nine times its best year ever. Drug pipelines, the years clinical trials take, and the agency's own staffing make that kind of jump nearly impossible by 2031. The main thing that would change this is a sweeping law, or a shift driven by artificial intelligence, letting regulators approve hundreds of drugs on far less human testing.
Where things stand
- 2026 so far: FDA's Novel Drug Approvals for 2026 table (content current as of 10/09/2026) has 46 entries, the latest being Hovilpri on 10/9/2026. That works out to about 60 for the full year (https://www.fda.gov/drugs/novel-drug-approvals-fda/novel-drug-approvals-2026, fetched 2026-10-13). Reaching 500 by 31 December would take about 454 more approvals in 11 weeks, so the first quarter-end is effectively zero.
- Recent history: I fetched FDA's year pages and counted table rows: 2025 = 46, 2024 = 50, 2023 = 55, 2022 = 37, 2021 = 50 (https://www.fda.gov/drugs/novel-drug-approvals-fda/novel-drug-approvals-2025 and the matching pages for the other years). The 2024 annual report's chart gives 2015–2024 as 45, 22, 46, 59, 48, 53, 50, 37, 55, 50, an average of about 47 a year (https://www.fda.gov/files/drugs/published/new-drug-therapy-2025-annual-report.pdf).
- Size of the jump: The record is 59 (2018). 500 is about 8.5 times that.
Base rate (computed)
I used annual counts for 2005–2025. The 2015–2025 figures come from FDA; the 2005–2014 figures are approximate values from memory.
- The standard deviation of year-to-year changes, measured as log ratios, is 0.33.
- The largest one-year rise was 2.09 times (22 to 46, 2016 to 2017). The largest absolute rise was 24.
- I ran a Monte Carlo simulation treating counts as a log random walk with trend drift, starting from about 57 in 2026. It gives about 0.55% that any year from 2027 to 2031 reaches 500.
- That overstates the tail, because the series bounces back toward its average after dips like 2016. A model where counts stay stable around the trend gives essentially zero.
- A related Metaculus question, Drugs approved by FDA in 2035, has a community estimate of about 69. I saw this only in a search snippet; the page itself was blocked.
Supply limits
Each approval needs a submitted application. FDA's paper on first-cycle rejection rates says 31 FY2025 applications, about 25% of original NDA/BLA submissions, were still pending. That implies roughly 120–125 original applications of all kinds in FY2025, and only about 50–70 a year are for new molecular entities or new biologics (https://www.fda.gov/media/194858/download). Getting 500 approvals in one year by 2031 would need roughly ten times more late-stage programs, files from sponsors, and FDA review capacity. Clinical trials, manufacturing and stability data, and companies' ability to launch products all build in multi-year delays.
Paths to YES (all remote)
- AI plus deregulation. Very capable AI arrives around 2028–29, and FDA and Congress allow approval on mechanism or biomarker evidence at scale. AI-discovered drugs are only now reaching Phase 3: Insilico started the first such Phase 3 in September 2026.
- Reciprocity. A law lets FDA mass-approve drugs already approved abroad. No such bill is moving, and in September 2026 FDA moved to tighten its scrutiny of foreign trial data.
- Personalized therapies. Individualized treatments under the "plausible mechanism" framework (draft guidance, February 2026) get counted one by one. This is a narrow, rare-disease approach.
- Changing the count. FDA reclassifies the series, for example by folding in generics. That is essentially excluded by the question's rules on the series and its successors.
Paths to NO
FDA lost staff to layoffs. The PDUFA VIII deal includes a fee cut and money to rehire reviewers. CDER and CBER only got permanent directors in September 2026. Operation TrialBlazer aims at Phase 1 timelines, not at multiplying approvals tenfold.
Strongest case against my estimate
This administration has repeatedly created new fast pathways: the plausible mechanism framework, priority vouchers with 1–2 month reviews, and a single-trial default. A year-to-date count reaching 500 in any single year is enough. If short AI timelines play out, a political push to approve far more drugs is conceivable. Even so, physical and commercial limits make 500 in one year by 2031 implausible, so I keep the tail at about half a percent rather than above 1%.
Forecast
Quarterly chances of first resolution start at 0.02% for the rest of 2026 and rise to about 0.04–0.045% per quarter in 2031. Cumulatively that gives 0.06% by end-2027, 0.15% by end-2028, 0.27% by end-2029, 0.41% by end-2030 and 0.57% by end-2031.
Resolution criteria
Question. By the last day of each quarter, has FDA's Center for Drug Evaluation and Research (CDER) publicly reported approving 500 or more novel drugs in a single calendar year?
Definitions.
- Novel drugs: drugs counted in CDER's official annual Novel Drug Approvals series, including new molecular entities approved under New Drug Applications and new therapeutic biological products approved under Biologics License Applications. FDA's classification in that series controls; the word “novel” in a press release alone does not establish inclusion.
- Count: the number of distinct novel-drug entries in that series, not the number of indications, doses, presentations, active ingredients, or approval actions. A combination or co-pack listed as one entry counts once. Both traditional and accelerated approvals count. Subsequent withdrawal does not erase an approval from the historical count unless FDA corrects the series itself.
- Single calendar year: January 1 through December 31 of one year, assigned using FDA's approval dates, not a fiscal year or rolling twelve-month period. The Observation period convention applies; 2026 is the first eligible calendar year, and its entire annual count is eligible.
Resolves YES if an eligible FDA publication reports an annual total of at least 500, or its official year-specific novel-drug table contains at least 500 distinct entries. A published year-to-date count reaching 500 also qualifies before year-end, provided its observation period ends after the opening date. A later FDA correction that first raises an eligible year's count to 500 qualifies on the correction's publication date. AI involvement in discovery, development, or review is not required.
Resolves NO if no qualifying publication establishes the threshold by the quarter-end. Do not combine years or add approvals by FDA's Center for Biologics Evaluation and Research. Generic approvals, biosimilars, supplemental indications, reformulations, and other approval actions count only if FDA includes the product as a distinct novel-drug entry in this series. Applications, trial authorizations, emergency-use authorizations, advisory-committee recommendations, and projected approval totals do not count.
Resolving source. FDA only: the year-specific Novel Drug Approvals tables, CDER's annual New Drug Therapy Approvals reports linked from those tables (see the 2025 table and report link), or an FDA publication explicitly reporting the same series. An explicitly corrected FDA figure supersedes an earlier figure when testing for the first qualifying publication; otherwise the annual report controls a conflict with the web table. A renamed successor series counts only if it preserves CDER's novel-drug category and reporting scope. This is a narrower resolving-source rule than Public confirmation (A2). Apply Resolution date (A3), the Resolution window, and the Observation period conventions. For a web-table update, use the first documented publication date of the qualifying version, not the approval date of its last entry. Once YES, always YES under the Resolution window convention.