Why Have We Stopped Finding New Antibiotics?

The method Selman Waksman pioneered in the late 1930s was simple: grow soil microbes, see which ones killed bacteria, purify the active compound. It produced streptomycin, tetracyclines, erythromycin and many more, and almost two-thirds of all antibiotic classes came from that era. But each new screen found mostly the same molecules again, a problem chemists call rediscovery. By the 1980s the method was exhausted.

The second barrier is biological. Gram-negative bacteria wrap themselves in two membranes and actively pump foreign molecules back out. A compound that kills bacteria in a dish often cannot reach a high enough concentration inside these cells. That is why the WHO’s list of critical threats is dominated by Gram-negatives, and why activity against resistant Gram-negatives is a separate, essential measure of pipeline quality.

The arithmetic is stark. About $1.3 billion to reach approval, counting the candidates that fail along the way, followed by roughly $350 million over ten years to manufacture, monitor and support the drug once it is on the market. Against that, new antibiotics have averaged $46 million a year in sales. Achaogen won approval for plazomicin in June 2018, earned about $800,000 from it by the end of that year, and filed for bankruptcy in April 2019. One analysis estimated that a drug for hospital-acquired pneumonia could be worth over $12 billion to society and negative $4 million to the company that made it.

The value of a reserve medicine is the option to use it

An antibiotic can be enormously useful while treating relatively few patients. Its availability makes a dangerous operation, transplant or course of chemotherapy safer because clinicians have a rescue option if a resistant infection develops. Much of that value accrues to hospitals and patients who never need the drug. A per-prescription payment collects only a small part of it. Stewardship is therefore one reason the market fails, alongside short treatment courses, cheap older alternatives and the small patient populations for some resistant infections. PASTEUR sponsors’ June 2026 statement

Discovery remains difficult even if payment is repaired. Gram-negative bacteria have an outer membrane that excludes many molecules, pumps that remove others, and enzymes that disable drugs that get through. Zosurabalpin attacks the machinery that moves lipopolysaccharide into that membrane, a different route from simply finding another conventional antibiotic. Its narrow focus on Acinetobacter also makes rapid organism identification more important. A targeted drug, an effective diagnostic and reliable access need to arrive together. Zosurabalpin discovery paper, Nature, 2024

The reserve-value explanation is economic interpretation. A new mechanism does not guarantee clinical benefit or prevent bacteria from evolving resistance.

Paying for readiness

The fix most economists favour is called a pull incentive: a payment that rewards a useful antibiotic for existing, decoupled from how much is sold. England pioneered it with a subscription model and in 2024 expanded it to £100 million a year, with contracts worth up to £20 million a year for the most important new drugs. In the United States, the PASTEUR Act, reintroduced in Congress for the fourth time in 2026, proposes contracts of $75–300 million a year, with payments reduced by net product revenue. The model above shows why size matters: at a typical development cost, the top payment band can still fail to cover the illustrative discounted investment, especially because the 2026 proposal offsets payments by product revenue. Large pull payments combined with cheaper, faster discovery are what make the investment clearly worthwhile.

Count useful innovation, and pay for dependable access

WHO’s 2025 pipeline is a snapshot with a 15 February cutoff: 90 clinical products, including 50 conventional antibiotics and 40 non-traditional approaches. Its short release reports 15 innovative candidates and a five-candidate critical-threat headline. The detailed dashboard lists 21 of 27 priority-pathogen antibiotics with expected activity against at least one critical pathogen. The summary does not clearly reconcile these counts, so the chart uses the explicitly defined dashboard category; five must not be read as the total number with any critical-pathogen activity. Gepotidacin reached FDA approval in March 2025. Zosurabalpin remains an investigational programme, with Roche reporting phase III development, rather than an available treatment. WHO’s detailed pipeline dashboard; FDA’s gepotidacin trial snapshot; Roche’s 2025 annual report

England’s subscription guidance pays according to assessed clinical value, with a top band of £20 million per year. Delinking revenue from volume can reward supply, surveillance and readiness without encouraging unnecessary prescribing. The United States’ June 2026 PASTEUR reintroduction proposes annual contracts with product-revenue offsets; it is not revenue already flowing to developers. The calculator’s $300 million case is a policy-scale scenario: its near-break-even result depends on when development spending occurs, the discount rate, upkeep and the payment period. It cannot establish the correct contract price for a particular drug. NHS subscription commercial guidance

Contracts also need enforceable supply and access terms. A capitalized development-cost estimate cannot be treated as an undiscounted cash budget without changing its meaning.

Who is building what

Drug developers, discovery labs and the funds and governments trying to repair the market. Search the record, or filter by role.

A compound that clears infection in mice is a discovery, not a medicine. Most candidates fail in human trials, and roughly one in five infectious-disease drugs entering phase 1 is eventually approved.

Four numbers to watch

First, the number of new antibiotic classes in phase 3 or approved, especially against Gram-negative bacteria. Second, the total value of pull incentives committed worldwide per year. Third, the size of the WHO clinical pipeline and its share of innovative agents, updated every two years. Fourth, deaths attributable to resistance, the outcome that matters.