The state of the pipeline
WHO's pipeline reviews count on the order of a hundred antibacterial agents in clinical development worldwide — but only a minority target the critical-priority Gram-negative pathogens, and only around a dozen are judged genuinely innovative (a new class, target or mechanism). For comparison, oncology has thousands of candidates. Still, after decades of drought, real breakthroughs are reaching patients:
- Cefiderocol — a “Trojan horse” antibiotic approved in 2019/2020 that hijacks bacteria's iron-uptake machinery to smuggle itself through the Gram-negative outer membrane.
- Sulbactam–durlobactam (approved 2023) — targets carbapenem-resistant Acinetobacter baumannii, a WHO critical pathogen.
- Gepotidacin (approved 2025) — the first genuinely new-class oral antibiotic for urinary tract infections in decades, also being developed against drug-resistant gonorrhoea.
- Zosurabalpin and darobactin (in trials) — candidates from entirely new classes that attack previously untouched machinery in Gram-negative bacteria.
- New TB regimens — the BPaL/BPaLM combinations have cut treatment of highly drug-resistant tuberculosis from ~18 months of gruelling therapy to 6 months of pills, with cure rates around 90% in trials.
Phage therapy: setting viruses on bacteria
Bacteriophages — viruses that infect only bacteria — were used therapeutically a century ago, before antibiotics eclipsed them. They are precise (often targeting a single bacterial strain), self-amplifying at the infection site, and can kill bacteria that resist every antibiotic. Long confined to compassionate-use rescues, phage therapy is now generating formal evidence:
- A 2024 review of 100 consecutive compassionate-use cases reported clinical improvement in 77% of difficult-to-treat infections and bacterial eradication in 61%.
- Randomised trials are underway; in 2025 a Staphylococcus aureus phage product (AP-SA02) reported positive results in a phase 1b/2a bacteraemia trial, and engineered phage cocktails against E. coli urinary infections are in clinical development.
- Challenges remain: phages must often be matched to a patient's specific strain, manufacturing and regulation are unsettled, and bacteria can evolve phage resistance too — though often at the cost of losing virulence or antibiotic resistance.
AI joins the hunt
Machine learning is transforming the slowest step of antibiotic discovery — finding new chemical starting points:
- Deep-learning screens identified halicin (2020) and abaucin (2023, active against Acinetobacter), molecules with mechanisms unlike existing classes.
- Generative AI models have since designed novel candidate compounds from scratch against gonorrhoea and MRSA, and AI mining of genomes and ancient proteins is surfacing thousands of candidate antimicrobial peptides.
- AI also accelerates phage matching, resistance prediction and diagnostic interpretation — compressing processes that took weeks into hours.
Beyond new antibiotics
| Approach | Idea | Status |
|---|---|---|
| Vaccines | Prevent infections so antibiotics are never needed; pneumococcal, Hib and typhoid conjugate vaccines already avert huge volumes of antibiotic use. | Deployed; candidates against TB, Group B strep, gonorrhoea and hospital superbugs in trials. |
| Rapid diagnostics | Distinguish viral from bacterial illness, and identify the bug and its resistances in hours instead of days, so the right narrow drug is used first. | Molecular panels and MALDI-TOF in richer hospitals; cheap point-of-care tests are the frontier. |
| Monoclonal antibodies | Lab-made antibodies that neutralise specific pathogens or their toxins. | Approved for anthrax, C. difficile toxin and RSV; superbug candidates in trials. |
| Anti-virulence & potentiator drugs | Disarm bacteria (block toxins, quorum sensing, biofilms) or break resistance (beta-lactamase inhibitors) rather than kill outright — lowering the evolutionary pressure to resist. | Beta-lactamase inhibitor combos in wide use; others experimental. |
| Microbiome therapies | Restore healthy gut flora to crowd out resistant organisms; faecal-microbiota products for recurrent C. difficile were approved in 2022–23. | Approved for C. difficile; decolonisation uses under study. |
| Phage enzymes (lysins) & antimicrobial peptides | Use the bacteria-dissolving proteins of phages, or evolution's own antimicrobial molecules, as drugs. | Clinical trials. |
The missing ingredient: a market that works
Science is no longer the main bottleneck — economics is. Several recent antibiotics reached approval only for their makers to go bankrupt, because a drug society wants used sparingly cannot survive on sales volume. That is why “pull” incentives (subscription payments, market-entry rewards such as the proposed PASTEUR Act, and the AMR Action Fund's late-stage rescues) described on the Global Response page are considered as important as any molecule in the pipeline.
Bottom line
For the first time in a generation, genuinely new weapons — new drug classes, phages, AI-designed molecules, vaccines and fast diagnostics — are arriving together. Whether they reach patients at scale depends less on laboratories than on how the world chooses to pay for them.
Sources for this page
- WHO, “Antibacterial agents in clinical and preclinical development” (annual pipeline reports) — who.int
- “Antibiotics re-booted — time to kick back against drug resistance”, npj Antimicrobials and Resistance, 2025 — nature.com/articles/s44259-025-00096-1
- Labiotech, “Can phage therapy answer the booming antibiotic resistance problem?”, 2025 — labiotech.eu
- World Economic Forum, “AI is reviving an old solution against antimicrobial resistance”, 2025 — weforum.org
- “The global economic burden of antibiotic-resistant infections and the potential impact of bacterial vaccines: a modelling study”, 2025 — pmc.ncbi.nlm.nih.gov/articles/PMC12182023/