> ## Content Index > Fetch the complete content index at: https://nexi.fund/llms.txt > Use this file to discover other available public pages before exploring further. # AI-Designed Phage Therapy Targets Drug-Resistant Pneumonia in Landmark NIH Trial - URL: https://nexi.fund/ai-phage-therapy-pneumonia-2026/ - Published: 2026-07-12T15:30:42.000Z - Updated: 2026-07-12T15:30:42.000Z - Description: NIH awards Locus Biosciences $3.3M for Phase 1b trial of AI-designed bacteriophage therapy against antibiotic-resistant Pseudomonas aeruginosa hospital-acquired pneumonia. - Author: Nexi.fund Labs - Tags: Biotech & Health, #mode-3, #hook-number, #track-F **$3.3 million.** That's what the NIH just committed to a clinical trial for a therapy that doesn't use a single molecule a chemist designed. It uses AI-designed viruses (bacteriophages) reprogrammed to hunt and kill one of the deadliest hospital-acquired infections. 🎯 **What happened** The National Institute of Allergy and Infectious Diseases (NIAID) awarded Locus Biosciences $3.3 million (with up to $28 million available at milestones) to run a Phase 1b trial of LBP-PA01, an AI-designed bacteriophage therapeutic targeting antibiotic-resistant *Pseudomonas aeruginosa* pneumonia. **Why it matters** Antibiotic-resistant infections killed 1.14 million people in 2021\. *P. aeruginosa* alone accounts for up to 24% of ICU respiratory infections, and once it turns resistant, mortality reaches 50%. The last new class of antibiotics against Gram-negative bacteria was approved in the 1960s. No major pharma company has launched a novel Gram-negative antibiotic since. **The convergence** This is an AI plus synthetic biology and robotics story. A platform that compresses a decade of trial-and-error discovery into weeks of systematic computation. Antimicrobial resistance (AMR) is the quiet compounding crisis that the Covid-19 pandemic momentarily overshadowed. Four million deaths annually are already linked to resistant infections. The WHO calls it one of the top ten global health threats. And the pipeline for new antibiotics has been drying up for two decades. Big pharma exited the space because a course of antibiotics sells for a few hundred dollars, not the six-figure price tags oncology drugs command. Bacteriophages (viruses that evolved specifically to infect and kill bacteria) have been known for over a century. They were used as medicine in the Soviet republics and Eastern Europe while the West bet on penicillin. But phage therapy never scaled, for a simple reason: phages are exquisitely specific. A phage that kills one strain of E. coli may do nothing to the next. Finding the right phage for a patient's infection was a months-long manual process that could not be industrialised. That fundamental limitation is what has now changed. ## The Platform: From Years to Weeks Locus Biosciences, founded in 2015 as an NCSU spin-out in North Carolina's Research Triangle Park, spent its first years building something the phage therapy field had never seen before: an industrial pipeline. The core of the platform is a high-throughput screening loop driven by robotics and machine learning. Robots measure millions of phage-bacteria interactions in parallel, generating training data for multimodal large language models trained on genomic and functional datasets. Those models predict optimal genomic engineering configurations: which CRISPR-Cas3 payload to insert, which tail fibre to swap, which combination of phages in a cocktail maximizes killing while minimising resistance emergence. Locus's platform simulates over a quadrillion (one million billion) potential phage combinations in silico before any wet-lab experiment ever runs. What used to require years of graduate-student bench work now takes weeks of automated, AI-orchestrated testing and refinement. The "crPhage" technology that emerged from this pipeline adds CRISPR-Cas3 to the natural lytic activity of the phage. Cas3 is a DNA-shredding enzyme. Unlike the better-known Cas9 (which makes a single cut), Cas3 chews through the bacterial genome in a long, irreversible sweep. The phage delivers the construct, the natural lytic cycle begins, and if any bacteria survive that, the CRISPR-Cas3 payload finishes the job. It is a dual-mechanism kill switch with no known bacterial escape route. 🔬 **How crPhage works** 1\. The engineered phage binds to the target bacterium and injects its DNA 2\. The phage's natural lytic machinery begins replicating inside the host 3\. CRISPR-Cas3 expression triggers — a DNA-shredding enzyme that irreversibly degrades the bacterial genome 4\. The bacterium lyses, releasing more phage particles to infect neighbouring cells It already demonstrated this works in humans. The company's first candidate, LBP-EC01 targeting E. coli urinary tract infections, completed a Phase 1b trial, the world's first randomised, placebo-controlled trial of a recombinant bacteriophage therapy, meeting all primary and secondary endpoints. Phase 2 data published in The Lancet Infectious Diseases showed the platform's dual mechanism translated into real patient outcomes, including reduced bacterial load and symptom clearance. BARDA committed up to $93 million to take that LBP-EC01 program through Phase 3 and toward potential FDA approval. ## The Target: Pseudomonas aeruginosa in the ICU LBP-PA01, the candidate funded by the new NIH award, targets a harder problem than its first candidate. P. aeruginosa is a Gram-negative bacterium that proliferates on hospital surfaces, ventilator tubing, and the lungs of immunocompromised patients. It is the leading cause of ventilator-associated pneumonia (VAP) in intensive care units worldwide. Up to 24% of ICU patients on mechanical ventilation develop a respiratory infection; when the pathogen is resistant P. aeruginosa, mortality climbs to 50%. The US Centers for Disease Control classifies carbapenem-resistant P. aeruginosa as a "serious threat," the second-highest tier in its AMR classification. Existing treatment options are limited to a handful of older antibiotics with significant toxicity and growing resistance. Colistin, the drug of last resort, causes kidney failure in a substantial fraction of patients. The Phase 1b trial funded under contract No. 75N93025C00035 will evaluate safety, efficacy, and optimal dosing in patients with P. aeruginosa infections that have become difficult or impossible to treat with conventional antibiotics. If the data mirror what it saw with LBP-EC01 (clean safety, clear efficacy signal), it opens a pathway to the first approved AI-designed bacteriophage therapeutic for a hospital-acquired infection. ## The Economics: Where the Money Comes From Phage therapy has a structural capital problem. Antibiotics are low-margin, short-course products, and venture capital has little appetite for them. Big pharma largely exited the space after the 2010s. Novartis, AstraZeneca, and Sanofi all shuttered or sold their antibiotic R&D divisions. The funding that keeps the field alive comes from government biosecurity and pandemic-preparedness mechanisms. It has received support from BARDA (up to $93 million for the UTI program), CARB-X ($12.5 million for a Klebsiella candidate), NIAID (the new $28 million contract), and it has a research collaboration with Viatris for ophthalmic infections. This is not a traditional venture-back story. It is a public-private biodefense funding model that has quietly become the dominant capital source for precision antimicrobials. 💰 **Locus Biosciences funding breakdown** BARDA: up to $93M for LBP-EC01 (UTI) Phase 2/3 NIAID: $3.3M–$28M for LBP-PA01 (pneumonia) Phase 1b CARB-X: up to $12.5M for LBP-KP01 (Klebsiella) Johnson & Johnson: up to $818M (2019 collaboration, milestone-based) Viatris: research collaboration for ophthalmic infections (2025) The broader phage therapy market is also attracting private capital. France's Phagos raised a €25 million Series A in 2025 for AI-enabled phage therapies in animal health and secured the first EU authorisation to market personalised veterinary phage treatments. Japan's Arrowsmith raised $7 million in Series A funding in mid-2026 and completed a successful FDA pre-IND meeting for its own P. aeruginosa phage cocktail. Tolka AI Therapeutics (Miami) is building a robotics and machine learning platform for personalised human phage therapy. Each of these startups is betting that AI can solve the specificity bottleneck that kept phage therapy confined to compassionate use for decades. The field is moving from laboratory curiosity to investable infrastructure. ## The AI Revolution in Phage Discovery The Locus platform is not an isolated case. The convergence of large language models, high-throughput robotics, and synthetic biology has produced a Cambrian explosion of AI-native phage startups in the last 18 months. What unites them is a shared thesis: the specificity problem that killed phage therapy in the 20th century was not a biological limitation. It was an information-processing bottleneck. The tools to solve that bottleneck now exist, and they are being deployed across the entire pipeline, from genome design to cocktail optimisation to manufacturing scale-up. The landmark proof of concept came from outside the startup world. In September 2025, a team led by Brian Hie at Stanford University published the first end-to-end generative design of complete, functional bacteriophage genomes in Nature. Using AI models trained on billions of nucleotide pairs (the same transformer architecture behind GPT), the team generated 300 candidate phage genomes, of which 16 produced viable viruses that killed E. coli. Some of the AI-designed phages outperformed the natural benchmark, and cocktails of them evolved to overcome phage-resistant bacterial strains faster than any single phage could. The finding demonstrated that AI can not only design individual proteins but entire genomes, a step change from the earlier wave of protein-structure prediction tools like AlphaFold. Alongside the genome-design breakthroughs, a parallel revolution in high-throughput wet-lab automation has closed the validation loop. Where traditional phage isolation required a microbiologist to hunt through environmental samples for weeks, platforms like Locus's own robotics pipeline can experimentally measure millions of phage-bacteria interactions in a single run. The AI trains on the data, designs the next candidate cocktail, and the robots build and test it in a closed loop of design-build-test-learn that collapses the discovery cycle from years to weeks. The field is still small enough that individual breakthroughs matter disproportionately. The Phage Therapy Summit held in Valencia in June 2026 brought together clinicians, regulators, and industry leaders to address the central question: how does phage therapy move from compassionate-use cases, last-resort treatments for dying patients, to a standard therapeutic option with a clear regulatory pathway and scalable manufacturing. The answer, increasingly, is AI. The same transformer architectures that write code and generate images are now being repurposed to design living medicines, viruses that hunt and kill specific bacteria while leaving the rest of the microbiome untouched. ## The Limits: Regulatory and Manufacturing Phage therapy faces a fundamental tension with the regulatory system. The FDA is built to evaluate fixed-composition drugs: a pill that contains the same molecule in every batch. Phages are living biologicals that evolve and change over time. A phage cocktail that works against one patient's infection may not work against the next patient, and the same cocktail can drift as the phages replicate. Regulators have had to create ad-hoc pathways (compassionate use, single-patient INDs, emergency authorisations) to accommodate a modality that does not fit the standard regulatory framework. Manufacturing is the second bottleneck. Phage production currently relies on batch fermentation with bacterial hosts, followed by purification. Scaling this to commercial volumes while maintaining sterility and genetic stability is an unsolved engineering problem. It has invested in its own US-based cGMP manufacturing facility, a strategic advantage, but industry-wide manufacturing standards are still being defined. Resistance is a legitimate concern too. Bacteria can evolve resistance to phages, typically by mutating the surface receptor the phage uses to bind. Phage cocktails mitigate this by targeting multiple receptors simultaneously, but the arms race never stops. The advantage phages have over conventional antibiotics is that the phage population can also evolve: new phages can be isolated or designed to circumvent bacterial resistance, something a fixed small-molecule drug cannot do. ## Counter: What Else Is in the Pipeline AI-designed phages are not the only approach to the AMR crisis. Several alternative strategies are competing for regulatory attention and capital: Synthetic biology has produced engineered antimicrobial peptides, short proteins that disrupt bacterial membranes. Companies like Pep-Therapy and Spero Therapeutics are advancing these through Phase 2 trials, though systemic toxicity remains a concern. Antibody-antibiotic conjugates combine the targeting specificity of monoclonal antibodies with a potent antibiotic payload. Genentech's stake in this space is significant, but no candidate has reached Phase 3 yet. Traditional antibiotic discovery, accelerated by AI screening of large chemical libraries, is seeing a resurgence. The fact that the same AI revolution that enabled AlphaFold is now being applied to small-molecule antibiotic design means the phage approach will have competitors. But small molecules will still face the same resistance problem that emptied the existing antibiotic arsenal. Phages offer a fundamentally different evolutionary dynamic that small molecules cannot replicate. 📊 **Key signals to track** LBP-PA01 Phase 1b data readout (2027): safety and efficacy against P. aeruginosa in ICU patients LBP-EC01 Phase 2/3 Part 2 results: the controlled, blinded portion of the ELIMINATE trial FDA guidance on phage therapy regulation: any movement toward a dedicated approval pathway Phagos veterinary phage commercial rollout: scale-up of the first authorised personalised phage treatment in the EU BARDA follow-on funding: whether the agency exercises options beyond the current LBP-EC01 commitment Traditional antibiotic discovery is also getting an AI reboot. Companies like Insilico Medicine and Recursion are using generative models to screen chemical libraries for novel antibiotic scaffolds, and the first AI-discovered antibiotic candidates are entering preclinical testing. The difference is that small-molecule antibiotics will face the same resistance problem that emptied the existing arsenal within a decade of deployment. Phages offer a fundamentally different evolutionary dynamic: they co-evolve with their targets, and a platform capable of rapidly designing new phages against emerging resistant strains creates a therapeutic model that does not share the fixed-product vulnerability of conventional drugs. The $3.3 million NIH award to Locus Biosciences is small by pharma standards, a fraction of what a single cancer trial costs. But it signals something larger. The US government, which spends billions of dollars annually treating hospital-acquired infections, is betting that the solution will not come from another molecule tweaked from an existing class. It will come from a platform that treats biology as code: viruses designed by AI, manufactured by robots, and deployed with the precision of a guided weapon. The Antibiotic Era lasted seventy years. What comes next will not look like it. But a small biotech company in North Carolina now has a clinical trial, a contract number, and a check with the NIH seal. The question is no longer whether phage therapy actually works as a modality. The question is how fast the regulatory and manufacturing infrastructure can catch up to what the science can already do. [ Bacteriophage therapy against multidrug resistant bacterial infections demonstrates clinical advances and engineering innovations between 2020–2026 Comprehensive review covering Locus Biosciences' LBP-EC01 Phase 1/2 data and the broader phage therapy clinical landscape. Frontiers in Microbiology ](https://www.frontiersin.org/journals/microbiology/articles/10.3389/fmicb.2026.1865548/full?ref=nexi.fund) Published July 2026\. The most current systematic review of the field, covering the Locus platform and competing approaches. [ Locus Biosciences Receives $3.3M Award From NIH FinSMEs reports on the NIAID contract award for LBP-PA01, with context on Locus's AI-powered phage engineering platform and BARDA/CARB-X partnerships. FinSMEs ](https://www.finsmes.com/2026/01/locus-biosciences-receives-3-3m-award-from-nih.html?ref=nexi.fund) January 2026\. Startup funding and deal coverage — primary details on the $3.3M award and $28M milestone structure. [ $3.3M Trial of AI-Designed Bacteriophage Therapy for HAP/VAP Clinical trial announcement with three key takeaways on the LBP-PA01 indication, the AI platform, and the public health context. Respiratory Therapy ](https://respiratory-therapy.com/disorders-diseases/infectious-diseases/other-infections/ai-designed-bacteriophage-hap-vap?ref=nexi.fund) Industry trade publication covering the respiratory care market. Focused on the ICU impact and epidemiological context.