The Phage Paradox: The Drugs That Work Keep Dying

The Graveyard

PHAXIAM Therapeutics treated over 140 patients. Its PhageBank held more than 1,000 characterized bacteriophages. Its GLORIA trial had an FDA IND. Analysts projected €100 million in revenue by 2030.

In June 2025, PHAXIAM entered receivership. By March 2026, it was heading toward liquidation. The bid review hearing has been postponed indefinitely. No acquirer has materialized. The company sits on €150 million in debt against €3.6 million in cash. Its collaboration with TechnoPhage — announced in January 2025 — survived 35 days before the parent company collapsed. A decade of phage science, a thousand catalogued viruses, and clinical infrastructure built across multiple countries, all heading for a fire sale.

Then there is BiomX. In July 2025, Nature Communications published the BX004-A Phase 1b/2a data: a 500-fold (2.7 log₁₀) reduction in Pseudomonas aeruginosa density in the lungs of nine cystic fibrosis patients. Clinically meaningful. Biologically clean. The kind of data that makes infectious disease physicians sit up.

Five months later, BX004 was dead.

The full story is more complicated than “good science, no money” — and honesty about it matters. In August 2025, the FDA placed a clinical hold over a nebulizer device issue. BiomX believed that was resolved by October. But on November 25, a Data Monitoring Committee safety review of the Phase 2b trial found “unexpectedly high rates of adverse events.” The DMC initially recommended continuing with adjusted dosing. Thirteen days later, on December 8, BiomX discontinued the program entirely — citing both the adverse events and resource constraints. CEO Assaf Solomon: “projected timelines and resources required... were beyond the Company’s currently available resources.” Cash on hand: $8.1 million. The Israeli subsidiary filed for insolvency.

The Phase 1b/2a data was excellent. The Phase 2b hit real problems. The graveyard has multiple causes — and that complexity is the point. Good early data does not guarantee scalable therapy, and underfunded companies cannot survive the valley of discovery.

If this sounds familiar, it should. Achaogen developed plazomicin, won FDA approval, and filed for bankruptcy seven months later. Melinta Therapeutics filed for Chapter 11 with four approved antibiotics in its portfolio. Tetraphase and Paratek ran out of runway. Venatorx collapsed in 2024 with cefepime-taniborbactam, despite FDA priority review. The same economic gravity that crushed antibiotic companies is now crushing phage companies — amplified by phage-specific challenges: personalization, manufacturing complexity, regulatory ambiguity, and a reimbursement system that has no category for what phage therapy is.

The Evidence That Won’t Go Away

What makes this a paradox rather than simply a failure is that the clinical evidence keeps accumulating. Not theoretical. Not preclinical. Patients treated, outcomes measured, data published in high-impact journals.

This is not a pipeline of theoretical candidates. These are clinical results from patients who were treated and got better. Exhibits 1 through 4 are randomized or controlled data in peer-reviewed journals. Exhibit 6 covers 100 patients across 12 countries. The science is not the problem.

The Last Company Standing

Armata Biosciences operates the only Phase 3 phage therapy program in the world. AP-SA02 targets Staphylococcus aureus bacteremia — a disease that kills more people than AIDS, tuberculosis, and hepatitis combined. The Phase 2a diSArm trial showed 100% response at 28 days without relapse, versus 58% for standard of care (p=0.047). The FDA has aligned on a superiority design for Phase 3, with initiation planned for H2 2026. AP-SA02 received QIDP designation in February 2026. Armata’s 56,000-square-foot cGMP facility in Los Angeles was formally commissioned to produce clinical-grade phages.

All of this is real. And all of it sits on top of this:

The clinical data says Armata’s phage works. The balance sheet says the company might not survive to prove it. If Armata fails, there is no commercial Phase 3 phage program anywhere on Earth. The world’s entire regulatory pathway to a first-ever FDA-approved phage product runs through a single company with $14.8 million in cash against $140 million in current liabilities.

The Other Survivors

Locus Biosciences is the best-funded phage company alive. Its CRISPR-Cas3 engineered phages are 100–1,000 times more lethal than natural phages — they don’t just kill bacteria, they shred their chromosomes. LBP-EC01’s ELIMINATE trial is in Phase 2/3 Part 2: up to 288 patients, randomized, blinded, with registrational intent. BARDA backing has grown to $85 million within a $152 million total program — the single largest phage therapy investment ever made. Locus also has LBP-PA01 in Phase 1b ($28 million NIH) and a new ophthalmic research collaboration. But they are still in enrollment, not yet at a Phase 3 readout. The data is coming. It isn’t here yet.

SNIPR Biome is doing something different. SNIPR001 is a CRISPR-based four-phage cocktail designed not to treat active infections but to prevent them — specifically, gut decolonization of E. coli in hematologic cancer patients undergoing stem cell transplant. Phase 1 data from 36 patients showed good tolerability and numerically lower gut E. coli levels. It’s a different application of phage biology — oral dosing, prophylactic intent, cancer-adjacent — and it broadens the definition of what phage therapy could become.

TechnoPhage in Lisbon is running the RECOVER trial — a Phase 1/2a with 15 patients testing TP-122A against ventilator-associated pneumonia caused by P. aeruginosa. TP-102 for diabetic foot infections has a Phase 2b/3 planned. Small, independent, European-funded. The PHAXIAM collaboration collapsed with the parent company, but TechnoPhage carries on.

BiomX itself isn’t dead — it has pivoted. BX011, a fixed multi-phage cocktail targeting S. aureus in diabetic foot infections, is now the company’s sole clinical focus. The FDA supports a BLA pathway with no additional nonclinical studies required. $40 million in DHA/Navy non-dilutive funding keeps the program alive. But Phase 2a is entirely funding-dependent, and the company has $8.1 million in cash with an insolvent subsidiary.

Phagos in Paris raised a €25 million Series A in October 2025, built on its Alphagos AI platform for phage discovery. It’s early — veterinary to human pipeline — but it’s new capital entering the space when everyone else is dying.

Two Roads

The fundamental question in phage therapy is no longer whether it works. It’s which development model will make it available to patients. Two roads are diverging, and they lead to very different futures.

The pharma road has better data quality — randomized, controlled, publishable in top journals. It is also killing every company that walks it. The public road has treated more patients, faster, at lower cost — but through fragmented compassionate-use channels with a bottleneck that ranges from 28 to 386 days (median 171 days) from request to phage administration. Neither road is working well enough. But only one of them is generating survivors.

The Regulatory Moment

Something shifted in 2025. After decades of regulatory silence on phage therapy, four separate frameworks are now converging simultaneously.

The field is professionalizing at the regulatory level at the exact moment companies are dying at the commercial level. Regulators are building roads. Companies can’t afford to drive on them.

The Money Question

Phage therapy faces every financial obstacle that crushed antibiotic companies, plus extras. As I covered in The Global Experiment, the fundamental problem with antibiotics is the stewardship paradox: the better a drug works, the less it should be used, and the less money it makes. Phage therapy amplifies this. Each treatment is personalized — selected for a specific patient’s specific bacterial strain — which means lower manufacturing volume, higher per-unit cost, and a market that doesn’t scale the way pharmaceutical pricing models require.

The projected global phage therapy market is $44 million in 2025, growing to $162 million by 2033. For comparison, a single successful cancer immunotherapy generates more revenue in a quarter than the entire phage market generates in a year.

Government funding has become the lifeline:

• $85M+ BARDA for Locus Biosciences (within $152M total program) — the largest single phage investment in history
• $40M DHA/Navy for BiomX BX011 — keeping a program alive that the company cannot fund itself
• $26.2M DoD for Armata — military funding for a civilian disease because the market won’t pay
• $28M NIH for Locus LBP-PA01 Phase 1b

The Department of Defense is funding phage therapy because soldiers get wound infections that resist every antibiotic. BARDA is funding it because carbapenem-resistant infections are a biodefense priority. These are national security dollars filling a hole that the pharmaceutical market refuses to fill.

The UK subscription model — which delinks antibiotic revenue from volume — could theoretically expand to phage, and the UK’s National Action Plan on AMR has expressed commitment to novel therapeutics. But the model is designed for standardized drugs with predictable dosing, not personalized viral cocktails assembled per patient. Adapting it for phage would require rethinking the entire framework.

The ASM Phage Therapy Coordination Initiative is building a centralized clinical data registry — an attempt to aggregate the fragmented evidence base and make it usable by regulators. It’s infrastructure work, not glamorous, but exactly what the field needs.

What This Means

I won’t dress this up with false hope. The phage therapy field in March 2026 is a paradox that hasn’t resolved. The science is real — not proven at scale, but generating statistically significant clinical results across multiple indications, pathogens, and geographies. The commercial pathway is failing — not because the drugs don’t work, but because the economic structure of pharmaceutical development cannot sustain personalized, niche, stewardship-limited therapeutics.

Two futures are possible. In one, Armata survives, its Phase 3 succeeds, AP-SA02 becomes the first FDA-approved phage product, and a regulatory pathway is established for others to follow. Locus reaches Phase 3 readout. BARDA continues funding. The pharma road works — slowly, expensively, with enormous attrition, but it works.

In the other future, Armata runs out of money. The pharma road collapses entirely. And the public infrastructure model — France’s PHAG-ONE producing GMP phages, Belgium’s magistral framework compounding per-patient treatments, European regulatory harmonization enabling cross-border compassionate use — becomes the only path forward. Not through FDA approval. Through public health systems treating patients directly.

Neither future is guaranteed. Both are plausible. The Armata 10-K, due by March 31, will tell us which direction the needle is moving. If the filing reveals new financing or partnership, the pharma road lives another day. If it reveals further deterioration, the world’s only Phase 3 phage program enters survival mode.

Phage therapy is caught between what science can prove and what markets can sustain. That gap has killed every company in the antibiotic graveyard. The question is whether phages find a way out — or join the dead.