Developing a new antibiotic costs about $1.2 billion and returns $46 million a year. Ninety are in development against 39 million forecast deaths.
The Drug That Cannot Pay For Itself
Why the pipeline is empty in the middle of an emergency
The world had 90 antibacterial agents in clinical development in February 2025, down from 97 two years earlier [1]. Fifteen meet the WHO definition of innovative. Five are active against a pathogen the WHO classifies as critical [1]. ✓ Established The shortfall is not primarily scientific. Developing an antibiotic costs about $1.2 billion and returns about $100 million or less [7].
Start with the arithmetic, because everything else follows from it. A review of antibiotic development economics published in npj Antimicrobials and Resistance in 2025 puts the cost of bringing a new antibiotic to market at approximately $1.2 billion [7]. It puts the sales such a drug can expect at about $100 million or less — a gap of roughly ten to one between what the product costs and what the market will pay for it [7]. ✓ Established The average new antibiotic generates $46 million a year in revenue [7]. A single oncology product routinely earns that in a fortnight. No investment committee, no venture fund and no board allocating capital against a required rate of return will choose the antibiotic. That is not a failure of conscience. It is the correct answer to the question the capital markets are asking.
The pipeline is what that arithmetic looks like from the outside. The WHO counted 90 antibacterial agents in clinical development as of February 2025 — 50 traditional small molecules and 40 non-traditional approaches such as bacteriophages, antibodies and microbiome-modulating agents [1]. Only 15 of the 90 satisfy the WHO innovation criteria, which require the absence of known cross-resistance, a new target, a novel mode of action or a new chemical class [1]. Only five innovative agents are active against at least one pathogen on the WHO bacterial priority pathogens list [1]. ✓ Established Of the 17 antibacterials authorised anywhere in the world since July 2017, two belong to a new chemical class [1]. The pipeline is not merely thin. It is thin in precisely the places where resistance is worst.
Behind the clinical pipeline sits a preclinical layer of 232 programmes spread across 148 research groups worldwide [1]. Ninety per cent of that work is carried out by organisations employing fewer than 50 people [1]. ✓ Established The entire early-stage discovery capacity of the field now rests on micro-enterprises and academic spin-outs, most with no revenue, no marketed product and a cash runway measured in quarters. Attrition is correspondingly brutal: between the 2023 and 2025 editions of the WHO review, four agents were authorised, one entered regulatory review and ten were withdrawn from development altogether [14]. The pipeline is losing candidates faster than it is gaining them.
The ten-to-one gap between development cost and expected sales is the structural fact that governs every other feature of this market. A 2023 analysis in The Lancet Regional Health Europe estimated the net present value of an antibiotic development project at minus $50 million, against plus $720 million for a neurological drug and plus $1.15 billion for a musculoskeletal one [13]. Estimates differ — the 2025 npj review puts the risk-adjusted figure closer to plus $100 million [7] — but no serious analysis places antibiotics near the returns available in adjacent therapeutic areas.
Capital has responded exactly as theory predicts. Since the 1990s, 18 major pharmaceutical companies have exited antibacterial research [7]. ✓ Established Antibiotics accounted for roughly 20 per cent of drugs approved by the US Food and Drug Administration in 1980; four decades later the share had fallen to about 6 per cent [7]. The four large firms that maintained credible antibacterial programmes into the modern era — GSK, Novartis, Sanofi and AstraZeneca — wound them down or divested them between 2016 and 2019 [7]. The field did not lose its scientists to a better idea. It lost its financiers to a better return.
Economists have a precise name for this configuration. An antibiotic generates enormous value that its price cannot capture: it protects people who never take it by suppressing transmission, it underwrites surgery and chemotherapy performed by clinicians who never prescribe it, and it holds option value as a reserve against a resistance event that may not arrive for a decade. None of that appears on an invoice. Meanwhile the same drug carries a negative externality for its owner, because every dose sold accelerates the resistance that destroys the asset. A product whose social value exceeds its private value by an order of magnitude, and whose private value declines with use, is the textbook definition of a market that will not clear on its own.
The consequence is a pipeline calibrated to profitability rather than to need. The GRAM Project estimates that bacterial resistance was directly responsible for 1.14 million deaths in 2021 and associated with 4.71 million more [2]. Its central forecast puts 39 million deaths directly attributable to resistance in the quarter century between 2025 and 2050 [2]. ◈ Strong Evidence Against that, the world is developing 90 candidate drugs, of which 15 are novel. The mismatch is not a failure of scientific ambition. It is a failure of accounting.
The Reward For Working Is Not Being Used
Stewardship, the patent clock and the paradox at the centre of the market
Every health system in the wealthy world instructs its clinicians to hold new antibiotics in reserve. That instruction is correct, and it is also why nobody will finance them: the better a new antibiotic is, the less of it should be sold [1] [5]. ✓ Established The market rewards volume. Medicine demands restraint. Nothing in the present system reconciles the two.
Antimicrobial stewardship is the practice of using the narrowest effective agent for the shortest effective period, and of holding the newest agents back for infections that nothing else will treat. The WHO codifies this in its AWaRe classification, which places the newest and most valuable agents in the Reserve category — drugs to be used sparingly, under specialist control, and only when first-line options have failed [1]. Hospitals in the United Kingdom, Germany, Japan and the United States all operate formularies built on that principle. A new antibiotic that reaches the market having demonstrated superiority over the standard of care is therefore rewarded with a restriction order. Its commercial launch consists of being locked in a cupboard.
The patent clock does not stop for stewardship. A composition-of-matter patent runs for 20 years from filing, and antibiotic development consumes a decade or more of that term before approval [6] [7]. What remains is a short window of market exclusivity during which the manufacturer is expected to recover $1.2 billion, while every professional body in medicine instructs prescribers to use as little of the drug as possible [7]. The moment a reserve antibiotic becomes genuinely indispensable, because resistance has eliminated the alternatives, typically arrives after exclusivity has lapsed. At that point the revenue accrues to generic manufacturers, not to the company that carried the risk.
Compare this with any other therapeutic category. A statin becomes more valuable to its owner the more widely it is prescribed, and its makers are free to say so. An oncology product is priced per course at five or six figures because the alternative is death within months and the treated population is small. An antibiotic is priced against generic comparators that cost a few dollars a course and still work most of the time, and the health system buying it will do everything in its power not to use it. There is no other product in medicine where the manufacturer, the regulator, the payer and the prescriber all agree that sales should be minimised.
A new antibiotic that fails is worthless. A new antibiotic that succeeds is placed under restriction, prescribed to a few thousand patients a year and held in reserve until resistance makes it indispensable, which is usually after the patent has expired. Excellence and revenue point in opposite directions. This is not a distortion introduced by regulation; it is correct clinical policy, faithfully implemented. The market failure is not that stewardship exists. It is that nobody arranged to pay for it.
The economics literature describes the missing element as insurance value and option value. A society holding an effective agent against carbapenem-resistant Enterobacterales is buying protection the way it buys a fire service: the value lies in availability, not in consumption. Delinked payment models exist precisely to purchase that availability, with the payer contracting for access and paying a fixed sum irrespective of doses dispensed [5] [13]. ✓ Established The DRIVE-AB consortium, which modelled the problem for the European Union and reported in 2018, recommended a market entry reward of roughly $1 billion per novel antibiotic and estimated that sustained investment on that scale would yield 16 to 20 genuinely innovative agents over 30 years [6].
Stewardship policy in every high-income health system requires that new agents be reserved. The WHO AWaRe framework formalises the restriction [1] and national formularies enforce it. The result is that a drug of exceptional clinical value generates minimal sales during its exclusivity period, and delivers its largest clinical benefit after generic entry, when the innovator captures nothing [6] [13].
The standard account of this market can be overstated, and it is worth being precise about what did and did not collapse. Aggregate antibiotic revenues did not fall away: global antibiotic sales grew from $22 billion in 1998 to $25 billion in 2009 [8]. ⚖ Contested What collapsed was the return on novel agents specifically, the narrow class of reserve drugs held back for resistant infections. The generic antibiotic market remains large, low-margin and broadly functional in high-income countries. The broken segment is the one at the frontier, and it is broken because reserve status and commercial viability are mutually exclusive under volume-based pricing.
The logical remedy has been obvious for at least fifteen years: pay for availability rather than for doses, and set the payment at a level that makes the development programme financeable. The obstacle has never been conceptual. It is that no single country wants to fund a global fix out of a national budget, and that the companies which would have benefited from the fix went bankrupt while the committees deliberated. What follows is a ledger of who did not survive the wait.
The Ledger of Companies That Succeeded and Died
Plazomicin, Achaogen and the bankruptcy of winning
Achaogen won US approval for plazomicin in June 2018 after roughly $800 million of public and private investment. In its first period on the market the drug sold $0.8 million [3] [8]. ✓ Established The company filed for bankruptcy in April 2019. The lesson investors drew was not that Achaogen executed badly. It was that approval itself is worth nothing.
Achaogen was, by the standards of the field, a success. It took an aminoglycoside programme through discovery, preclinical work and three phases of clinical trials with substantial US government biosecurity funding, and obtained approval from the Food and Drug Administration for complicated urinary tract infections. A financial post-mortem published in Humanities and Social Sciences Communications in 2024 reconstructed the flows: roughly $800 million of combined public and private capital went in, research and development expenditure peaked at $103 million in a single year, and sales in the drug's first period on the market came to $0.8 million [3]. ✓ Established The ratio between investment and revenue is approximately one thousand to one.
The equity market registered the outcome before the bankruptcy court did. Achaogen's valuation fell 95 per cent in the year preceding its filing, which closed off any possibility of raising further capital [3]. The post-mortem's broader finding is the one that matters for the pipeline: most small and medium-sized enterprises that sponsored an antibacterial approved by the FDA since 2010 have gone bankrupt or exited below the cost of their investment [3] [8]. ✓ Established Nor did public money buy public benefit. Rights to plazomicin were acquired after the bankruptcy by Indian and Chinese manufacturers, and the drug has not been widely commercialised, meaning US biosecurity investment produced neither a viable company nor reliable access [3].
Postwar antibiotic innovation ecosystems did not dissolve because global antibiotic markets broke, but because assetized markets broke them.
— Wells, Alas Portillo, Paterson, Vagneron and Kirchhelle, Public Humanities, 2025The signal travelled further than the company. Achaogen's failure established, in the mind of every specialist investor in the sector, that a novel antibiotic carries an effective market value close to zero on the day it is approved [3]. Venture funds do not price drugs; they price exits. Once the canonical exit for a successful antibiotic developer became a bankruptcy auction, the asset class ceased to exist. Melinta Therapeutics filed the same year. The withdrawal of specialist capital was not a mood. It was a rational update to a prior.
That leaves an innovation system with a structural absurdity at its centre. Ninety per cent of preclinical antibacterial research is performed by companies with fewer than 50 employees [1], and such companies exist in order to be acquired, which is the entire logic of small-cap biotechnology. But the acquirers left the field between 2016 and 2019 [7], and the public markets have priced approval at close to nothing. A small company with a promising Gram-negative candidate now faces a development path with no buyer at the end of it.
The cost structure of the trials themselves compounds the problem. Antibacterial registration trials are typically designed to show non-inferiority to an existing agent, which requires large numbers of patients in order to demonstrate that a new drug is no worse than an old one. Where a sponsor instead studies the resistant infections that justify the drug's existence, recruitment collapses: Achaogen's trial in carbapenem-resistant infections enrolled 39 patients from 2,000 screened [7]. ✓ Established A programme that must screen fifty patients to enrol one cannot be run cheaply, quickly or at scale, and the resulting evidence base is thin precisely where clinicians need it to be thick.
Fifth-year sales figures put the outcome beyond argument. A new antibiotic that survives launch, wins formulary access and reaches its fifth year on the market can expect average sales of about $120 million [7], against development costs of $1.2 billion and a patent clock already two-thirds spent. There is no configuration of clever management, pricing strategy or commercial partnership that closes that gap. These companies did not fail because they were badly run. They failed because they were solvent only in a market that does not exist.
Who Left the Room
The multinational exodus and the micro-enterprises left holding the science
Eighteen major pharmaceutical companies have exited antibacterial research since the 1990s [7]. ✓ Established GSK, Novartis, Sanofi and AstraZeneca, the last large firms with serious programmes, left between 2016 and 2019 [7]. What remains is a discovery base of micro-enterprises, 90 per cent of them employing fewer than 50 people [1].
The withdrawal was neither sudden nor secret. Companies announced it in earnings calls, framed as portfolio rationalisation. Eighteen major pharmaceutical firms have left antibacterial research since the 1990s [7]. The four that held on longest, GSK, Novartis, Sanofi and AstraZeneca, shifted out of the field between 2016 and 2019 [7]. ✓ Established Each exit removed not just a budget line but a discovery platform: compound libraries assembled over decades, medicinal chemists who understood Gram-negative permeability, screening infrastructure, and the institutional memory of which approaches had already failed and why. Those assets were not transferred to anyone. They were dispersed.
What remains is structurally unlike any other area of drug development. The WHO's 2025 preclinical review counted 232 programmes across 148 research groups, about 90 per cent of them run by organisations with fewer than 50 employees [1]. ✓ Established In oncology, early discovery is dominated by large integrated firms with balance sheets that can absorb a decade of failure. In antibacterials it is performed by companies that must raise money every eighteen months from investors who have watched the sector's successful products end in bankruptcy auctions [3]. The science is being carried by the least financially resilient organisations in the industry.
It would be dishonest to present this as a purely financial problem. Gram-negative bacteria, the group that includes the carbapenem-resistant Enterobacterales, Acinetobacter and Pseudomonas species at the top of the WHO priority list, are genuinely hard targets. Their outer membrane sharply limits which molecules can enter; efflux pumps actively expel those that do; and extended-spectrum beta-lactamases and carbapenemases destroy many of the compounds that survive both barriers [7]. ◈ Strong Evidence These are chemistry problems that money alone does not solve. The honest formulation is that the field faces a hard scientific problem and has simultaneously dismantled the institutions best equipped to work on it.
Antibacterial discovery now sits almost entirely with companies of fewer than 50 people, whose business model requires an acquirer. The acquirers left between 2016 and 2019. Public markets price a newly approved antibiotic at close to zero, as Achaogen demonstrated in the most expensive way available. A small company therefore develops a drug it cannot sell, for a market that will not pay, hoping to be bought by firms that no longer buy. Every element of that sentence is documented. Together they describe an innovation system with no terminal state other than failure.
Attrition data from the WHO reviews make the direction of travel explicit. Between the 2023 and 2025 editions, four antibacterial agents were authorised, one entered regulatory review and ten were withdrawn from development [14]. ✓ Established The clinical pipeline fell from 97 agents to 90 over the same period [1]. Non-traditional approaches, including bacteriophages, monoclonal antibodies and microbiome-modulating products, now account for 40 of the 90 candidates; the science is genuinely promising, but attrition among them is higher still and none has yet produced a broadly deployed product for a priority pathogen [1].
The clinical pipeline contracted from 97 agents to 90 over the same two years [1] [14]. Withdrawals outnumbered authorisations by more than two to one. In a field where the WHO identifies only five clinical-stage innovative agents active against a critical priority pathogen, a net loss of candidates is not portfolio pruning. It is contraction of the only supply the world has [1].
The loss of large-firm capability has a second-order effect that rarely appears in pipeline counts: manufacturing and regulatory competence. Bringing an antibacterial through chemistry, manufacturing and controls review requires expertise in sterile injectable production that small companies routinely lack and cannot afford to build. Approval is not the last hurdle. A sponsor that clears efficacy and safety can still fail on manufacturing documentation, and a company of 40 employees has no depth with which to absorb that. The large firms that departed were the ones that could have supplied it.
The result is an inversion of the normal relationship between risk and capacity. The organisations bearing the highest scientific risk in modern medicine are the ones with the least capital, the shortest runway and the weakest manufacturing base. The organisations with capital, runway and manufacturing left the field for portfolio reasons that were, from a shareholder's perspective, entirely defensible. Nobody in this story behaved irrationally. That is what makes it a market failure rather than a morality tale.
The Bill That Arrives When Prophylaxis Fails
What a post-antibiotic surgical ward actually costs
Between 39 and 90 per cent of surgical site infections in the United States are already caused by organisms resistant to the recommended prophylactic antibiotic [9]. ◈ Strong Evidence A 30 per cent loss of prophylaxis efficacy would add 120,000 infections and 6,300 deaths a year in the United States alone [9]. Modern surgery is an antibiotic-dependent technology.
Antibiotic prophylaxis is invisible infrastructure. A single dose given before incision is what makes a hip replacement a routine procedure rather than a gamble, and the same logic underwrites caesarean section, pacemaker implantation, spinal surgery, appendectomy, hysterectomy, colorectal surgery, transrectal prostate biopsy and the neutropenic phase of cancer chemotherapy. A 2015 modelling study in The Lancet Infectious Diseases examined the ten most common procedures in the United States that depend on prophylaxis and asked what happens as the prophylactic agents stop working [9].
The baseline was already worse than most clinicians assume. The study found that between 39 per cent of infections after caesarean section and 50 to 90 per cent after transrectal prostate biopsy are caused by organisms resistant to the recommended prophylaxis [9]. ◈ Strong Evidence Modelling a further 30 per cent reduction in prophylactic efficacy produced an additional 120,000 infections and 6,300 infection-related deaths per year in the United States [9]. Those are not deaths from exotic organisms in intensive care. They are deaths following planned, elective, ordinary operations.
The global mortality picture is now measured rather than guessed at. The GRAM Project's 2024 analysis in The Lancet attributed 1.14 million deaths in 2021 directly to bacterial resistance and found resistance associated with a further 4.71 million [2]. ✓ Established Its forecast for 2050 is 1.91 million attributable and 8.22 million associated deaths a year, increases of 67.5 and 74.5 per cent respectively [2]. Cumulatively, the analysis projects more than 39 million deaths directly attributable to resistance between 2025 and 2050, and 169 million with resistance involved [2].
The internal structure of that burden has changed in a way that alters the policy problem. Deaths from resistant infections in children under five fell by about 50 per cent between 1990 and 2021, a genuine public health achievement driven by vaccination, sanitation and neonatal care [2]. Over the same period, deaths among adults aged 70 and over rose by more than 80 per cent, and are forecast to rise a further 146 per cent by 2050 [2]. ✓ Established Resistance is becoming a disease of the elderly and the surgically treated in wealthy ageing societies, which is precisely the population whose care depends on the reserve agents nobody will finance.
The same GRAM analysis estimates that better care and better access to existing antibiotics could avert 92 million deaths between 2025 and 2050. That figure is not a projection of new science. It is a measure of the distance between what medicine can already do and what it is funded to do. The drugs required to close it are the drugs whose developers went bankrupt on approval. A seventy-year-old waiting for a hip replacement in 2040 will not be harmed by a scientific limit. They will be harmed by a discount rate applied in 2019.
The macroeconomic estimates are older but have not been superseded in scale. The World Bank modelled global output under two resistance scenarios in 2016. In the low-impact case, global GDP growth loses 1.1 per cent a year by 2050 and the annual output shortfall exceeds $1 trillion after 2030. In the high-impact case, growth loses 3.8 per cent and the shortfall reaches $3.4 trillion [12]. ◈ Strong Evidence Healthcare costs rise by 25 per cent in low-income countries, 15 per cent in middle-income countries and 6 per cent in high-income countries [12]. The distributional pattern matters: the countries least able to absorb the cost face the largest proportional increase.
Set those figures against the cost of the remedy. The modelled global pull incentive required to make antibiotic development financeable is on the order of $310 million a year, or $2.2 to $4.8 billion in total for full delinkage [15]. The World Bank's conservative case puts the annual cost of inaction above $1 trillion [12]. The ratio between the price of the fix and the price of the failure is roughly three thousand to one. No government has yet been willing to pay the smaller number.
Seven Countries, Seven Answers
The national experiments in paying for availability rather than volume
Since 2020, seven jurisdictions have tried to pay for antibiotics without paying per dose [5] [13] [15]. ✓ Established Sweden guarantees a floor, the United Kingdom pays a fixed annual subscription, Japan assures revenue, Italy routes antibiotics into an innovation fund, and the United States has failed to pass its bill in four consecutive Congresses [10].
Sweden went first. In July 2020 the Public Health Agency contracted for guaranteed availability of five antibiotics from four manufacturers, paying each roughly 400,000 euros a year in return for maintaining a security stock in the country [13]. Regions continued to buy doses normally, and central government topped the payment up to the guaranteed floor when regional purchasing fell short. The evaluation found the model achieved what it was designed to achieve, with Sweden obtaining access to several new agents earlier than comparable European countries, while being explicit that a payment of this size buys availability rather than innovation [13].
The United Kingdom built the first fully delinked model at scale. NHS England and NICE began subscription-style contracts in April 2022 for cefiderocol and ceftazidime-avibactam, paying about 10 million pounds per antibiotic a year regardless of volume dispensed [13]. The expanded scheme opened for tender in August 2024 with an annual budget of 100 million pounds and four value bands of 5, 10, 15 and 20 million pounds a year, assigned by a NICE expert panel against 17 criteria covering clinical need, pharmacological benefit and health system benefit [5]. ✓ Established Contracts run for an initial three years, extendable to a maximum of 16, with the tendered value approaching 1.9 billion pounds over that horizon [5].
Japan chose revenue assurance over subscription. The Ministry of Health, Labour and Welfare launched an antimicrobial securement support programme in fiscal 2023 and designated cefiderocol as its first product in November of that year, guaranteeing the manufacturer a revenue level in exchange for maintained supply and cooperation with stewardship policy [15]. The ministry framed the pilot as a test of three propositions at once: whether a pull incentive delivers appropriate use, whether it promotes research and development, and whether reimbursement can be set at a defensible level [15]. Industry bodies in Japan had previously proposed delinked rewards of 20 to 80 billion yen over ten years [15].
Italy took a different route again, using an existing instrument rather than building a new one. Its 2025 budget law grants new and recently approved anti-infectives, classed as Reserve under the WHO AWaRe system or targeting a WHO priority pathogen, direct access to the national fund for innovative medicines, with up to 100 million euros a year available for their reimbursement [15]. Canada opened a three-year antimicrobial economic incentives pilot running from the 2024/25 fiscal year to 2026/27 [15]. The European Union settled its own approach in December 2025, agreeing a transferable data exclusivity voucher: the developer of a priority antimicrobial receives 12 additional months of data protection, usable once, on that product or another, subject to a blockbuster clause excluding products with gross annual sales above 490 million euros [11]. ✓ Established
| Approach | Effectiveness | Assessment |
|---|---|---|
| Fully delinked subscription, United Kingdom | The only model paying a fixed, volume-independent fee at national scale. Bands reach 20 million pounds a year and contracts run to 16 years, but the United Kingdom alone accounts for a small fraction of the modelled global requirement. | |
| Market entry reward at PASTEUR scale, proposed | Contracts of $750 million to $3 billion over five to ten years would, on their own, approach the modelled global target. The mechanism has never been enacted and the bill has failed in four consecutive Congresses. | |
| Innovation fund routing, Italy | Reuses an existing reimbursement instrument, so it needed no new legislation and started quickly. Italy is one of only two countries currently on track for its mid-range fair share target. | |
| Revenue guarantee, Sweden | Guaranteed payments of about 400,000 euros a year secured earlier access to new agents than in comparable countries, but the sums are orders of magnitude below what changes a development decision. | |
| Transferable exclusivity voucher, European Union | Twelve months of transferable data exclusivity moves cost onto payers and patients in unrelated therapy areas through delayed generic entry. The blockbuster clause limits but does not remove the transfer, and the innovation effect is unproven. |
The United States, which carries the largest single share in any fair share calculation, has done the least. The PASTEUR Act would authorise the Department of Health and Human Services to sign subscription contracts worth $750 million to $3 billion each over five to ten years, backed by $6 billion in appropriations [10]. It was first introduced in 2019 and has been reintroduced in every Congress since, most recently in the 119th [10]. ✓ Established It has never reached a floor vote. Its headline appropriation was cut from $11 billion to $6 billion across successive drafts in an attempt to improve its prospects, which has not worked.
Read together, the national programmes describe a genuine policy shift and an insufficient one. The mechanism works where it has been implemented: the United Kingdom pays, Sweden secured access, Italy reached its target, Japan is testing the proposition on a real product. What is missing is scale and coordination, a set of national schemes that individually make sense and collectively fall short of the sum required to change a single development decision. The next section quantifies how short.
Does Any of It Work
The fair share gap and the argument about what actually broke
For a new antibiotic to be financeable, the G7 and EU27 together would need to generate about $363 million a year in global revenue for it [4]. ⚖ Contested Only the United Kingdom and Italy are on track for their share of that mid-range target, and no G7 member meets the high-end target at current prices or volumes [4].
The most rigorous attempt to quantify the shortfall was published in eClinicalMedicine in September 2025. Using cefiderocol and ceftazidime-avibactam as representative reserve antibiotics, and allocating responsibility by GDP, it calculated global annual revenue targets of $258 million at the low end, $363 million mid-range and $562 million at the high end, sustained over ten years [4]. National shares follow from that: $105 to $228 million for the United States, $104 to $227 million for the EU27 collectively, $24 to $52 million for Japan, $22 to $49 million for Germany, $15 to $34 million for the United Kingdom, $13 to $29 million for Italy and $9 to $21 million for Canada [4]. ✓ Established
The single most clarifying number in that analysis is a ranking. An antibiotic reaching the mid-range target of $363 million a year would rank approximately 230th among all drugs worldwide by revenue [4]. ✓ Established The world is failing to finance a product category whose commercial requirement is to be moderately successful. This is not a demand for blockbuster economics. It is a demand that the 230th-best-selling drug on earth be viable, and the answer, as things stand, is that it is not.
The Case That Pull Incentives Will Work
NHS England has run fully delinked contracts since 2022 and has expanded to a 100 million pound annual budget with bands rising to 20 million pounds per product.
The United Kingdom through its subscription programme and Italy through higher use and reimbursement reform are on track for their mid-range fair share.
A qualifying antibiotic would rank about 230th in the world by revenue. Full delinkage is modelled at $2.2 to $4.8 billion in total.
The Swedish guarantee secured earlier availability of new agents than in comparable European countries, at roughly 400,000 euros per product a year.
PASTEUR contracts of $750 million to $3 billion would, on their own, approach the modelled global target if the bill were enacted.
The Case That They Will Not
PASTEUR has been introduced in four consecutive Congresses without reaching a floor vote, and its appropriation has been cut from $11 billion to $6 billion.
Incentives are geographically fragmented and legally incompatible. No G7 member meets the high-end fair share target at current prices or volumes.
The EU transferable exclusivity voucher is funded by patients in unrelated therapy areas through delayed generic entry, and its innovation effect is unproven.
Gram-negative outer membranes, efflux pumps and carbapenemases are scientific barriers that a revenue guarantee does not address.
Historians of the field argue the ecosystem was destroyed by the assetisation of pharmaceutical research, not by an absence of market demand.
The transferable voucher deserves particular scrutiny, because it is the mechanism the European Union actually chose. Under the December 2025 agreement, the developer of a priority antimicrobial receives 12 months of additional data protection, transferable once to any centrally authorised product, subject to a blockbuster clause excluding products with gross annual sales above 490 million euros over the preceding four years [11]. ⚖ Contested The economics are unusual. The reward is funded not by health systems buying antibiotics but by patients and payers in whatever therapy area the voucher is ultimately applied to, through a year of delayed generic or biosimilar entry. Critics note that the size of the reward is therefore unrelated to the value of the antibiotic, and that the cost falls on people who will never take it.
Antimicrobial resistance is escalating. But the pipeline of new treatments and diagnostics is insufficient to tackle the spread of drug-resistant bacterial infections.
— Dr Yukiko Nakatani, WHO Assistant Director-General, October 2025A more fundamental challenge comes from historians of the field. A 2025 study in Public Humanities traced the empty pipeline metaphor from its emergence in the mid-1990s and argued that the market failure framing is itself a historical artefact, one that reframed a public health crisis as an incentives problem and thereby narrowed the range of admissible solutions [8]. ⚖ Contested Its evidence includes the awkward fact that global antibiotic revenues rose from $22 billion in 1998 to $25 billion in 2009, during the very period the pipeline was said to be collapsing [8]. Its conclusion is that postwar antibiotic innovation systems dissolved not because antibiotic markets broke, but because the assetised model of pharmaceutical research broke them [8]. If that diagnosis is right, paying more into the same structure will produce more of the same outcome.
The scientific counter-argument runs in a different direction and is equally uncomfortable for incentive advocates. The chemistry of Gram-negative penetration has defeated well-funded programmes repeatedly: the outer membrane restricts influx, efflux pumps expel what enters, and beta-lactamases and carbapenemases inactivate much of the rest [7]. ◈ Strong Evidence A large share of the clinical pipeline consists of derivatives of existing classes, which is a rational response to difficulty rather than a failure of ambition [1]. A pull incentive changes who pays for a programme. It does not make the outer membrane permeable.
The honest position holds all three arguments at once. The financial constraint is real and quantified: development costs $1.2 billion against $46 million a year in revenue [7]. The structural critique is also real: an industry organised around asset appreciation will not sustain a product whose value lies in not being sold [8]. And the scientific constraint is real: the hardest targets remain hard whoever funds the work [7]. What distinguishes the financial constraint is that it is the only one of the three that can be removed by a decision taken in a single budget cycle.
An Infrastructure Problem Filed Under Investment
What the evidence actually tells us
The modelled cost of repairing the antibiotic market is about $310 million a year globally [15]. The World Bank's low-impact estimate of the cost of not repairing it is an annual output shortfall above $1 trillion after 2030 [12]. ◈ Strong Evidence That ratio, roughly three thousand to one, is the entire argument, and it has been available since 2016.
Very few public policy questions arrive with a cost-benefit ratio this lopsided and this well documented. On one side: a global pull incentive of $2.2 to $4.8 billion in total to fully delink antibiotic revenue from sales volume [15], or $363 million a year in global revenue per antibiotic to reach the mid-range fair share target [4]. On the other: 39 million deaths directly attributable to resistance between 2025 and 2050 [2], 120,000 additional surgical infections a year in one country under a moderate loss of prophylaxis efficacy [9], and a trillion-dollar annual drag on global output [12]. ◈ Strong Evidence None of these figures is disputed by any serious party. The money has still not been appropriated.
The mismatch persists because antibiotics are filed in the wrong conceptual category. They are treated as pharmaceutical assets, evaluated on projected revenue, discounted for risk and compared against alternative uses of capital. They function as public infrastructure, the thing that makes surgery, transplantation, neonatal intensive care and chemotherapy possible, and their value lies mainly in availability rather than consumption [8] [9]. No treasury asks a flood barrier to demonstrate an internal rate of return proportional to the water it holds back. Applied to antibiotics, that standard produces a rational decision not to build the barrier.
The global pull incentive modelled as sufficient to restore antibiotic development costs roughly $310 million a year. The World Bank's conservative estimate of the annual output lost to resistance after 2030 exceeds $1 trillion. The ratio is about three thousand to one, before counting a single death. What blocks the trade is not cost but structure: the payment must be made now, by national treasuries, in exchange for a benefit that is global, diffuse and realised decades later by other governments. Every incentive in the system points toward waiting for somebody else to pay.
Four findings in this report are not seriously contested. Development costs approximately $1.2 billion and returns approximately $46 million a year [7]. The clinical pipeline shrank from 97 agents to 90 between 2023 and 2025, with 15 classed as innovative and five active against a critical priority pathogen [1]. Eighteen major companies have left the field since the 1990s, and 90 per cent of remaining preclinical work is done by firms of fewer than 50 people [1] [7]. And most small companies that successfully brought an antibacterial to market since 2010 went bankrupt or exited below cost [3]. ✓ Established
Three things are genuinely open. Whether pull incentives at achievable scale will refill the pipeline is unproven, because the United Kingdom's contracts are recent, Italy's fund is new, and no country has yet run a delinked model long enough to observe a development decision taken in response to it [4] [5]. ⚖ Contested Whether the market failure diagnosis is even correct is disputed by historians who locate the cause in the assetisation of pharmaceutical research rather than in market size [8]. And whether more money would solve the Gram-negative chemistry problem is doubtful on the evidence [7].
The United Kingdom commits 100 million pounds a year across all products, Italy up to 100 million euros, Sweden about 400,000 euros per product, and Japan runs a single-product pilot [5] [13] [15]. The modelled global requirement is about $310 million a year per qualifying antibiotic, sustained over a decade [15]. Only the United Kingdom and Italy are on track for their mid-range fair share, and no G7 member reaches the high-end target [4].
There is a second failure that the incentive debate largely ignores. Approval is not access. Ninety per cent of preclinical antibacterial research takes place in high-income regions, while the countries carrying the heaviest resistance burden face registration gaps, absent suppliers and prices set for wealthy health systems [8] [1]. A pull incentive paid by the G7 secures supply for the G7. The GRAM analysis estimates that better care and better access to drugs that already exist could avert 92 million deaths between 2025 and 2050 [2], a larger prize than anything the pipeline can deliver over the same period, and one that requires no new chemistry at all.
The antibiotic pipeline is the clearest case in modern health policy of a solved problem left unfunded. The science is difficult but tractable, the mechanism is designed, modelled and running in at least four countries, and the sum required would not register against a single defence procurement line. What is missing is the willingness of any government to pay a national price for a global good in advance of the emergency that would justify it. The deadline is not abstract. It arrives the day a surgeon in an ordinary hospital, preparing for an ordinary hip replacement, has nothing left to give before the incision.