1. Introduction - A Deficit Imaginary

Narratives have power. Social scientists have long highlighted how language structures humans’ capacity to perceive the world, reflects power imbalances, and impacts policymaking.Footnote ¹ This historical review examines the rise, circulation, and impact of one of the most influential narratives in antibiotic governance—the metaphor of the “empty antibiotic pipeline.” Starting in the 1990s and gaining traction since 2010, the empty pipeline metaphor spurred calls to “fix” the “broken antibiotic market” and “restore” research and development (R&D). The resulting inflows of public and philanthropic funding have outpaced other forms of antimicrobial resistance (AMR) investment and intervention, including those in vaccines and diagnostics.Footnote ² The underlying linear imagery has galvanised a directed innovation ecology consisting of publicly funded incentives to “push” or “pull” compounds through the empty pipeline alongside dedicated public–private partnerships (PPPs) and product development partnerships (PDPs).Footnote ³ Yet antibiotic innovation remains insufficient to address AMR, large developers have continued to exit, and economic volatility within the pipeline has not diminished.Footnote ⁴ Meanwhile, questions of affordability and access-oriented design of novel technologies have long been neglected.Footnote ⁵ Have public interventions been insufficient? Or is there a deeper problem with the central metaphor structuring global action?

Our historical review of antibiotic innovation since 1980 illuminates the sociopolitical roots of the “empty pipeline” metaphor. Each part addresses a distinct question: (1) How did the historical ecosystem for antimicrobial innovation function, and why did it rupture? (2) Who shaped the emergent “empty antibiotic pipeline” narrative? (3) How did remaining R&D companies adapt to antibiotics’ declining commercial status? (4) What were the geopolitical dynamics impacting the uptake of the pipeline metaphor and the resulting ecosystem of incentives, PPPs, and PDPs? (5) Can assetised public–private R&D align with what anthropologist Clare Chandler describes as antimicrobials’ infrastructural role within essential health and food production systems? (6) To what extent do current attempts to “fix” the innovation pipeline take into account questions of drug access and R&D ecosystems outside of high-income countries (HICs)?Footnote ⁶

To answer these questions, we draw on the substantial body of social sciences research on discursive performativity in biomedical R&D, and combine a quantitative thematic-trends evaluation of antibiotic innovation-centred publications (Figure 1), with a qualitative analysis of relevant historiography, policy reports, and U.S. Securities and Exchange Commission (SEC) records.Footnote ⁷ We also draw on a 2024 historical witness seminar recording reflections of key decision-makers from clinical medicine, industry, and international organisations who shaped AMR policy and financial models for antibiotic development from the 1990s.Footnote ⁸

Figure 1. Thematic trends analysis search strategy.

Part One reconstructs antibiotic innovation prior to the 1980s. It highlights a diversity of public and private R&D approaches and factors contributing to the dissolution of the early innovation ecology.

Part Two examines the rise of the “empty antibiotic pipeline.” Although the use of a pipeline metaphor to describe drug development dates to the mid-twentieth century, our analysis of scientific literature and policy reports indicates that the application of the “empty pipeline” metaphor to the antibiotic R&D space only began during the mid-1990s.Footnote ⁹ Implicit in the metaphor’s use was an assumption that innovation was a unidirectional process primarily entailing proprietary development of novel molecular compounds.Footnote ¹⁰ Characterisation of the antibiotic pipeline as “empty,” despite ongoing R&D, created what social scientists have termed a “deficit model of innovation” that justified selective market-focused public interventions.Footnote ¹¹ Industry use of this deficit model to reframe AMR as an innovation challenge coincided with expiring blockbuster patents and controversies about intellectual property (IP) impacts on drug access. Company histories show that many large developers were uninterested in returning to or reprioritising antibiotic R&D.Footnote ¹²

Part Three reconstructs how, from the 2000s, the remaining large companies began prioritising mergers and acquisitions over in-house R&D, leaving antibiotic innovation increasingly dominated by small- and medium-sized enterprises (SMEs).Footnote ¹³ While SMEs focused on adapting known compounds for emerging hospital “superbugs” like Methicillin Resistant Staphylococcus aureus (MRSA), their financially vulnerable business models—and dependence on speculative investment and value-based pricing—contributed to a more volatile R&D ecosystem.Footnote ¹⁴

Part Four analyses how the failure of this volatile innovation ecosystem to keep up with AMR became evident around 2010, amidst surging Gram-negative infections. Funders’ preference for incentives-based interventions led to a doubling-down on the “empty pipeline” metaphor’s market failure diagnosis and attempts to reattract large companies via “push” subsidies for early-stage R&D, and market-shaping “pull” incentives for licensed compounds.Footnote ¹⁵ “Delinkage” emerged as a second key narrative to reconcile stewardship proponents (who initially resisted prioritising innovation) with industry strategists (who were sceptical about state intervention).Footnote ¹⁶ Proponents argued that value-based public subscriptions, or prizes for access to novel compounds, could rejuvenate commercial R&D and avoid overmarketing.Footnote ¹⁷

Part Five studies how “fixing” the pipeline with incentives led to a growing divide between what novel antibiotics should cost as an essential infrastructure for health service delivery according to value-based pricing models, and what societies can afford to pay.Footnote ¹⁸ The article discusses the SME Achaogen, which “got everything right” in funding terms, raising almost $800 million of public and private investment to develop its antibiotic plazomicin, but suffered bankruptcy after generating only $0.8 million in revenues.Footnote ¹⁹

Part Six ends by assessing how the rise of the pipeline metaphor and incentives-based innovation interventions have impacted antibiotic drug access. It argues that chronic shortages of generic antibiotics are indicative of a wider crisis of antibiotic supply chains and markets. These issues are not adequately being addressed by current innovation initiatives and may require novel forms of decentralised R&D.Footnote ²⁰

The conclusion argues that attempting to satiate open-ended investor expectations using public incentives is not sustainable. Diversifying R&D ecosystems is key to securing sustainable antibiotic innovation and equitable access.

2. Part One: the “Golden Era” of antimicrobial innovation

Between 1900 and the 1970s, antimicrobial innovation surged. Over the past decades, historians of science and medicine have published both witness seminars and detailed historical overviews of the rise of arsenic-based antimicrobials around 1900, synthetic sulphonamides in the mid-1930s, and biological and semi-synthetic antibiotics from the 1940s and 1950s.Footnote ²¹ According to scholars such as Vivianne Quirke, Robert Bud, Jean-Paul Gaudillière, and John Lesch, early innovation ecosystems were “big,” integrated, and spanned public and private sectors.Footnote ²² R&D was often undertaken in proximity to sites of testing and production. In an age of process rather than product patents, this combination of scale and proximity enabled compounds and know-how to circulate amongst institutions and across borders. In the case of fermentation-based production, close interaction between scientists and engineers also enabled a rapid upscaling of production.Footnote ²³ Infrastructural integration of innovation ecosystems also facilitated retention of expertise and repurposing of compounds for novel targets.Footnote ²⁴ The described combination of comparatively open knowledge flows, long-term expertise, and integrated infrastructures also meant that innovation often occurred in unforeseen non-linear ways, as in the case of the 1939 decision to turn penicillin—an unpatented compound discovered eleven years earlier—into a therapy.Footnote ²⁵

Molecular and use-based innovation involving the discovery of new compounds and the repurposing of existing compounds for novel targets continued after 1945.Footnote ²⁶ Far from engendering gloom about the future of antibiotics, Christoph Gradmann has shown how rising AMR was seen as an “exciting” development in an anti-infective market that was already considered saturated during the 1950s.Footnote ²⁷ From the 1960s, many R&D departments thus focused on modifying particularly effective compounds to target evolving resistance mechanisms.Footnote ²⁸ It would, however, be wrong to believe that the post-war surge of antibiotic compounds and applications implied a standardised innovation model or concepts of a wider “pipeline” of antimicrobial R&D. Instead, histories of companies such as Glaxo and Eli Lilly (first-generation cephalosporins), LEO Pharma (fusidic acid), or Bayer (ciprofloxacin) reveal often haphazard and disconnected approaches to innovation.Footnote ²⁹

Addressing the needs of low- and middle-income countries (LMICs) was not a major priority for most R&D centres. In contrast to the rapid saturation of HIC markets, most LMICs’ experience of the “Golden Era” of antibiotic innovation was characterised by trickle-down access to new compounds.Footnote ³⁰ With few exceptions, such as post-war UN attempts to spread penicillin production, access depended not only on one’s ability to pay for novel compounds, but also on Cold War geopolitical alignment.Footnote ³¹ Historical research on West and East-African health systems shows that early twentieth-century arsenic-based treatments remained common well into the late 1960s.Footnote ³² Sustained supplies of newer and safer biological antibiotics were often limited to disease-specific international health campaigns.Footnote ³³ Starting in the 1970s, HIC companies increasingly tried to develop LMICs into markets for now off-patent—occasionally more toxic—human and veterinary compounds, as well as combination therapies that had been abandoned in the West.Footnote ³⁴ In LMICs themselves, shortages drove scarcity-based innovation, such as alternative fermentation media in China, East-Germany’s development of nourseothricin to replace tetracyclines, Spain’s development and distribution of fosfomycin to South America, and the Yugoslavian pharmaceutical company Pliva’s discovery and out-licensing of azithromycin to Pfizer.Footnote ³⁵

The integrated innovation ecosystems underlying the “Golden Era” started dissolving around 1980. Declining attention to microbial infections led to funding shortfalls and the closure of public anti-infective innovation centres.Footnote ³⁶ Companies also began divesting from antibiotic innovation due to perceived market saturation and new managerial doctrines focusing on maximising shareholder value, streamlining product portfolios, and concentrating on cancer and chronic diseases.Footnote ³⁷ Having mostly abandoned large-scale screening, remaining companies saw little return from genetic and high-throughput analyses of pathogens and compound libraries.Footnote ³⁸ Meanwhile, increased emphasis on protecting proprietary knowledge in industry and academia slowed circulation and adaptation of compounds and expertise.Footnote ³⁹ Knowledge cycling was also constrained due to growing distances between R&D and production sites, which reduced interactions between researchers and engineers.Footnote ⁴⁰ Although structural adjustment programmes disrupted many LMIC pharmaceutical industries, Good Manufacturing Practice (GMP) certification intensified outsourcing of manufacturing to regions with cheaper labour costs and weaker environmental regulations.Footnote ⁴¹

By the 1990s, the diverse post-war innovation and manufacturing ecosystem had fragmented along geographic and proprietary silos. Whereas global manufacturing and packaging were rapidly shifting to China and India, R&D remained concentrated in HICs, but was also transforming due to intensifying mergers of traditional developers and the growing prominence of SMEs.Footnote ⁴² The 1995 Agreement on Trade-Related Aspects of Intellectual Property Rights further narrowed definitions of what counted as valuable innovation by universalising product patents associated with molecular novelty.Footnote ⁴³

3. Part Two: a pipeline born empty

Consequences for antibiotic innovation were profound, with R&D activity falling from the 1980s onwards.Footnote ⁴⁴ However, widespread public and scientific alarm about declining antimicrobial innovation was relatively slow to emerge. Alarm about failing bug-drug combinations had been voiced since the 1940s, and both antibiotic activist Stuart Levy and the 1992 Institute of Medicine report on Emerging Infections reacted to slowing innovation by noting that AMR required “replacement drugs to be in the ‘pipeline’.”Footnote ⁴⁵ However, few early commentators conceived of antibiotic innovation as an interconnected—let alone empty—pipeline. Indeed, our thematic trends analysis shows that the pipeline metaphor only gradually emerged as an overarching deficit imaginary for antibiotic R&D from the mid-1990s onwards (Figure 2).Footnote ⁴⁶

Figure 2. References to “antibiotic pipeline” in sorted articles (n = 210).

The rise of the “empty pipeline” metaphor coincided with surging international attention for AMR as a problem exceeding individual “superbugs.”Footnote ⁴⁷ However, the implied deficit imaginary was slow to gather steam. In the infectious disease community, many contemporaries prioritised stewardship interventions over branding AMR as an innovation challenge.Footnote ⁴⁸ In 1994, John La Montagne, director of microbiology and infectious diseases at the U.S. National Institute for Allergy and Infectious Diseases, noted that industry was still “producing drugs at a reasonable clip.”Footnote ⁴⁹ Four years later, at the annual Infectious Diseases Society of America (IDSA) meeting, physician Dennis Maki hyperbolically compared calls to prioritise innovation over stewardship to “providing finer brandy to your alcoholic patients.”Footnote ⁵⁰

Who was behind the push to reframe AMR governance around the “empty pipeline” metaphor? Authorship analysis of 92 sorted Web of Science articles on antibiotic innovation between 1995 and 2005 shows that 77% were from HIC universities (Figure 3). LMIC voices (1%) were remarkably absent despite expanding antibiotic industries (e.g., China) and vocal contemporaneous participation in HIV debates.Footnote ⁵¹ The global divide in reporting indicates that, for most of the world, antibiotic access rather than molecular innovation remained the primary challenge. Meanwhile, industry authors (19%) played a notable role in warning that “the pipeline may be drying up” and calling for incentives-based market fixes for innovation, such as extended patents and reduced trials requirements, “to stem the tide of companies leaving the area.”Footnote ⁵²

Figure 3. Author affiliations in sorted antibiotic innovation articles (1995–2005) (n = 92).

Industry advocacy for public interventions was not linked to an acute crisis of the antibiotic market. Between 1998 and 2009, global antibiotic revenues continued to grow from approximately $22 billion to over $25 billion.Footnote ⁵³ However, antibiotic profitability compared increasingly unfavourably with other fields such as cancer therapy. Whereas antibiotics had long ranked amongst the most lucrative medicines, a 2002–04 list of U.S. pharmaceutical bestsellers featured none.Footnote ⁵⁴ Seeking to maximise return on investment, many R&D companies with lucrative existing antibiotic portfolios were completely abandoning the field. Historical analysis of Bayer, a long-standing antibiotic innovator, reveals that in-house R&D had been dismantled by 2002.Footnote ⁵⁵ Similar divestments occurred at Eli Lilly, Aventis, Bristol Myers Squibb, Proctor & Gamble, and Lederle.Footnote ⁵⁶ Roche provoked particular attention when it announced that it would not pursue R&D on a successor to Rocephin (ceftriaxone) despite the drug generating well over $1 billion per year.Footnote ⁵⁷ Even within companies continuing antibacterial research, scientists were said to face “a daily battle for resources compared with colleagues working in other therapeutic areas.”Footnote ⁵⁸

Why then call for public interventions in a market that was neither in crisis nor of particular strategic interest? One reason may be that mobilising the spectre of an “empty pipeline” enabled industry to push back against wider problems facing contemporary R&D. In the context of stricter clinical trial requirements, expiring blockbuster patents, and public backlash against high pricing for patent-protected antiretrovirals, the empty pipeline metaphor offered a way to reframe the emergent health threat of AMR as a market problem.Footnote ⁵⁹ This pipeline could be “fixed” via deregulation, incentives, and exclusivity without a wider systems critique of proprietary HIC drug innovation.

The market and innovation reframing of AMR aligned with broader trends within the field of Global Health. During the 1990s, the World Bank and new non-governmental donors, such as the Bill and Melinda Gates Foundation, challenged the traditional leadership of the World Health Organization (WHO) within international health.Footnote ⁶⁰ WHO Director-General Gro Harlem Brundtland (1998–2003) tried to regain influence by co-designing policies with donors and intensifying industry engagement.Footnote ⁶¹ In the case of AMR, HIC stewardship and biosecurity concerns engendered WHO emphasis on “vertical” technology-based solutions, including innovation.Footnote ⁶² With strong U.S., U.K., and Japanese input, the WHO published the first Global Strategy for Containment of Antimicrobial Resistance in September 2001, which called for R&D incentives, fast-track authorisation, orphan drug schemes, time-limited exclusivity, aligned IP rights, and industry partnerships.Footnote ⁶³ Although the strategy’s launch was overshadowed by the attacks of 9/11, WHO endorsement of an IP- and incentives-based response to AMR two months before the Doha Declaration on compulsory medicines licensing and weeks before Bayer’s high-profile Amerithrax defence of ciprofloxacin patents was a consequential political decision.Footnote ⁶⁴ It also helped embed the empty pipeline narrative of market failure within the wider AMR community.

4. Part Three: volatile values

Although funders such as the US National Institutes of Health (NIH) increased grants for early-stage research and influential IDSA reports called for market interventions, actual public investment in antibiotic innovation was slow to take off.Footnote ⁶⁵ Remaining companies profited from mounting concerns about “superbugs” and an abundance of older compounds, and knowledgeable researchers created by exiting developers. Meanwhile, innovation finance and underlying business models became dominated by financial markets and speculation on compounds’ downstream earning potential.

In a market where off-patent generic antibiotics often constrained pricing, MRSA presented an attractive target for commercial innovation. Emerging in the 1960s, MRSA gained notoriety as a hospital “superbug” in HICs during the 1990s.Footnote ⁶⁶ As a Gram-positive single-walled bacterium, MRSA was easier to target than double-walled Gram-negative “superbugs,” such as Pseudomonas or Acinetobacter spp.Footnote ⁶⁷ Meanwhile, MRSA’s healthcare-associated infection (HAI) status, complicating treatment of vulnerable patients, made it an ideal target for value-based pricing. Value-based pricing was rooted in the post-1970s rise of health technology assessment (HTA) and aligned shifts of pharmaceutical marketing, which priced drugs according to their projected value to patients or healthcare systems with no necessary relation to R&D or manufacturing costs.Footnote ⁶⁸ Subject to shifting norms, HTA and value-based pricing drove a surge of industry interest in high-value targets such as cancer, but also transformed anti-infective pricing.Footnote ⁶⁹

Cubist and Pfizer were two companies that successfully adapted abandoned compounds for emerging high-value bacterial targets. Pfizer’s Zyvox (linezolid) is an oxazolidinone analogue. Dupont synthesised oxazolidinones while working on agrochemicals in 1978 and accidentally discovered their antibacterial effects before abandoning R&D due to toxicity concerns in the late 1980s.Footnote ⁷⁰ 1990s MRSA epidemiology prompted Upjohn Laboratories (latterly Pharmacia) to resurrect development and synthesise linezolid before Pfizer acquired the company.Footnote ⁷¹ Cubicin (daptomycin) is a cyclic lipopeptide discovered by Eli Lilly in 1984, which was also abandoned due to toxicity concerns, until Cubist adapted the dosage regimen, aided by former Lilly scientists.Footnote ⁷² Turning both compounds into lucrative medicines depended on robust patent protection and demonstrating value via cost savings relative to older drugs such as vancomycin, which had emerged as an important MRSA treatment in the 1980s and was now off-patent.Footnote ⁷³ Cubist’s adapted daptomycin regimen was approved in 2002. Following the 2003 FDA licensing, Cubicin pricing for a 10-day treatment regime of 6 mg/kg per dose was approximately $1,500.Footnote ⁷⁴ The price reflected projected savings from shorter MRSA hospitalisations versus generic vancomycin-based treatment, and was calibrated to compete with Pfizer’s Linezolid (licensed 2000).Footnote ⁷⁵ Noting the high number of patients with staph-based endocarditis (infection of the lining of heart valves and chambers) and bloodstream infections, Cubist expected to “very easily get to a billion-dollar opportunity in the United States.”Footnote ⁷⁶ Pfizer similarly justified Zyvox’s price by referencing its superior performance and savings versus vancomycin.Footnote ⁷⁷

Blockbuster profits (at the time defined as approximately >$1 billion) rewarded acquisitions-based development. Annual worldwide revenue from Zyvox totalled $1.1 billion in 2008 (with $4.4 billion total sales 2000–08).Footnote ⁷⁸ In the case of Cubist, Cubicin revenues soared from approximately $40–50 million projected in 2004 to $966.7 million in 2013 before the company was acquired by Merck for $9.5 billion in 2014.Footnote ⁷⁹ Legal challenges, however, revealed the fragility of the value and IP claims underlying these profits. In 2009, Pfizer paid a record-breaking $2.3 billion settlement after a whistle-blower lawsuit alleged fraudulent U.S. marketing of multiple drugs, including off-label marketing of Zyvox with claims that it was superior to vancomycin.Footnote ⁸⁰ Merck’s hopes for sustained blockbuster income were similarly dented when a U.S. court voided four of five Cubicin patents in 2015, resulting in a major fall in daptomycin revenue.Footnote ⁸¹ On the supply side, companies’ shift to acquisitions-based development also failed to reinvigorate large-scale multi-pronged antibiotic R&D. Following Merck’s takeover, all of Cubist’s 120 scientists were let go, prompting Science to criticise large companies for “treating the scientists as annoying husks keeping you from sinking your corporate teeth into the delicious acquisition of corn.”Footnote ⁸² Critics also highlighted that remaining commercial R&D nearly exclusively targeted high-value targets in HICs, that value-based pricing could make novel antibiotics unaffordable, and that less profitable targets in LMICs and double-walled Gram-negative targets were still being neglected.Footnote ⁸³

The ability of the for-profit ecosystem to provide several first-in-class Gram-positive treatments nonetheless strengthened academic and political assessments that science was not the main barrier to antibiotic innovation and that the market could fix itself, given suitable rewards.Footnote ⁸⁴ Meanwhile, the departure of additional large companies like Bristol Myers Squibb meant that the antibiotic innovation ecosystem gradually became dominated by SMEs. Speaking in 2008, former Glaxo Head of Research Sir Mark Richmond noted that SMEs offered traditional antimicrobial developers the opportunity to outsource high-risk innovation by “using the originality and excitement of the people that run those companies to find new molecules, at which point [large companies] will buy them.”Footnote ⁸⁵ For Richmond, SMEs signalled an “evolution of ‘big Pharma,’ which is the setting up of their out-of-house research activities, but under control, …”Footnote ⁸⁶ One year later, AstraZeneca’s former Global Project Director Keith J. Williams similarly observed that SMEs were filling niches left by larger companies, but that loss of expertise in end-to-end development and “raising capital and finding major companies prepared to invest in clinical development will be a challenge.”Footnote ⁸⁷

Williams’ and Richmond’s comments point to the high-risk, high-return business model at the heart of SME-based innovation.Footnote ⁸⁸ As described by social scientists Kaushik Sunder Rajan and Victor Roy, SME-led biotech R&D often involves passing ownership over a small number of knowledge assets from one investor to another. Some companies retain control of IP and raise capital via successive investment rounds in the hope of out-licensing or selling assets to a big pharmaceutical company. To successfully raise capital and reach an eventual buyout, SMEs need to continuously project assets’ increasing earning potential for investors. Patents and market exclusivity do not merely function to recoup R&D outlays via eventual product sales, but as tradable monetised assets whose downstream rent and value evolution is speculated upon. In other words, speculation on a treatment’s future worth is what defines its value in the present.Footnote ⁸⁹ SME attrition in this speculative—or assetised—innovation ecology is high.Footnote ⁹⁰ Promising compounds can be lost behind IP walls if companies fail, and there is a temptation to (over)emphasise compounds’ projected earnings in line with investor expectations for continuous value evolution (see Part Five).

5. Part Four: a directed ecosystem

Stabilising and reinvigorating the increasingly volatile antibiotic innovation ecosystem became subject to a profound governance shift towards public interventionism from 2010. This shift coincided with surging Gram-negative resistance and a globalisation of AMR discourse, which no longer saw AMR as a problem that could be mitigated via HIC stewardship alone.Footnote ⁹¹ In line with the 2015 WHO Global Action Plan’s renewed call for “incentives for innovation,” the empty pipeline metaphor functioned as the structuring principle for an increasingly directed “public–private” innovation ecology, whose various initiatives were subject to geopolitical contestation.Footnote ⁹²

Nearly all early innovation initiatives were based on PPPs and targeted pre-registration R&D stages (Figure 8 in Supplementary Appendix Page 2). Although public–private sector collaboration had long been common, the 1990s saw PPPs emerge as a formal policy tool for private sector delivery of public infrastructure and services.Footnote ⁹³ In Global Health, PPPs gained traction due to inefficiency allegations against international organisations, successes in vaccine, tuberculosis, and malaria R&D, and PPPs’ popularity amongst HIC decision-makers and funding bodies, which were often headed by former industry executives or originated in pharmaceutical companies.Footnote ⁹⁴

Antibiotic PPPs emerged on both sides of the Atlantic (Timeline Figure 7 in Supplementary Appendix Page 1). While the TB Alliance (est. 2000) was the first antibiotic PPP, the European Union’s (EU) Innovative Medicines Initiative injected significant capital into broader antibiotic research programmes and PPPs between 2008 and 2021.Footnote ⁹⁵ In the United States, the Obama administration also favoured PPPs. From 2010, the Biomedical Advanced Research Defence Authority (BARDA), which had initially focused on stockpiling critical antibiotics, also started subsidising research on new antibiotics relevant to national security.Footnote ⁹⁶ Funded by BARDA, with contributions from the Wellcome Trust, the Combating Antibiotic Resistant Bacteria Biopharmaceutical Accelerator (CARB-X) was established to bridge gaps between basic research and clinical development in 2016. CARB-X provides developers with non-dilutive investment and advice from former industry experts to “push” compounds through early development stages.Footnote ⁹⁷ Emulating regulations for neglected diseases, U.S. officials also introduced the Limited Population Pathway for Antibacterial and Antifungal Drugs to reduce regulatory burdens and speed licensing in 2016.Footnote ⁹⁸

Having previously resisted state intervention, pharmaceutical companies formed consortia to steer PPP debates and lobby for further public investment. Distinct platforms emerged for SMEs and for larger companies (Figure 8 in Supplementary Appendix Page 1), with the BEAM alliance representing EU SME interests and the AMR Industry Alliance representing a coalition of larger and smaller commercial enterprises.Footnote ⁹⁹ New joint funding initiatives resulted from these efforts. In 2020, the industry- and EU-funded AMR Action Fund set out to bring two to four antibiotics to market by 2030 and “facilitate needed long-term policy solutions.”Footnote ¹⁰⁰

Surging public investment occurred amidst political tensions about financing models. As highlighted by participants of our 2024 witness seminar, senior WHO officials had seen AMR as an opportunity to breathe new life into plans for a proposed global R&D fund and generate support for a PDP for antibiotics.Footnote ¹⁰¹ Although frequently used interchangeably with PPPs, PDPs emerged around 2000 as a new form of non-governmental R&D. PDPs function as non-profit intermediary hubs to develop and widen access to health products via targeted service partnerships with industry.Footnote ¹⁰² Despite finding a receptive audience amongst officials planning Germany’s 2014–2015 G7 Presidency, the WHO plan failed to make it onto the official policy agenda due to competing national innovation concepts.Footnote ¹⁰³

Britain, in particular, advocated a more incentives-focused approach. Under Prime Minister David Cameron, AMR had emerged as a foreign policy priority in which the United Kingdom could assert leadership. In addition to financing stewardship and surveillance via the Fleming Fund (2015-2025), British efforts focused on “fixing” the antibiotic pipeline. This emphasis was supported by influential AMR “policy entrepreneurs,” Chief Medical Officer (2010–19) Sally Davies, and Wellcome Trust Director (2013–23) Jeremy Farrar.Footnote ¹⁰⁴ In her 2011 annual report, Davies’ described the dearth of novel antibiotics as a “market failure” to be overcome by public investment commensurate to societal value: “If we are to secure a ‘pipeline’ of new antimicrobial drugs for the future, then we must align the private and societal risks, and the costs and benefits of research and development of these agents.”Footnote ¹⁰⁵ In 2014, Davies prompted Cameron to commission Goldman Sachs economist Jim O’Neill to calculate AMR’s economic burdens, propose solutions, and galvanise action. Based in Wellcome Trust headquarters, the final O’Neill Review was submitted in 2016 prior to the first UN high-level meeting on AMR.Footnote ¹⁰⁶

At our 2024 witness seminar, key stakeholders, including senior former WHO officials, noted that the British desire to maximise attention for the O’Neill Review was one reason why no AMR R&D fund made it onto Germany’s G7 policy agenda.Footnote ¹⁰⁷ A second reason highlighted by several witnesses was a preference from U.K.—and U.S.—actors for fiscal investment incentives, like those proposed by the O’Neill Review, over more direct market interventions.Footnote ¹⁰⁸ This was also true for influential funders, such as the U.K.-based Wellcome Trust, described by witnesses as “industry-friendly,” who had opposed WHO R&D Treaty plans and instead favoured fiscal incentives for industry to take action.Footnote ¹⁰⁹ Witnesses’ memories align with political economists’ contrast of an Anglo-American shareholder value-based liberal market economy versus a more interventionist German coordinated market economy model based on investment in corporatist partnerships.Footnote ¹¹⁰ Reflecting on differences around 2015, witnesses emphasised that “political ownership matters”: despite there being “two super-committed governments… the two most committed European governments could not agree on what they would actually do.”Footnote ¹¹¹

In 2016, Anglo-American funders and governments thus invested in CARB-X’s PPP model with BARDA cumulatively providing $200 million, the Wellcome Trust $155 million, and U.K. agencies £21.1 million by 2023. In contrast, Germany became the primary early sponsor of the WHO-founded PDP, the Global Antibiotic Resistance Development Partnership (GARDP), providing €116.8 million (60% of cumulative funding by 2023) (Figure 4, and Figure 9 in Supplementary Appendix Page 2). While official Anglo-German push-funding differences gradually subsided, Germany remains GARDP’s largest sponsor. Meanwhile, the U.S. and philanthropic (Wellcome Trust, Gates Foundation, and Novo Nordisk Foundation) foundations have continued to nearly exclusively support CARBX (Figure 4).

Figure 4. GARDP and CARB-X cumulative funding sources (2016–23, € million).

Source: GARDP and CARB-X 2023 reports. CARB-X funding reported in USD/GBP converted to Euros using the 2023 average exchange rate from http://www.ofx.com.

Note: Chart displays funds received since inception and commitments announced by 2023, including those made for future periods up to 2027. Not shown is a potential additional commitment to CARB-X from BARDA of $300 million 2022–32. More information is available at https://gardp.org/publications/annual-report-2023/; https://carb-x.org/wp-content/uploads/2024/07/CarbX_AnnualReport_FINAL-compressed.pdf; https://carb-x.org/partners/funding-partners/; and https://carb-x.org/wp-content/uploads/2023/05/2023-0502-CARBX-annual-report-combined-sm.pdf (accessed April 25, 2025).

Disagreement over funding models also arose regarding accelerating calls for market-shaping post-registration “pull”-incentives (Figure 5).Footnote ¹¹² In 2012, the United States introduced the Generating Antibiotic Incentives Now (GAIN) Act, which extended nonpatent market exclusivity for Qualified Infectious Disease Products by five years. Although the GAIN Act prompted some SMEs to reinvestigate off-patent drugs, its perceived failure to regalvanise R&D by larger companies prompted calls for more ambitious market entry rewards (MERs).Footnote ¹¹³ MERs advocacy was often linked to the concept of “delinkage” (Figure 5). Delinkage had initially evolved around 2000 as a way to improve LMIC drug access by delinking drug prices in poorer countries from companies’ R&D outlays and pricing in HICs.Footnote ¹¹⁴ In the AMR space, delinkage became a way to reconcile formerly distinct stewardship and industry profitability concerns (Part Two) by delinking profits from volume of sales and extending the concept of value-based pricing (Part Three).Footnote ¹¹⁵ Health systems would try to simultaneously stimulate innovation and avoid overmarketing by using MERs or subscriptions to reimburse patent holders for novel antimicrobials’ projected societal worth rather than via traditional product sales.Footnote ¹¹⁶

Figure 5. Mentions of pull (n=114), push (n=58), and delinkage (n=51) in the coded innovation literature (2000–20). Some articles mention multiple intervention categories.

The concept of “pulling” compounds through the “empty pipeline” via incentives such as MERs again proved most popular in the United Kingdom. Between 2014 and 2015, London’s Royal Institute of International Affairs (Chatham House) proposed a delinked business model for antibiotics.Footnote ¹¹⁷ Following similar MERs proposals in the 2016 O’Neill Review, then Wellcome Trust Director Jeremy Farrar highlighted the need for “finding creative new models to stabilise the antibiotics market and stimulate private-sector innovation without exposing public funders to all the risk.”Footnote ¹¹⁸ In the same year, the United Kingdom embarked on a policy experiment, which paid industry for access to, rather than use of, compounds for very drug-resistant infections.Footnote ¹¹⁹ U.K. officials expanded their subscription scheme in 2024 by committing to pay fees of up to £1.9 billion over sixteen years for new antibiotic drugs.Footnote ¹²⁰ Since then, significant advocacy has focused on convincing other countries to develop additional pull incentives with frameworks estimating a “fair share” contribution from HICs.Footnote ¹²¹ In contrast to the fully delinked U.K. model, pull proposals in the EU and United States are based on blended public–private contributions and conventional market-based structures.Footnote ¹²² In the EU, the introduction of Transferable Exclusivity Vouchers would see companies rewarded for the development of novel antimicrobials by market exclusivity extensions transferable to lucrative drugs (Transferable Exclusivity Extensions).Footnote ¹²³ These could either be applied to an in-house drug or sold to another company. In the United States, the PASTEUR Act would lead to subscription contracts with individual companies.Footnote ¹²⁴ However, at the time of writing, both initiatives remain stalled in the legislature.

Varying innovation cultures may again be at play. According to a witness at the 2024 seminar, U.K. funders were convinced “that if they only lobby enough for pull incentives the industry is going to fix the problem,” whereas “the Germans wanted to put public money into something which is real, and they can touch.”Footnote ¹²⁵ Critics have also highlighted that pull incentives by themselves are no magic bullet. Focusing on the PASTEUR Act, the Médecins Sans Frontières Access initiative warned that poorly designed HIC pull-incentives can detrimentally impact drug access in LMICs by redirecting constrained global drug supplies and raising prices in other markets.Footnote ¹²⁶ In the context of HICs, proposed transferable exclusivity extensions may lead to European health systems overpaying for antibiotic innovation via higher reimbursement for other important drugs.Footnote ¹²⁷ In the US, the PASTEUR Act’s intended cap on antibiotic sales may impede domestic access by raising prices for patients not covered by Federal procurement mechanisms.Footnote ¹²⁸ Scholars have also warned that paying private companies for access rather than use of drugs is vulnerable to political challenges.Footnote ¹²⁹ Others have highlighted that investing in open innovation and public antibiotic development may be cheaper than financing market-shaping incentives for proprietary innovation.Footnote ¹³⁰ Indeed, a recent study on the German pull mechanism found that Germany would either need to increase volumes of antibiotics covered or increase prices by as much as 3.3 times to provide a de-linked revenue model for proprietary developers.Footnote ¹³¹

Despite the outlined criticism, the past fifteen years have nonetheless seen remarkable HIC consensus on the need to “fix” the “broken antibiotic market.” While some advocates claim that existing incentives only provide a fraction of funding needed, over $2 billion of public and philanthropic investment already targeted antibiotic innovation between 2017 and 2023.Footnote ¹³² In parallel, a heterogeneous public funding system has evolved to cover nearly all aspects of the “empty pipeline” from pre-clinical development to post-registration income. The pipeline metaphor has also enabled funders and developers to stake out remits. An informal division of labour has emerged with CARB-X funding earlier stages of development, GARDP de-risking more advanced development and access projects, and WHO trying to direct R&D with its priority pathogens list (est. 2017) and pipeline reviews (est. 2018).Footnote ¹³³ So far, however, pipeline fixes have not restored self-sustained commercial innovation but have dramatically expanded public intervention and created a “transnational bureaucracy” of non-governmental entities disbursing significant public and philanthropic funds.Footnote ¹³⁴ Meanwhile, core drivers of R&D fragility and precarious antibiotic access remain unresolved.Footnote ¹³⁵

6. Part Five: divergent values

The most recent 2024 WHO review concludes that antibiotic innovation remains insufficient to meet public health needs.Footnote ¹³⁶ Meanwhile, the R&D exodus of large companies, such as Novartis, AstraZeneca, Sanofi, and Johnson & Johnson, has continued.Footnote ¹³⁷ The remaining large companies have licensed a limited number of compounds, such as Avycaz (ceftazidime/avibactam, Pfizer, 2015), Fetroja (cefdicerol, Shionogi, 2019), Recarbrio (imipenem/cilastatin/relebactam, Merck, 2019), and Gepotidacin (GSK, 2025).Footnote ¹³⁸ However, molecular innovation remains limited, with 10 of 13 U.S.-licensed traditional antibiotics between 2017 and 2023 belonging to classes for which AMR is known. Rather than fixing the “broken” pipeline, fifteen years of end-to-end incentives and subsidies have mostly stabilised an innovation ecosystem dominated by volatile “micro” or “small” companies with limited molecular portfolios.Footnote ¹³⁹ Worryingly, every SME with approval for its compounds has suffered bankruptcy or unfavourable market exit since 2010.Footnote ¹⁴⁰

The high-profile 2019 bankruptcy of SME Achaogen illustrates that proposals to incentivise innovation, for example, by providing one-off payments at registration, may be inadequate to support sustainable patient access or survival of SMEs critical to novel antibiotic development. Achaogen adapted an aminoglycoside compound, sisomicin, discovered in 1970 by Bayer. In 2009, the company began trials of its sisomicin analogue, Zemdri (plazomicin), against complicated urinary tract infections (cUTIs) before achieving fast-track licensing in 2018 as a “breakthrough therapy” under the GAIN Act. Achaogen mobilised push investment, largely from BARDA and U.S. biosecurity agencies. Private investor interest was garnered with a 2014 forecast that plazomicin would generate $63 million per annum in the United States and additional EU royalties of $3 million.Footnote ¹⁴¹ Estimating approximately $15,000 incremental cost per U.S. hospital patient of antibiotic-resistant HAIs, the market entry price was $4,955 per treatment course.Footnote ¹⁴² This proved overambitious in a market where existing alternative treatments cost between $21 (polymyxin B) and $56 (colistin) per day.Footnote ¹⁴³ Failure to gain FDA approval for Carbapenem-resistant Enterobacterales further constrained earning potential. Generating only $0.8 million of sales in 2018, Achaogen filed for bankruptcy in 2019, wiping out almost $800 million of investment.Footnote ¹⁴⁴

Plazomicin was subsequently criticised for providing limited clinical benefits over existing cUTI treatments and, to date, has only been reintroduced to India.Footnote ¹⁴⁵ However, during a time of strong support for push/pull incentives, this did not prevent Achaogen from attracting significant public and private investment at every R&D stage. In a 2020 interview, Achaogen’s Ryan Cirz noted: “We got everything right, and it still don’t work.”Footnote ¹⁴⁶ This was correct—at least regarding SMEs’ assetised business model.

As the case of Achaogen and other SME failures shows, the income new antibiotics can realistically generate had become incommensurable with the speculative dynamics shaping a compound’s supposed commercial worth and earning potential.Footnote ¹⁴⁷ According to social scientists, antibiotics constitute an essential infrastructure propping up the basic functioning of global health and food production systems.Footnote ¹⁴⁸ Although improved countermeasures, such as infection control and stewardship, will hopefully reduce antibiotic dependency, continuous access to affordable and effective antibiotics will remain a prerequisite for human and animal health.Footnote ¹⁴⁹ Similar to other essential services, such as water or electricity, where value-based pricing is capped to ensure access, there is also a cap on what societies can afford to pay for the perpetual and equitable renewal of antibiotic infrastructures in the face of AMR.Footnote ¹⁵⁰ This cap poses an ultimately unsolvable problem for an assetised mode of R&D dependent on investors and companies seeking open-ended earning potential from anti-infective innovation. It also lends credence to calls by social scientists and development economists to treat essential medicines such as antibiotics as a commons requiring access and reimbursement mechanisms beyond the “classic commodity framework” underpinning HIC drug development.Footnote ¹⁵¹

MERs and delinked value-based reimbursement seemingly provide a solution to the increasing divide between public health and market values. If, as in the case of daptomycin, health systems are willing to pay high prices, then the innovation system works (see Part Three). However, profit expectations change over time. In 2000, a drug expected to generate approximately $500 million per annum would have been considered a worthwhile investment.Footnote ¹⁵² This estimate has long changed: in 2016, revenues of the top cancer drug totalled more than all antimicrobials combined, while two semaglutides, Ozempic and Rybelsius, generated $17.8 billion for Novo Nordisk in 2023.Footnote ¹⁵³ Although prices for novel antimicrobials have increased significantly since 2010, they remain an order of magnitude below other drugs: Teflaro (licensed 2010) and Nuzyra (licensed 2018) only generated between $66.4 and $58 million in the first nine quarters after registration (Figure 6).Footnote ¹⁵⁴

Figure 6. Comparison of antibiotic and non-antibiotic drug revenues by year of registration.

Source: Antibiotic revenues from “Antimicrobial Drugs Market Returns Analysis: Final Report, Office of the Assistant Secretary for Planning and Evaluation (ASPE), 2022,” mAbs and GLP-1 revenues from company reports 2023. Novo Nordisk revenues in Danish Krone converted to USD using the 2023 average exchange rate from http://www.ofx.com.

Note: The bubble size reflects the relative revenues, and the number represents the drug revenues ($millions), followed by the name of the drug.

To remain attractive to executives considering the opportunity cost of allocating capital, MERs need to evolve accordingly. In 2016, the O’Neill Review calculated an $800 million to $1.3 billion partially delinked MER for a new antimicrobial meeting unmet need.Footnote ¹⁵⁵ One year later, another group calculated $1–2.5 billion per drug in a fully delinked scheme spread over multiple years.Footnote ¹⁵⁶ By 2022, another estimate for a fully delinked subscription amounted to $2.2–4.8 billion for ten years per drug.Footnote ¹⁵⁷ Although some of the described rise can be explained by changing push-incentive estimates, mobilising even a fraction of these ever-increasing pull-incentives is challenging amidst failing multilateralism and health systems stress. By failing to address underlying business models, it will also do little to overcome the assetised market dynamics contributing to the fragmentation of the historic R&D ecosystem (Part One).

7. Part Six: access as innovation

Almost twenty-five years after the WHO’s 2001 Strategy on AMR, it moreover remains unclear how much incentives-based attempts to fix the empty antibiotic pipeline will benefit LMICs.Footnote ¹⁵⁸ Similar to other areas of AMR governance, LMIC voices have often been absent from relevant fora (Part Two).Footnote ¹⁵⁹ Instead, global R&D frameworks and incentives-based interventions have been shaped by HIC priorities and values. In line with the pipeline metaphor’s deficit imaginary, this means that innovation has often been framed as creating, rather than widening, access to novel molecular entities. It also means that international policy frameworks have often not adequately engaged with non-HIC R&D models.

The public health importance of improving global access to antimicrobials has regularly been referenced since the first WHO Global Strategy on AMR in 2001.Footnote ¹⁶⁰ Access considerations have also been embedded in PPP and PDP frameworks. CARB-X mandates that each company receiving funding must plan how their product will be accessible.Footnote ¹⁶¹ Meanwhile, a major focus of GARDP’s work lies in making older and new drugs available in LMICs. Announced in 2022, a DNDi-inspired landmark agreement between GARDP and Japanese company Shionogi is designed to make cefiderocol (brand name: Fetroja) available via sublicensed manufacturing in a large range of countries (n = 135) at risk of delayed or no access.Footnote ¹⁶² However, putting manufacturing agreements in place can be challenging, and so far, no other large R&D companies have introduced similar schemes. Meanwhile, the dominance of assetized SME-based R&D (Part Five) means that lucrative HIC markets—particularly the United States—remain the main targets of innovation.Footnote ¹⁶³ According to the most recent WHO Pipeline review, around 90% of preclinical development is occurring in WHO Europe and Americas regions.Footnote ¹⁶⁴ This divide has increased over recent years, with the number of drugs targeting LMICs and low-income countries either decreasing or flatlining despite an overall increase in R&D.Footnote ¹⁶⁵ The ongoing geographic R&D bias (see Part Two) not only drives up price points, but also means that many newer antibiotics are designed without considering LMIC constraints (e.g., requiring cold chains and parenteral administration), which further restricts access to treatment.Footnote ¹⁶⁶ Prioritising molecular innovation over access and supply chain security can also have negative consequences for HIC patients. In 2023, a U.S. Senate hearing revealed that “between 2021 and 2022, drug shortages increased by approximately 30 percent” with stockouts also impacting European patients.Footnote ¹⁶⁷ Supply chain resilience for both old and new antibiotics is likely to be further strained by contemporary trade wars.Footnote ¹⁶⁸

Other countries have prioritised direct state investment over incentives-based innovation. Having established dominance of global active pharmaceutical ingredient production since the 1990s, the Chinese State has used direct investment to implement a series of policy projects to enhance drug discovery.Footnote ¹⁶⁹ In 2008, a National Mega-Project for Innovative Drugs was implemented, supporting over 3,000 development pathways by 2020, including for emerging antibiotic resistant infections and multidrug-resistant tuberculosis (MDR-TB).Footnote ¹⁷⁰ By 2022, Chinese R&D represented 20% of the global innovation pipeline, ranked second only to the United States, with three Chinese companies in the top 25.Footnote ¹⁷¹ Although many compounds are derivatives of existing molecules, 17 new antibiotics are currently at different clinical development stages, primarily focused on WHO priority pathogens and/or MDR-TB.Footnote ¹⁷² In contrast to HICs’ focus on “repairing markets” via fiscal incentives, joint focus on using state policy and investment to steer manufacturing and innovation is making China critical to global innovation and supply, although some scholars speculate that China may eventually need a subscription model to cover return on investment for both generic and R&D companies.Footnote ¹⁷³

The rise of MIC pharmaceutical manufacturing and biotechnology expertise points to the role direct investment, state-owned companies, and long-term national planning can play in creating innovation ecosystems that are less reliant on HIC R&D companies or financial markets. This does not mean that MIC innovation will necessarily translate into more equitable global drug access. Indeed, it is unlikely that any one-size-fits-all R&D model will work across different bio-geographies and socio-economic contexts. Similar to the diversified vaccine R&D and manufacturing called for in the International Pandemic Agreement, some scholars have therefore proposed using current HIC incentives to build antimicrobial innovation and manufacturing hubs in underserved parts of the world.Footnote ¹⁷⁴ Others have suggested diversifying HIC R&D cultures via new trial regulations, open access innovation, and further valorising the repurposing of older off-patent compounds.Footnote ¹⁷⁵ While analysing or adding to this list is beyond the scope of our historical review, enabling local populations to shape R&D priorities seems a logical way of ensuring that access is the actual target of innovation. It also entails abandoning unhelpful one-size-fits-all approaches to drug innovation.

8. Conclusion: beyond the pipeline

As our historical review shows, the accelerating divide between what a compound must be worth to be considered financially worthwhile and its actual public health value points to a flaw at the heart of the “empty pipeline” metaphor: postwar antibiotic innovation ecosystems did not dissolve because global antibiotic markets broke, but because assetized markets broke them.

Following the rupturing of integrated R&D systems during the 1980s, the rise of the pipeline metaphor during the mid-1990s strategically reframed an emergent public health challenge as an incentives problem. The metaphor’s narrow focus on unidirectional molecular innovation and focus on technical solutions suited contemporary Global Health philosophies and industry interests. However, it failed to capture the diverse world of mid-century R&D, which included non-linear R&D flows within and between public and private development sites, as well as distinct innovation cultures beyond HICs. Emphasis on “broken” antibiotic markets also reduced proposed solutions to a question of fiscal stimuli. Doing so glossed over the eroding speculative dynamics of financial markets themselves—with larger firms’ exit increasing reliance on SMEs for innovation and SME dependence on speculative investment for survival.Footnote ¹⁷⁶ Despite disagreeing on finance models, public funders have mostly accepted the empty pipeline metaphor with PPPs and PDPs designed to work within the presumed pipeline rather than rethink it. Meanwhile, delinkage has emerged as a key concept to reconcile stewardship and profitability concerns by expanding value-based reimbursement and satisfying investor demand for downstream earning potential (Figure 10 in Supplementary Appendix Page 3).Footnote ¹⁷⁷

Current HIC incentives will undoubtedly lead to a limited number of new compounds. However, the volume of existing R&D is unlikely to satisfy the global need for sustained and accessible antibiotic innovation.Footnote ¹⁷⁸ Although most PPPs and PDPs incorporate access conditions, the prioritisation of HIC needs by global R&D appears to be increasing rather than decreasing.Footnote ¹⁷⁹ Meanwhile, rising supply chain vulnerabilities for both old and new drugs remain unaddressed.Footnote ¹⁸⁰ As global health funding and multilateralism falter, decision-makers have the chance to rethink and diversify core parts of their innovation and access strategies—including learning from non-HIC systems. A more diverse and equitable innovation era may lie beyond the self-imposed limits of the “empty antibiotic pipeline” imaginary.