Part 1. The Erosion: How American Biotech Lost Its Preeminence By Michelle Longmire, MD
We are on the eve of the greatest era of medical science in human history. Artificial intelligence is compressing target discovery, structure prediction, and molecular design. That compression should unlock the most productive period American biotechnology has ever seen. But science is not medicine. Someone has to translate it. Who will lead that race: the country that invented the biotechnology industry, or a new entrant?
@longmireMD . Agree? Disagree? Better ideas? This topic matters too much for a one way conversation.
Imagine it is a Tuesday in March 2035. You have just been diagnosed with glioblastoma, an aggressive brain cancer. Your oncologist walks you through the standard of care: the Stupp protocol, a regimen that has barely changed in thirty years. Median survival is measured in months.
You ask about clinical trials. There is a drug. It doubles the survival benefit of anything this clinic can offer, with documented remissions in advanced disease. China approved it in early 2033. Tens of thousands of patients there have already received it. Here, it is still in Phase 3. An FDA decision is expected sometime in 2037. Your institution is not a trial site, and your oncologist cannot tell you how to enroll at one that is.
So you ask the obvious question. How does the United States not have this medicine?
I. The Industry Was Born from American Innovation and Capital In 1976, in the San Francisco Bay Area, venture capitalist Robert Swanson and scientist Herbert Boyer founded Genentech and turned a laboratory result into an industry.1 The underlying breakthrough was American as well. Boyer at UCSF and Stanley Cohen at Stanford had developed recombinant DNA technology, splicing genes into bacteria to produce the first recombinant DNA organisms.2 That work opened a route to manufacturing human insulin without animal derived sources3 and launched the biopharmaceutical era.
For half a century thereafter, the United States converted bench science into medicine faster and more ambitiously than any other country. It built the machinery that made the conversion repeatable: venture capital, deep public markets, and a regulatory authority whose approvals carry global weight. The returns compounded across five decades in approvals, in company formation, and in capital markets that recycled successful exits into new science (Figure 1).
FIGURE 1: Fifty Years of Compounding Returns on American Invention Approximate decade totals, 1976 to 2025, rounded from published annual data. Approvals: FDA novel drug and new biologic approvals.4 IPOs: U.S. biotech initial public offerings.5 2016 to 2025 cross checked against the BioPharma Dive tracker.6
II. Transforming Science into Medicine In fifty years the biotechnology industry has become a major driver of improvements in human health and longevity. Emerging biopharmaceutical companies, which IQVIA defines as those with annual R&D spending under $200 million or prescription sales up to $500 million, originated 65% of the novel active substances launched worldwide from 2020 to 20247 and start about 63% of clinical trials.8 Large pharmaceutical companies increasingly source assets from outside and carry them through approval and commercialization, as Section IV shows. Invention and early risk taking happen in biotech. Big pharma pays for scale and reaps the returns of commercialization.
Biotech does not simply produce incremental improvements. It is where new therapeutic paradigms are born. Genentech made recombinant human insulin manufacturable in 1982, and with Herceptin in 1998 established that a cancer could be treated by its molecular target rather than its organ of origin. Amgen turned recombinant proteins into an industry with Epogen in 1989 and Neupogen in 1991. Medarex and Bristol Myers Squibb brought checkpoint blockade to patients with Yervoy in 2011, and Merck's Keytruda made immunotherapy the backbone of oncology. Pharmasset's sofosbuvir, carried through development by Gilead, made hepatitis C curable in 2013. The first CAR T cell therapies arrived in 2017, from Kite Pharma and from Novartis with University of Pennsylvania science. Vertex addressed the underlying defect in cystic fibrosis with Kalydeco in 2012 and Trikafta in 2019, and with CRISPR Therapeutics delivered the first approved gene editing therapy in 2023.4 Moderna showed in 2020 that a vaccine could be designed within days of a pathogen's sequence being published9 and manufactured at global scale, and more recently that an individualized neoantigen vaccine combined with pembrolizumab can reduce recurrence in resected high risk melanoma.10
Each of these breakthroughs redefined a disease category and the trajectory of every life diagnosed within it. Together with reduced smoking and earlier detection, better treatment is why the age adjusted cancer death rate in the United States fell 34% from its 1991 peak through 2022, which translates to nearly 4.5 million deaths that did not occur.11 HIV, the leading cause of death among Americans aged 25 to 44 in 1994,12 now carries a near normal life expectancy when treated promptly.13
III. American Dominance Is Rapidly Eroding In 2015 the United States originated 48.2% of the world's early stage drug development programs and China originated 8.0%. By 2024 the figures were 37.4% and 32.3%.14 By 2020 China had surpassed the United States in annual registered clinical trial volume.15 In 2021 China's share of global Phase 1 trials passed the U.S. share for the first time, and in 2024 China registered more than 7,100 clinical trials, 39% of the global total.16,17 In 2024 a Chinese company, Jiangsu Hengrui, overtook AstraZeneca as the world's top clinical trial sponsor.18 China's share of first launches of new active substances rose from 25% in 2023 to 34.6% in 2024, against 44.4% for the United States, counting a drug launched in several markets at once in each.19
The pipeline evidence extends beyond volume into modality and complexity. In 2023, investigational new drug applications to China's Center for Drug Evaluation rose 33% over the prior year, and approved applications for innovative therapeutic biologics rose 26% to 695.20 All four bispecific antibody drug conjugates in Phase 3 worldwide are being developed for East Asian markets, three of them for China.21 An NBER working paper, not yet peer reviewed, reaches the same conclusion: this is not a low complexity pipeline pursuing well trodden targets.15
Origination follows expected market access, and Europe shows what happens when access deteriorates. Europe's share of pharmaceutical R&D investment among the major regions has fallen by roughly a quarter over two decades, from 41% in 2001 to 31%, according to EFPIA, the industry's European trade association.22 Access has deteriorated with it. EFPIA's W.A.I.T. indicator documents a mean delay of nearly 600 days between approval and patient access, with median delays ranging from 56 days in Germany to more than three years in Romania. Nearly half of newly approved medicines (49%) are unavailable to patients in the average European country.23 Patients elsewhere receive them.
If the trends observed from 2015 to 2024 continue, early stage origination in 2035 stands at 62% for China and 24% for the United States, a ratio of roughly five to two. This is a linear extrapolation of the published data, not a forecast. It describes a future in which the country that invented the biopharmaceutical industry becomes a secondary market for its output (Figure 2).
FIGURE 2: The Trend Line, Extended Share of global early stage drug development programs. Solid segments are observed values reported by Kang and Ji for 2015 and 2024.14 Dashed segments are the author's linear extrapolation of the observed slopes, shown to make the implication of the current trajectory explicit. This is not a forecast. It assumes no saturation, no policy response, no competitive adjustment, and no change in the underlying cost structures, assumptions that are individually unlikely and collectively the point of this paper.
IV. Pharma Is Shopping in China The most consequential signal is not what China develops but what the rest of the industry now buys. Out licensing deals signed by companies in greater China rose nearly tenfold, from $13.9 billion in 2021 to a record $137.7 billion in 2025.24 Mainland China alone recorded 157 transactions worth $135.7 billion, up from 94 transactions worth $51.9 billion in 2024.25 First quarter 2026 reached $60 billion, up 73% year over year,26 and the first half of 2026 reached about $110 billion across 81 deals.27 By Jefferies' count, as reported by BioWorld, China's share of global out licensing deals rose from about 5% in 2020 to 21% in 2024 and 32% in the first half of 2025 (Figure 3).28
FIGURE 3: Where Large Pharmaceutical Companies Are Sourcing Their Pipelines Bars: out licensing deal value and transaction count for mainland China origin assets (left axis); January to June 2026 is a partial year. No bar is shown for 2020 because a comparable mainland series was not located. Line: China's share of global out licensing deals by transaction count (right axis); the 2025 point covers January to June. Sources differ by bar: 2024 and 2025 are PharmCube NextBiopharm data;25 January to June 2026 is National Medical Products Administration data,27 which may define transactions differently. The share line is Jefferies' count of deals, not value.28
Individual transactions show the scale. GSK licensed Hengrui's PDE3/4 inhibitor for COPD in a collaboration covering up to 12 programs, with $500 million upfront and up to about $12 billion in milestones and option fees, roughly $12.5 billion in total.29 Pfizer licensed 3SBio's oncology candidate SSGJ-707 for $1.25 billion upfront against up to $4.8 billion in milestones.30 Pharmaceutical Executive reports that by 2025, 38% of large pharma licensing deals originated with Chinese partners.31 Morgan Stanley analysts project that drugs from Chinese developers could account for 35% of FDA approvals by 2040.32
V. Root Cause: Velocity and Cost, Not Scientific Discovery What is driving the pivot to China? At face value the answer would seem to be better discovery science. Recent cuts to basic research funding in the United States threaten the American discovery engine, but they are not what moved a third of the world's early stage origination to China over the past decade. The pivot is driven by Chinese innovation in the clinical development machine: the trials themselves, and the regulatory and operational infrastructure that carries a molecule through testing in humans.
Consider the operating parameters. By GlobalData's estimates, a Phase 1 trial in China runs approximately seven months against seventeen or more in the United States. Phase 1 costs run 32% to 52% lower.33 Early discovery to IND cycles in China are now 50% to 70% faster than in the rest of the world (Figure 4).34,17 China has adopted ICH guidelines, so both countries run these studies under the ICH E6 good clinical practice standards required for FDA submission.35,36 The gap is not in the standard of evidence. It is in the speed and cost of producing it.
FIGURE 4: The Mechanism: Velocity and Cost, Not Discovery Median Phase 1 trial duration, months, per GlobalData estimates as reported by Clinical Trials Arena. Associated differentials: Phase 1 trial costs 32% to 52% lower;33 discovery to IND cycle times 50% to 70% faster.34,17
VI. How China Actually Wins China's advantage lies in operations and systems engineering, not scientific superiority. Three mechanisms drive it.
1. Study startup is one obligation, not disconnected phases The intuitive explanation, that Chinese regulators simply approve faster, is wrong. China's implied approval mechanism, established by NMPA Announcement No. 50 in 2018, deems a clinical trial application approved if the Center for Drug Evaluation raises no objection within 60 working days.37 NMPA reports that average review under this mechanism has fallen to about 50 working days, and a 2025 reform adds an optional 30 working day track for eligible innovative drugs.38 An FDA IND takes effect 30 calendar days after receipt.39 On the regulatory clock alone, the United States is the faster jurisdiction.
The American disadvantage lies in what surrounds that clock. The Department of Health and Human Services' 2026 roadmap reports that the average interval between a U.S. sponsor's pre-IND meeting request and IND submission is 380 days, with a range of almost 700. In April 2026 FDA's then commissioner compared that interval with about 220 days in China.40 Institutional review board approval and contract negotiation can add up to thirteen further months before a single patient is enrolled.17 Activating a site for a Phase 1 to 3 trial at an academic medical center or hospital takes a median of 8.12 months (about 247 days), measured from site selection to contract completion, against the National Cancer Institute's recommended target of 90 days.41 The same source reports a median of 4.37 months at independent sites and physician practices. This paper models the academic figure because the trials at issue here, oncology and first in human studies, are commonly run at academic centers, and benchmarking of North American cancer centers finds activation well beyond the 90 day target for most trial types.42 Budget and contract negotiation are among the leading barriers to activation.17
China's structural difference is that approval and initiation are treated as one continuous obligation. An applicant using the 2025 expedited pathway must commit at the time of application to initiating the trial within twelve weeks of approval.38 The American system optimizes the regulatory decision but not what follows. Once approval is granted, the pace is set by the policies of individual institutions and sites, which vary widely and face no mandated timeframe.
2. The enrollment infrastructure was built for research, not retrofitted Enrollment favors China for a structural reason: research there is concentrated in very large medical centers. Trial activity clusters in Grade 3A tertiary hospitals such as Sun Yat-sen University Cancer Center, Fudan University Shanghai Cancer Center, Peking Union Medical College Hospital, and their peers. These institutions combine enormous catchment populations with dedicated research units and direct relationships with principal investigators. The domestic system itself adds speed. Within China, single country trials enrolled about seven times faster per site than multinational trials that included Chinese sites in Phases 1 and 3, a difference GlobalData attributes to a single regulatory and ethics authority.43 That comparison is within China and predates 2018, but it points to the same conclusion as the rest of this section: the advantage lies in the system, not in the patients.
The patient pool is also larger. About 226,000 new lung cancer cases were diagnosed in the United States in 2022,44 compared with about 1.06 million in China the same year.45 Chinese patients are also more often treatment naive, which matters in oncology, where eligibility criteria routinely exclude prior lines of therapy.46
American research infrastructure is dispersed across many more institutions. At the academic centers where trials do concentrate, concurrent protocols often compete for the same specialist referred population, so trials can slow one another. Clinicians are often disincentivized from referring patients to trials, because identification, explanation, and coordination take time for which there is no reimbursement structure.17 The combined effect, in a 2019 analysis, was patient enrollment two to three times faster in China.47
3. Operations cost a fraction as much The cost gap follows from the first two mechanisms and from the price of labor. In an analysis of Phase 3 non-small cell lung cancer trials published in 2019, direct cost per patient ran approximately $69,000 in the United States against roughly $25,000 in China.47 That analysis is also the source of the enrollment estimate above, so both figures rest on a single study. Its data predate the pandemic and are not adjusted for inflation, but recent single country comparisons point the same way: average Phase 1 cost is $3.29 million in China against $5.80 million in the United States, with Phase 2 costs 15% to 30% lower.48
The composition of that gap matters more than its size. GlobalData attributes China's savings to lower site and operational costs, and finds that the savings narrow as trial complexity increases.48 The differential concentrates in the operational lines examined later in this paper: monitoring visits, study coordinator and data management time, site administration, and the per visit overhead charged by institutions. The two countries generate evidence to the same standard and pay radically different prices to coordinate the process of generating it (Figure 5). That pattern is consistent with an edge in execution rather than discovery.
FIGURE 5: How China Wins: Startup, Enrollment, Cost A: modeled calendar days from regulatory application to first patient enrolled, divided into regulatory review and everything that follows. United States: the 30 calendar day IND clock39 plus the 8.12 month median site activation interval,41 which ends at contract completion rather than first enrollment and therefore understates the U.S. total; at the independent site median of 4.37 months, the U.S. total would be about 163 days. China: 60 working days of implied approval (about 84 calendar days before public holidays)37 plus the twelve week initiation commitment of the 2025 expedited pathway;38 under that pathway's own 30 working day review, the total falls to about 126 days. Neither bar includes the pre-application period, which averages a further 380 days in the United States.17 B: new lung cancer cases in 2022.44,45 C: direct cost per patient in Phase 3 non-small cell lung cancer trials, late 2010s data.47
The outsourcing models differ as well. Sponsors in both countries rely heavily on contract research organizations (CROs), but the structures differ. The American outsourcing industry grew from the 1980s and consolidated over the following decades into a small number of global providers. The global CRO market is now estimated at $56 billion to $100 billion, depending on scope.49 Those providers serve pre-revenue ventures largely through full-service contracts. An American sponsor therefore absorbs layered vendor margins on top of higher facility overhead, then incurs governance costs to monitor vendor performance across fragmented sites.
China's ecosystem originated in the 2000s as an offshore service hub, built by WuXi AppTec in discovery and Tigermed in execution. Regulatory reforms after 2015, including ICH accession in 201736 and implied approval of trial applications in 2018,37 reoriented that capacity toward domestic originators and produced a dense, low-cost infrastructure organized around high-volume medical centers with far fewer handoffs.
Three qualifications are necessary. First, the arbitrage is being competed away at the top of the market. Average upfront payments for China-originated assets rose from $52 million in 2022 to $172 million in 2026 as Western buyers bid against one another.31 Second, a cost advantage realized within one jurisdiction does not transfer automatically to a global registration package. Regulators scrutinize single-country data for applicability to their own populations. In 2022, the FDA's Oncologic Drugs Advisory Committee found that a pivotal trial conducted almost entirely in China was insufficient to support U.S. approval of sintilimab.50 The practical response is a multiregional trial with a substantial China component, the design ICH E17 was written to govern, rather than a China-only study.51 Sponsors capture much of the operational savings while still paying for Western sites. That lowers the blended cost without eliminating the American operational premium.
Third, speed has not always come with integrity, and policy risk now runs in both directions. In July 2015 China's regulator ordered sponsors to self-inspect the clinical data behind 1,622 pending applications; by the end of 2016, 1,165 had been voluntarily withdrawn.52 Inspection standards tightened afterward, but the episode is recent. In the United States, the BIOSECURE Act, enacted in December 2025, restricts federal contracting and grants involving designated biotechnology companies of concern.53 The cost advantage described here is real. A Western sponsor must still price these risks.
China treated startup, enrollment, and operations as engineering problems worth solving, and spent a decade solving them. The United States, over the same decade, treated those problems as unavoidable friction.
VII. Time Is the Asset The economic case for a pharmaceutical buyer is more interesting than the price tag. Chinese assets have historically commanded lower upfront payments and smaller total deal values than comparable Western programs, although that discount is narrowing. Price is the visible part of the transaction. It is not the most important driver of the decision. Large pharmaceutical companies are not capital constrained. They are time constrained, and time is what a Chinese asset delivers.
Consider what a buyer is purchasing. In drug development, absolute cost matters far less than cost measured against market upside, and market upside depends heavily on when a medicine reaches patients. A compound's commercial life is bounded by patent expiry and regulatory exclusivity. For small molecules, patent term restoration returns part of the time lost to development, roughly half of the clinical testing period plus the review period, up to five years and within 14 years of post-approval patent life.54 For biologics, the 12 year reference product exclusivity runs from licensure regardless of how long development took.55 A year saved in development therefore does not convert one for one into a year of added exclusivity. It always converts into earlier revenue, and often into longer protected sales, worth far more than the cost of the trial that produced it. Development savings are measured in tens of millions. Time is measured in billions.
Speed compounds a second time through the cost of capital. Drug development is a race for evidence. Each milestone cleared converts uncertainty into information, lowers the risk-adjusted discount applied to the program, and unlocks the capital needed to fund the next phase. An asset that reaches human proof of concept in half the time has not merely spent less. It has become financeable sooner, on better terms, with more protected commercial life ahead of it. In a market where capital is abundant and de-risked assets are scarce, that is the competition that matters. It is decided on operations, not on chemistry.
The consequence for American biotech is not that its science is being outbid. Its clock is. Two companies may hold molecules of identical merit and identical probability of success. The one that can generate the same evidence in half the time is worth more to an acquirer, cheaper to finance, and bought first. The licensing shift is not a verdict on American discovery. It is a verdict on American development.
VIII. The Structural Vulnerability of the Emerging Biotech An emerging biotechnology company is exposed to operational cost in a way a large pharmaceutical company is not, for three compounding reasons. First, it is pre-revenue. Estimates of the capitalized cost of bringing a new medicine to approval, including the cost of failures, range from a median of about $985 million per drug based on public filings56 to a mean of roughly $2.6 billion based on confidential company data.57 The median pivotal trial costs about $19 million, with a long upper tail in which some trials exceed $100 million.58 A single program consumes capital faster than most such companies can raise it.
Second, its financing is milestone gated. Capital arrives in tranches released against data readouts. Time is therefore not merely expensive but structurally convertible into ownership. A quarter lost to operational drag is a quarter of additional dilution, or an earlier raise at a lower valuation than a cleaner readout would have commanded. These companies do not pay their operational costs in cash alone. They pay in equity and in runway.
Third, and most consequentially, the emerging company rents the clinical development machine. Lacking an internal clinical operations organization of meaningful scale, it outsources monitoring, data management, safety, regulatory operations, and project management to CROs. Its constraints are structural: lean teams in which individuals cover work spread across whole departments at a large sponsor; dependence on vendors whose systems the biotech either cannot see into or must stitch together by hand; no headcount leverage, because a fixed runway forecloses hiring one's way out of drag; and no tolerance for delay, because a slipped timeline is an existential threat.
IX. Where the Capital Actually Goes Follow the biotech's dollars through clinical development. In the canonical federal cost driver analysis, the activities that generate evidence, clinical procedures (15% to 22% of trial expenditure) and central laboratory work (4% to 12%), account for roughly 30% of spend. Administrative staff consume 11% to 29%, site monitoring 9% to 14%, and site retention 16% to 22%.59 The remainder, roughly 70%, funds coordination. Source data verification, the practice of sending a person to a site to confirm that values in a database match values on a source record, has been reported to account for up to a quarter of a trial's budget.60 In a post hoc analysis of three Phase 3 trials, complete verification reduced the overall data error rate by only 0.26 percentage points compared with partial verification.61 Data collection and management, site management and retention, recruitment and contracting, and regulatory and document handling account for most of the rest (Figure 6).
FIGURE 6: Where the Clinical Development Budget Goes Author's allocation of a late phase development budget, derived from the cost component ranges reported by Sertkaya et al.59 The point values are illustrative midpoints, not reported figures, and the underlying data predate widespread risk based monitoring. Approximately 70% of spend funds coordination and operations rather than evidence generation.
The decisive observation is not the size of this coordination spend but its content. Very little of it is scientific or medical judgment. The prevailing activity of the clinical workforce is reconciliation: bridging software systems that do not communicate, re-keying values between platforms, chasing and resolving queries, verifying entries against source documents, filing and classifying records, and coordinating handoffs across a vendor stack never designed as a coherent whole. The workforce exists in substantial part to compensate for fragmented tooling. That is not a failure of the workforce. It is a failure of the operating model that employs it.
That is why the competitive gap cannot be closed by depressing domestic wages or centralizing hospital systems. It must be closed by eliminating the manual coordination friction that the current operating model imposes on every domestic trial. That friction is exactly what agentic AI is positioned to eliminate.
X. Two Biotechs, Same Molecule, Different Outcomes Suppose two biotech companies hold similar molecules against the same target, with comparable preclinical packages, and file to begin first in human studies in the same month. One is in Boston, one in Shanghai. Thirteen months later, the Shanghai company has a human efficacy signal. The Boston company is still more than a year from its own.
The direct saving on the Phase 1 study is about $2.5 million, which is close to irrelevant. The consequential number is the roughly $21 million of additional capital the Boston company consumes to arrive at the same result. Its composition is the whole argument of this paper. It is not principally a discount on procedures or assays. It is the cost of paying for an entire organization, its people, its facilities, and its vendors, for about fourteen additional months. The Boston company is not buying more science for its money. It is renting time.
Now consider what a pharmaceutical buyer sees in month thirteen. One asset has human data. The other has a promise. That is not a choice between two prices, and no discount on an upfront payment reverses it. For a medicine reaching $1 billion in peak annual sales, fourteen additional months of protected sales at peak would be worth about $1.1 billion. Patent term restoration may return part of that to a small molecule program, but even if half is returned, the value of the time exceeds the Phase 1 saving by more than two hundred to one. It explains a licensing market that cost discipline alone cannot.
The result does not depend on the choice of American site. If the Boston company activates at an independent site instead, its start-up roughly matches Shanghai's. The remaining gap, about ten months and $16 million, comes from Phase 1 conduct itself (Table 1).
The consequence for the Boston company is measured in ownership. Twenty one million dollars raised before a value inflection point, at a pre proof of concept valuation, is a materially larger share of the company sold to reach the same milestone. Its investors own less of the same molecule, and they own it later. Compounded across a portfolio and a decade, that is what the migration of origination looks like from the inside. None of it reflects a difference in the quality of the science. In this example the science is identical by construction.
TABLE 1: The Same Molecule, Two Development Systems
From Regulatory Application to Human Proof of Concept
From regulatory application to human proof of concept
Boston, academic site
Boston, independent site
Shanghai
Application to first patient enrolled
9.1 months
5.4 months
5.5 months
Phase 1 conduct
17 months
17 months
7 months
Elapsed time to first human efficacy signal
26.1 months
22.4 months
12.5 months
Direct cost of the Phase 1 study
$5.8M
$5.8M
$3.3M
Total capital consumed to reach the milestone
~$42M
~$37M
~$21M
Illustrative model. Boston start-up assumes the 30 day FDA IND clock plus the median site activation interval: 8.12 months at academic medical centers and hospitals, 4.37 months at independent sites.39,41 Both intervals end at contract completion rather than first enrollment, so both Boston columns understate elapsed time. Shanghai start-up assumes about 84 calendar days for implied approval plus the twelve week initiation commitment.37,38 Phase 1 durations and direct costs are GlobalData averages.33,48 Total capital assumes organizational burn of $1.4 million per month for all three plus the direct Phase 1 cost. Equal burn is conservative, because Shanghai operating costs are lower.
XI. Where This Leaves Us American biotechnology has not lost its scientific advantage. It has lost its operational one. The molecules originated here are as good as the molecules originated anywhere, and in many categories they are better. What differs is the machine that carries a molecule into humans. That machine adds roughly a year between pre-IND meeting request and IND, most of another year to activate a site, and it staffs itself with a workforce whose largest single activity is confirming that one system agrees with another.
The machine is expensive in a specific and correctable way. Most of the budget funds coordination rather than evidence, and almost none of that coordination is scientific or medical judgment. The cost compounds rather than adds, because every month the machine consumes is a month of capital burned before a value inflection point and, often, a month of protected commercial life lost at the other end. For an emerging company financed against readouts, that is paid in ownership rather than in cash.
How can American biotechnology regain its edge in translating leading science into world class therapeutics? The answer is feasible today. Operational reform and agentic AI can level the playing field on cost and time, and shift the global competition from bureaucracy and bottlenecks to the competition we want: whose medicine works best.
Part 2 sets out what a clinical development system driven by agentic AI looks like: how the labor model changes line by line, what it costs to reach the same milestone, how the same infrastructure addresses an enrollment problem that capital efficiency alone does not solve, and what regulatory discipline it demands. The question is no longer whether the American development machine can be rebuilt. It is whether it is rebuilt before the trend line in Figure 2 becomes the structure of the market.
About the author. Michelle Longmire, MD, is a physician scientist and the co-founder and CEO of Medable, a clinical trial technology company building agentic AI systems for clinical development. Dr. Longmire is a Stanford trained, board certified dermatologist and the author of numerous peer reviewed publications. Her work spans entrepreneurship, translational science, and clinical research.
References 1. Hughes SS. Genentech: The Beginnings of Biotech. University of Chicago Press; 2011.
2. Cohen SN, Chang ACY, Boyer HW, Helling RB. Construction of biologically functional bacterial plasmids in vitro. Proc Natl Acad Sci USA. 1973;70(11):3240–3244. doi:10.1073/pnas.70.11.3240.
3. Goeddel DV, Kleid DG, Bolivar F, et al. Expression in Escherichia coli of chemically synthesized genes for human insulin. Proc Natl Acad Sci USA. 1979;76(1):106–110. doi:10.1073/pnas.76.1.106.
4. U.S. Food and Drug Administration, Center for Drug Evaluation and Research. Novel drug approvals and biological product approvals, 1976–2025. Silver Spring, MD: FDA.
5. Ritter JR. Initial public offerings: updated statistics. Warrington College of Business, University of Florida; updated 2026.
6. BioPharma Dive. Biotech IPO tracker. Accessed 2026.
7. IQVIA Institute for Human Data Science. Expanding Options for Emerging Biopharma in the US: A Decade of Change. IQVIA Institute; 2025.
8. IQVIA Institute for Human Data Science. Global Trends in R&D 2025. IQVIA Institute; 2025.
9. Corbett KS, Edwards DK, Leist SR, et al. SARS-CoV-2 mRNA vaccine design enabled by prototype pathogen preparedness. Nature. 2020;586(7830):567–571.
10. Weber JS, Carlino MS, Khattak A, et al. Individualised neoantigen therapy mRNA-4157 (V940) plus pembrolizumab versus pembrolizumab monotherapy in resected melanoma (KEYNOTE-942): a randomised, phase 2b study. Lancet. 2024;403(10427):632–644.
11. Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA Cancer J Clin. 2025;75(1):10–45. doi:10.3322/caac.21871.
12. Centers for Disease Control and Prevention. Update: mortality attributable to HIV infection among persons aged 25–44 years, United States, 1994. MMWR Morb Mortal Wkly Rep. 1996;45(6):121–125.
13. Trickey A, May MT, Vehreschild JJ, et al. Survival of HIV-positive patients starting antiretroviral therapy between 1996 and 2013: a collaborative analysis of cohort studies. Lancet HIV. 2017;4(8):e349–e356.
14. Kang S, Ji Y. Geographic shifts in early-stage biopharmaceutical innovation. JAMA. 2026;335(15):1355–1357. doi:10.1001/jama.2026.1962.
15. Barwick PJ, Xia H, Xia T. From free rider to innovator: the rise of China's drug development. NBER Working Paper 34977. National Bureau of Economic Research; 2026. doi:10.3386/w34977.
16. National Security Commission on Emerging Biotechnology. Charting the Future of Biotechnology. Final report. April 2025. https://www.biotech.senate.gov/final-report/. Accessed September 2026.
17. U.S. Department of Health and Human Services. Operation TrialBlazer: HHS Roadmap to Maintaining U.S. Leadership in Early Clinical Research and Development. HHS; June 22, 2026. https://www.hhs.gov/sites/default/files/operation-trialblazer.pdf. Accessed September 2026.
18. Citeline. Clinical Trials Roundup 2025. Citeline; 2025. Reported in: PharmaVoice. China's Jiangsu Hengrui overtakes AstraZeneca as top clinical trial sponsor. 2025.
19. Citeline. Pharma R&D Annual Review Supplement: New Active Substances. 2024 and 2025 editions. Citeline; March 2024 and April 2025. (Drugs launched simultaneously in more than one market are counted in each.)
20. Center for Drug Evaluation, National Medical Products Administration. 2023 China Drug Evaluation Report. CDE; February 2024. English summary: Cisema, April 2024.
21. GlobalData. Four drugs race for first bispecific ADC approval. Pharmaceutical Technology; 2025. https://www.pharmaceutical-technology.com/analyst-comment/four-drugs-first-bispecific-adc-approval/
22. European Federation of Pharmaceutical Industries and Associations; Charles River Associates. Europe's share of global medicines R&D shrinks by a quarter in 20 years. EFPIA; November 2022.
23. European Federation of Pharmaceutical Industries and Associations; IQVIA. EFPIA Patients W.A.I.T. Indicator 2025 Survey. EFPIA; May 2026.
24. Wu K, Silver A. Analysis: China biotech licensing boom to hit record in 2026 as pipeline swells. Reuters. February 13, 2026. (PharmCube data, greater China.)
25. Valencia JS. China's out-licensing deals bear fruit in 2025. DIA Global Forum. March 2026. (PharmCube NextBiopharm data, mainland China.)
26. Ren D. China biotech deals hit record as innovative drugs draw interest of multinationals. South China Morning Post. April 2026. (National Medical Products Administration statement.)
27. Reuters. China innovative drug out-licensing deal value reaches new high in first half of 2026. July 2026. (National Medical Products Administration data via CCTV.)
28. Cui C. China biopharma dealmaking analysis. Jefferies; July 2025. Reported in: BioWorld. July 21, 2025.
29. GSK plc. Press release: GSK and Hengrui Pharma enter licence agreement for PDE3/4 inhibitor and collaboration on up to 11 further programmes. July 2025.
30. Pfizer Inc. Press release: Pfizer enters exclusive license agreement with 3SBio for SSGJ-707. May 2025.
31. Deepening ties: why China is becoming Big Pharma's most essential R&D partner. Pharmaceutical Executive. June 9, 2026.
32. Morgan Stanley. China's biotech boom: from generics to innovators. Morgan Stanley Insights; 2025. https://www.morganstanley.com/insights/articles/china-biotech-boom-generics-to-innovators. Accessed September 2026.
33. Clinical Trials Arena. Congressman proposes draft legislation to accelerate US clinical trials. May 2025. (Reports GlobalData estimates of Phase 1 duration and cost, China versus United States; underlying GlobalData data are proprietary.)
34. McKinsey & Company. The emerging epicenter: Asia's role in biopharma's future. https://www.mckinsey.com/industries/life-sciences/our-insights/the-emerging-epicenter-asias-role-in-biopharmas-future. Accessed September 2026.
35. International Council for Harmonisation. ICH E6(R3) Guideline for Good Clinical Practice. ICH; 2025.
36. International Council for Harmonisation. Press release: ICH Assembly, Montreal, Canada, June 2017 (admission of China's drug regulatory authority as a regulatory member).
37. National Medical Products Administration (China). Announcement on adjusting the review and approval procedures for drug clinical trials (NMPA [2018] No. 50). NMPA; 2018.
38. National Medical Products Administration (China). Announcement on optimizing the review and approval process for clinical trials of innovative drugs ([2025] No. 86), with policy interpretation. NMPA; October 14, 2025. https://english.nmpa.gov.cn/2025-10/14/c_1138494.htm and https://english.nmpa.gov.cn/2025-10/14/c_1138493.htm
39. Investigational New Drug Application: when an IND goes into effect. 21 CFR §312.40(b).
40. Cooley LLP. HHS and Congress push to streamline and onshore clinical trials. July 13, 2026. https://www.cooley.com/news/insight/2026/2026-07-13-hhs-and-congress-push-to-streamline-and-onshore-clinical-trials. Accessed September 2026.
41. Accelerating clinical trial activation. Applied Clinical Trials. June 21, 2024. (Citing WCG data on median activation time, 2021–2023.)
42. Kasner M, Fritsche AH, Leong MC, et al. North American cancer center clinical research capacity and benchmarking in the postpandemic era. JCO Oncol Pract. 2024. doi:10.1200/OP.24.00164.
43. GlobalData. Clinical Trials: Enrollment Trends in Emerging Markets. China. GlobalData; 2018. (Phase 1 to 3 trials, January 2001 to April 2018.)
44. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–263. doi:10.3322/caac.21834. Country data: International Agency for Research on Cancer, Global Cancer Observatory, United States of America fact sheet, GLOBOCAN 2022.
45. Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022. J Natl Cancer Cent. 2024;4(1):47–53.
46. Liu B, Slusarczyk M, Peng S, Welte J. China's trial advantage: tracking the nation's growth in pharma innovation and global investment. Pharmaceutical Executive. 2025;45(5).
47. Baeder G, Zhang E. How China is changing the clinical development landscape. DIA Global Forum. October 2019. https://globalforum.diaglobal.org/issue/october-2019/how-china-is-changing-the-clinical-development-landscape/
48. GlobalData. Clinical trial cost in China is lower than US: Phase I and Phase II single-country trials. Factsheet. November 20, 2025.
49. Grand View Research. Healthcare Contract Research Organization Market Report. 2024; Fortune Business Insights. Contract Research Organization Market Report. 2026.
50. U.S. Food and Drug Administration. Oncologic Drugs Advisory Committee meeting, February 10, 2022: sintilimab biologics license application.
51. International Council for Harmonisation. ICH E17: General Principles for Planning and Design of Multi-Regional Clinical Trials. ICH; 2017.
52. Storm surrounds clinical trial data in China. PDA Letter. Parenteral Drug Association; February 2017. https://www.pda.org/pda-letter-portal/home/full-article/storm-surrounds-clinical-trial-data-in-china. Accessed September 2026.
53. National Defense Authorization Act for Fiscal Year 2026, Pub L No. 119-60, §851 (BIOSECURE Act). Enacted December 18, 2025.
54. Extension of patent term. 35 U.S.C. §156.
55. Biologics Price Competition and Innovation Act, exclusivity for reference products. 42 U.S.C. §262(k)(7).
56. Wouters OJ, McKee M, Luyten J. Estimated research and development investment needed to bring a new medicine to market, 2009–2018. JAMA. 2020;323(9):844–853.
57. DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: new estimates of R&D costs. J Health Econ. 2016;47:20–33.
58. Moore TJ, Zhang H, Anderson G, Alexander GC. Estimated costs of pivotal trials for novel therapeutic agents approved by the US Food and Drug Administration, 2015–2016. JAMA Intern Med. 2018;178(11):1451–1457.
59. Sertkaya A, Wong H-H, Jessup A, Beleche T. Key cost drivers of pharmaceutical clinical trials in the United States. Clin Trials. 2016;13(2):117–126.
60. Andersen JR, von Sehested C, Byrjalsen I, Popik S, Follin AB, Bihlet A. Impact of monitoring approaches on data quality in clinical trials. Br J Clin Pharmacol. 2023;89(6):1756–1766. doi:10.1111/bcp.15615.
61. Andersen JR, Byrjalsen I, Bihlet A, et al. Impact of source data verification on data quality in clinical trials: an empirical post hoc analysis of three phase 3 randomized clinical trials. Br J Clin Pharmacol. 2015;79(4):660–668. https://pmc.ncbi.nlm.nih.gov/articles/PMC4386950/
