Can artificial intelligence conquer most human diseases within a decade? Silicon Valley’s answer is yes\u2014but the pharma chiefs who actually run new-drug pipelines are collectively reaching the opposite conclusion. According to The Wall Street Journal, Vertex Pharmaceuticals (VRTX) Executive Chairman Jeff Leiden offered a blunt assessment of the claim that “AI will cure 90% of human diseases”: “It’s complete nonsense.”<\/p>\n
Leiden disclosed that Vertex organized roughly 50 AI companies to comb through its dataset of 10,000 drug compounds using their technology, and not one produced an effective drug. “It just doesn’t work on its own,” he said. “Any company that claims it can\u2014at least in our experience, and we looked at most of them\u2014isn’t true.”<\/p>\n
The remarks stand in sharp contrast to Silicon Valley’s recent optimistic narrative. Anthropic CEO Dario Amodei posted on X in September that “curing most human diseases within 5 to 10 years is actually possible, even though it sounds crazy.” Anthropic has since acquired a stealth-mode biotech and a one-person biotech operation run almost entirely by AI; OpenAI announced in October that it was acquiring an AI drug discovery company.<\/p>\n
Capital-side enthusiasm is equally notable. According to McKinsey, global venture capital and private equity investment in AI drug discovery reached $8.4 billion in 2025, double the roughly $4 billion of two years earlier. Major drugmakers are also racing to tie up with AI companies: Eli Lilly and Nvidia announced in January a commitment of up to $1 billion over five years for drug discovery and robotics research, while Bristol Myers Squibb, Novo Nordisk and Sanofi have each formed partnerships with OpenAI or Anthropic.<\/p>\n
Japanese drugmakers are stepping up as well. Takeda Pharmaceutical signed a multiyear collaboration with Iambic Therapeutics in February worth up to $1.7 billion in potential total value, focused on small-molecule design in oncology and gastrointestinal and inflammatory diseases; in July it reached a deal worth up to about $600 million with Insilico Medicine, securing global exclusive development and commercialization rights to therapies selected under the collaboration. Nvidia has also partnered with Astellas, Fujifilm and other Japanese companies in AI healthcare.<\/p>\n
Yet one unavoidable fact remains: to date, no drug designed entirely by artificial intelligence has won marketing approval from the U.S. Food and Drug Administration (FDA).<\/p>\n
A ‘Co-Pilot,’ Not a ‘Driver’<\/p>\n
Executives with deep roots in new-drug development do not deny AI’s value, but they stress an auxiliary role. AstraZeneca (AZN) CEO Pascal Soriot said: “If we stay in this industry long enough, we quickly learn humility. Disease is complex, biology is complex. Saying AI will cure cancer may be a bit of a leap.”<\/p>\n
AstraZeneca recently signed an agreement with AI biotech firm Insitro, with $40 million upfront, to discover novel cardiovascular and weight-loss drugs, with milestone payments of up to about $1.46 billion; the company is also working with Oracle to optimize patient recruitment for clinical trials and has opened a new R&D site in Cambridge, Massachusetts, to use AI to accelerate experimental work.<\/p>\n
Regeneron (REGN) Chief Scientific Officer George Yancopoulos, who holds more than 1,000 patents, drew a more specific boundary: “AI won’t miraculously tell us what to do, and it won’t shorten a 15-year development cycle to five years.”<\/p>\n
Bayer Chief Operating Officer Sebastian Guth offered an example of where AI stands today: the company’s gene therapy team used AI to sift through 100 million patient records within days, analyzing genetic markers and medical history to identify five diseases best suited to its gene-editing technology. “Is this AI curing disease? No. The magic happens when you combine the brilliance of human scientists with AI.”<\/p>\n
The Technical Bottleneck: Stuck on ‘Off-Target Effects’<\/p>\n
The drugmakers’ caution is not conservatism but points to a specific scientific obstacle. Alex Devereson, a McKinsey senior partner focused on AI drug R&D, noted that about a quarter of late-stage clinical trial failures can be traced to safety issues\u2014drugs producing “off-target effects” in the body, acting on unintended targets and causing adverse or even harmful side effects\u2014and existing AI models remain limited in predicting such problems.<\/p>\n
“Investment is flowing to the easiest problems to solve, not the most important ones,” Devereson said. He acknowledged that AI does design experimental drugs faster than before, but faster design does not equal viable drugs. On the regulatory front, the U.S. FDA and the European Medicines Agency (EMA) jointly issued “Guiding Principles of Good AI Practice in Drug Development” in January, setting out 10 principles covering evidence generation and monitoring; an earlier FDA draft guidance centered on “model credibility.” But these frameworks govern how AI is used and cannot accelerate some of the thornier steps, such as finding enough suitable patients and monitoring them throughout a study.<\/p>\n
Companies genuinely pursuing AI-centric R&D have also made limited progress. Among the more advanced candidates, one has just completed a Phase 2a trial and another has entered Phase 3, but neither has yet filed for marketing approval.<\/p>\n
CompanyCandidateTarget\/IndicationLatest ProgressInsilico Medicinerentosertib (ISM001-055)TNIK\/idiopathic pulmonary fibrosisPhase 2a results published June 2025; now in Phase 3 in ChinaGenerate BiomedicinesGB-0895TSLP\/severe asthmaGlobal Phase 3 (SOLAIRIA-1, SOLAIRIA-2)<\/p>\n
Insilico Medicine’s rentosertib was discovered\u2014both target and molecule\u2014by a generative AI platform. The Phase 2a trial published in Nature Medicine in June 2025 covered 71 patients with idiopathic pulmonary fibrosis; the 60 mg dose group showed a mean increase in forced vital capacity of 2.82% after 12 weeks, within the 2% to 6% minimal clinically important difference range for that measure; but the trial sample was small, and some patients discontinued due to liver injury. Generate Biomedicines’ GB-0895 is an AI-optimized anti-TSLP antibody designed for dosing once every six months, now in two global Phase 3 studies in patients with severe asthma.<\/p>\n
Generate Biomedicines co-founder and Chief Technology Officer Gevorg Grigoryan said in an email interview that Nvidia’s GPUs are “indispensable” for designing AI-generated drugs, but “compute alone does not constitute a differentiated advantage”; the company most likely to lead in AI drug discovery will need to combine computing power with “proprietary biological data, the scale of experimental measurement, and the ability to close the loop between generation and measurement.”<\/p>\n
The Gap Between Capital Narrative and R&D Reality<\/p>\n
At the heart of this divide is a mismatch between two time scales. Silicon Valley measures progress at the pace of software iteration, while a new drug typically takes more than a decade from target discovery to approval, with persistently high failure rates. AI can compress the time spent on compound screening, literature review and patient matching, but it struggles to replace validation of biological mechanisms.<\/p>\n
For investors, this means the valuation logic of the AI-pharma sector needs recalibration: in the near term, companies that provide drugmakers with R&D tools, data processing and clinical-trial optimization are more likely to deliver revenue; the value of AI-native firms claiming they can independently produce blockbuster drugs still hinges on the long and uncertain gate of regulatory approval. Leiden’s phrasing may best capture the current industry consensus\u2014AI as a scientist’s “co-pilot” is genuinely valuable, but on the core proposition of independently driving new-drug discovery, between Silicon Valley’s promises and the laboratory’s practice still lies a list of questions about human biology that remains unanswered.<\/p>\n","protected":false},"excerpt":{"rendered":"Can artificial intelligence conquer most human diseases within a decade? Silicon Valley’s answer is yes\u2014but the pharma chiefs…\n","protected":false},"author":2,"featured_media":161098,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[20662],"tags":[62827,3424,62835,11153,48784,3523,62829,4658,62836,62832,62831,62826,62833,62834,47657,62825,26645,62837,11316,54881,62830,62828,47937,35873],"class_list":["post-161097","post","type-post","status-publish","format-standard","has-post-thumbnail","category-astrazeneca","tag-alex-devereson","tag-anthropic","tag-astellas","tag-astrazeneca","tag-bayer","tag-dario-amodei","tag-european-medicines-agency-ema","tag-fda","tag-fujifilm","tag-gb-0895","tag-generate-biomedicines","tag-george-yancopoulos","tag-gevorg-grigoryan","tag-iambic-therapeutics","tag-insilico-medicine","tag-jeff-leiden","tag-mckinsey","tag-nature-medicine","tag-pascal-soriot","tag-regeneron","tag-rentosertib","tag-sebastian-guth","tag-takeda-pharmaceutical","tag-vertex-pharmaceuticals"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/posts\/161097","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/comments?post=161097"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/posts\/161097\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/media\/161098"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/media?parent=161097"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/categories?post=161097"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/britain\/wp-json\/wp\/v2\/tags?post=161097"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}