AI-Driven Drug Development Poised to Halve R&D Timelines, Say Sanofi and Biotech Leaders
核心洞察
Sanofi's chief digital officer projects AI can compress drug development timelines from 10–12 years to 5–6 years by simulating clinical trials before they occur.
Industry leaders at theCUBE + NYSE Wired: AI in Bio event emphasized that 90–95% of drugs fail in the clinic, with AI targeting higher probability of clinical success as the key bottleneck.
Multiple biotech companies, including Noetik, Insitro, and Sen-Jam Pharmaceutical, are deploying AI platforms to build causal disease models, digital twins of tumors, and immunomodulators with 200–300% improved success rates.
The pharmaceutical industry's decade-long R&D marathon—stretching 10 to 12 years from hypothesis to regulatory approval—is on the cusp of dramatic compression as artificial intelligence reshapes how drugs are discovered, designed, and de-risked. At the Snowflake Summit 2026 and theCUBE + NYSE Wired: AI in Bio event, executives from Sanofi and a cohort of AI-native biotech companies laid out a unified vision: AI-driven drug development can halve timelines while significantly raising the probability of clinical success.
Emmanuel Frenehard, chief digital officer of Sanofi S.A. (搜索), articulated the ambition directly. "We believe, I believe, that we'll be able to see and we're starting to simulate even clinical trials before they happen," Frenehard said. "So we still have to do them, but we do them with more precision. We do them with a better appreciation of the kind of patients we need in those trials. My intention is to have the time it takes—so from 10 to 12 to five to six."
The clinical failure bottleneck
A core strategic principle driving the AI pivot is the stark reality that approximately 90% of drugs still fail at Phase 3 of clinical trials—close to the finish line, with patients already waiting years for a therapeutic outcome that never materializes. Ron Alfa, co-founder and CEO of Noetik Inc. (搜索), framed the problem in patient-centric terms: "Most drugs fail in the clinic because we don't know which patients are most likely to benefit. If we can train these models to understand patient biology and increase the probability of success, that's really the real bottleneck in getting new drugs to patients."
Noetik is tackling the 90–95% clinical failure rate by training AI models on thousands of real patient tumor samples to create digital twins that can predict which patients are most likely to respond to a given therapy. The company recently signed a major non-exclusive partnership with GlaxoSmithKline PLC (搜索) to deploy the technology across the pharma giant's pipeline.
Unified data as the operating system for AI
Frenehard emphasized that achieving AI precision demands a single, governed data environment. Fragmented data, he argued, is the enemy of AI precision. Sanofi's answer is to run all workflows directly on a unified platform, treating Snowflake as an operating system rather than merely a database.
"All our data shall be on Snowflake and shall be what we call AI ready data, which means it's governed, which means of quality we can trust," Frenehard said. "The workflow is what runs on top of the data. And so with Snowflake and the ability that Snowflake is an operating system, it's not just a data platform."
AI-native biotechs reshape the discovery landscape
Beyond Sanofi's enterprise-scale transformation, a wave of AI-native biotech companies is embedding computational approaches at every stage of the drug development pipeline.
Insitro Inc. (搜索) is generating vast amounts of fit-for-purpose data by modeling disease in human cells at high throughput, then using AI and genetics to create a causal model that predicts the effect of intervening at specific genes. Mary Rozenman, chief financial officer and chief business officer of Insitro, explained that this approach aims to identify genetically validated targets with a significantly higher probability of clinical success, with programs already underway in ALS (搜索) and other hard-to-treat diseases.
Sen-Jam Pharmaceutical LLC (搜索) is developing immunomodulators designed to restore the immune system to homeostasis and reduce chronic inflammation—or inflammaging—linked to a majority of age-related diseases and deaths. The company has advanced five assets to Phase 2 using traditional methods and is now applying its AI platform, Galaxy, with Atlas serving as its discovery layer. According to Jim Iversen, CEO of Sen-Jam Pharmaceutical, the platform can increase the probability of clinical success by 200–300% based on trial data.
Manifold Biotechnologies Inc. (搜索) has developed a platform that inverts the traditional drug discovery funnel by measuring hundreds of thousands of AI-designed drug candidates at once inside a single animal rather than testing them one at a time. This high-throughput in vivo approach generates the rich biological data needed to train better models and improve tissue targeting, including the ability to cross the blood-brain barrier, explained Gleb Kuznetsov, co-founder and CEO.
Expanding the therapeutic frontier
Other companies are applying AI to unlock entirely new therapeutic modalities. ExpressionEdits Ltd. (搜索) discovered during COVID research that the DNA instructions used to produce therapeutic proteins still resemble 50-year-old viral sequences rather than native human genes, causing cells to silence them. By using AI to redesign these instructions, the company enables significantly higher protein expression—a critical unlock for gene therapies, vaccines, and recombinant protein drugs. Co-founder and CEO Kärt Tomberg noted the company has already secured multiple large pharma partnerships.
Fauna Bio Inc. (搜索) is mining hibernating mammals that naturally repair heart damage, regrow muscle, and reverse neurodegeneration-like pathology, then combining that data with human patient data to identify high-potential drug targets. The company has built a knowledge graph and graph neural network to surface these insights and is advancing its lead heart-failure therapy toward the clinic next year, according to Linda Goodman, co-founder and CTO.
Concerto Biosciences Inc. (搜索), an MIT spinoff, is advancing a once-weekly topical multibacterial therapy designed to treat and prevent eczema (搜索) flares by restoring a healthier skin microbiome. The company has built a proprietary high-throughput data platform and AI models that predict effective microbial combinations, has completed Phase 1b for its lead eczema program, and is expanding into additional skin and women's health applications, said Cheri Ackerman Araromi, co-founder and CEO.
The investment thesis: drugs, not just tools
Jacob Oppenheim, venture partner at RA Capital Management L.P. (搜索), offered a grounded perspective on where AI creates the greatest near-term value. The only proven business model in biotech remains making successful medicines, he noted, so any AI or platform technology must clearly show how it produces a better therapeutic faster than alternatives. The greatest near-term value of AI lies in expanding the hypothesis space, accelerating experimental cycles, and better integrating human genetics data rather than solving only isolated steps in the discovery process.
Alloy Therapeutics Inc. (搜索), described as a "hyperscaler" for the biotech industry, is building the infrastructure layer that powers drug pipelines across the sector. Founder and CEO Errik Anderson explained that by aligning its business model with partner success through a mix of services, milestones, and royalties, Alloy aims to lower the cost and time of bringing medicines to patients without owning its own drug pipeline.
The convergence of unified data infrastructure, causal AI models, and high-throughput biological experimentation signals a fundamental shift: AI is moving from the edges of biotech into its core, with the shared goal of delivering approved therapies to patients faster and with greater certainty than ever before.