Small molecules remain the backbone of modern drug development, accounting for roughly two thirds of FDA-approved therapeutics, but the chemistry required to discover and manufacture them has changed little in decades.
While artificial intelligence (AI) has reshaped early-stage design and prediction, synthesis remains a persistent bottleneck, often slowing programs by years and limiting how effectively AI can be applied in practice.
Excelsior Sciences is betting that reengineering chemistry itself is the key to unlocking AI-driven drug discovery at scale. The New York–based company announced a $95 million financing to accelerate development of what it calls “machine-native chemistry,” an approach designed to allow both AI systems and automated platforms to design, make, and test small molecules in closed-loop workflows.
The funding includes a $70 million Series A co-led by Deerfield Management, Khosla Ventures, and Sofinnova Partners, along with a $25 million grant from New York’s Empire State Development. Eli Lilly and Company, Cornucopian Capital, Illinois Ventures, MIT, and other investors also participated.
“Our unique breakthrough is based on leveraging iterative carbon–carbon bond formation, analogous to peptide synthesis, which forms the foundation for our smart bloccs platform,” Michael Foley, cofounder and CEO of Excelsior Sciences, told DDN. “Smart bloccs are automated, synthesis-friendly chemical building blocks that enable iterative carbon–carbon bond formation.”
Unlike traditional automated chemistry systems, which attempt to replicate manual, reaction-specific workflows, Excelsior’s platform is built around a limited set of highly versatile reactions that can be run under standardized conditions. Smart bloccs act as modular components — a chemical “language,” as Foley describes it — that AI systems can use to guide molecular design and synthesis.
“Smart bloccs serve as a modular chemical ‘language,’ where bloccs act as tokens which AI can use to derive novel insights and guide discovery,” Foley said.
Breaking the synthesis bottleneck
Automated synthesis has long been viewed as a critical missing link in AI-driven drug discovery. Many generative chemistry platforms can propose thousands of molecules, but only a small fraction are ever synthesized and tested, limiting feedback and slowing optimization.
“Previous attempts to harness AI to chemistry have focused on wrapping AI and automation around traditional artisanal processes,” Foley noted. “This ‘brute force’ approach has only yielded marginal gains. Excelsior Sciences’ approach is unique because it creates a new form of chemistry that machines can do and AI can use.”
By contrast, Excelsior’s system is designed to generate and test molecules at a scale sufficient to support true closed-loop learning. “Our AI-centric competitors can generate hypotheses, but only Excelsior Sciences can make and test molecules on a scale that enables us to optimize multiple parameters simultaneously,” Foley said.
Technically, the platform relies on three to four highly generalizable reactions derived from foundational work in the laboratory of Marty Burke at the University of Illinois Urbana-Champaign. Those reactions allow the system to reliably produce diverse small molecules in quantities sufficient for purification and testing across as many as 20 assays, without constant reconfiguration.
“There are no examples of systems that can do thousands of different things thousands of different ways and do them well,” Foley said. “We have leveraged a chemical breakthrough … that allows us to use a small number of highly versatile reactions, each run under a standard procedure, to reliably produce highly diverse small molecules.”
Peer-reviewed validation and industry context
Excelsior’s approach builds on a body of peer-reviewed research demonstrating automated, closed-loop chemical discovery, including studies published in Science and Nature Chemistry. Key examples include work on automated iterative C(sp³)–C bond formation and closed-loop systems that allow AI to generate chemical knowledge directly from experimental feedback.
Synthesis has historically accounted for a substantial share of drug development timelines and costs, particularly in lead optimization. Industry analyses suggest that chemistry and manufacturing challenges contribute significantly to late-stage attrition and delays, especially for complex small-molecule programs. By standardizing synthesis and embedding it directly into AI-driven workflows, Excelsior aims to compress the discovery-to-lead cycle.
“Synthesis represents a major roadblock in small-molecule development and adds years to the process of bringing new drugs to market,” Foley said. “Our chemistry enables automated synthesis, shattering the bottleneck and driving benefits for every industry that relies on small molecules.”
Manufacturing, reshoring, and strategic implications
Beyond discovery, Excelsior is positioning its platform as a response to growing concerns about pharmaceutical supply chain resilience. The US pharmaceutical industry has announced plans to invest hundreds of billions of dollars in reshoring and supply chain diversification, driven by geopolitical risk and pandemic-era disruptions.
“By slashing the cost and time involved in discovering and synthesizing new molecules, Excelsior Sciences removes the biggest barriers to reshoring — expense and uncertainty,” Foley said. “This in turn has benefits that ripple all the way through the supply chain, providing the US with strategic immunity in a rapidly changing world.”
The company plans to use the new funding to scale its platform, advance an internal pipeline, and pursue partnerships across biotech and other sectors that rely on small molecules, including materials science. While the underlying chemistry could theoretically support other modalities, Foley said the near-term focus remains on small molecules.
“We are focused on scaling the smart bloccs platform and demonstrating our ability to revolutionize the discovery and manufacture of small molecules for a range of applications — primarily but by no means exclusively in biotech,” he added.
As AI continues to reshape drug discovery, Excelsior’s approach highlights a growing recognition across the industry: without equally radical advances in chemistry and synthesis, the promise of AI-designed medicines may remain constrained by what can actually be made, tested, and manufactured at scale.
