Excelsior Sciences, photographed in the language it actually speaks: building blocks, snapped together, waiting for a machine to read them.
The New York biotech that decided chemistry had been done by hand for long enough - and built a way for machines to do it instead.
In a custom-built lab inside the Cure building in Manhattan, a molecule is being made without a chemist standing over it. Synthesis, purification, and biological testing sit in one continuous line, roughly forty feet long. A robot snaps chemical building blocks together. Software decides which to try next. The molecule that comes out the other end was designed by an algorithm and built by a machine - and the people who run the place are mostly watching data, not pipettes.
This is Excelsior Sciences. It is small, around 26 people, and it is not shy about its ambition: it wants to end what it calls chemistry's 200-year artisanal phase. The pitch is unusual because it is literal. Most biotechs sell a drug, or a target, or a model. Excelsior is selling a way of doing chemistry at all.
"Our smart bloccs platform is the first system to seamlessly integrate discovery and scale-up chemistry."
- Michael Foley, Co-Founder & CEOHere is the awkward gap at the center of modern drug discovery. The design side has gone digital - AI models can propose vast libraries of candidate molecules in an afternoon. The making side has not. Turning a designed molecule into a real one still depends on skilled chemists, bespoke reactions, and a lot of time at the bench. Theory races ahead; synthesis trudges behind.
The founders watched this play out and reached a slightly inconvenient conclusion: the bottleneck was not the AI. It was the chemistry itself, which was never built to be read by a machine. And chemistry that only humans can do has a quiet vulnerability - when the humans stop, so does everything else.
"When that stops, everything stops - science stops, healthcare stops."
- Michael Foley, on chemistry's dependence on the benchThe pandemic made the point in a less abstract way. Supply chains for small-molecule drugs turned out to be long, opaque, and easy to disrupt. A field that prides itself on precision had outsourced much of its own manufacturing to places it could not see. Excelsior's answer is not to add more chemists. It is to make the chemistry simple enough that you need fewer of them.
Excelsior was spun out of Deerfield Management in 2024 and built around a piece of science that predates the company. In Marty Burke's lab at the University of Illinois, researchers discovered and patented a way to build complex molecules from standardized, automation-friendly building blocks - assembled iteratively, the way nature assembles peptides. That insight is the seed. Excelsior is the attempt to turn it into an industry.
The name is a wink. "Excelsior" - ever upward - is the official motto of New York State. A company built to reshore American chemistry chose American chemistry's most state-pride-coded word.
The platform is called smart bloccs, and the metaphor people keep reaching for is LEGO - though peptides are the more accurate one. Each blocc is a modular, function-infused chemical building block. Snap them together through iterative carbon-carbon bond formation and you assemble a small molecule the way you'd assemble a sentence from tokens. The chemistry is deliberately simplified so that the same set of reactions can build an enormous range of molecules.
That simplicity is the whole trick. Because the steps are standardized, a machine can run them. Because the blocks are AI-readable, an algorithm can learn from every attempt and propose the next one. Synthesis, high-throughput testing, and AI optimization run as one closed loop - design, make, test, repeat - without a human re-inventing the route each time.
"Excelsior is the missing piece - turning theory into tangible molecules."
- Edward Kliphuis, Sofinnova PartnersStandardized building blocks replace bespoke, one-off reactions.
Chemistry designed to be executed by automation, not hands.
Every block and reaction is data an algorithm can learn from.
Simplified steps map onto continuous-flow manufacturing.
The payoff Excelsior cares about most is the part that usually gets bolted on later: manufacturing. Because the chemistry is simple and continuous-flow-friendly, the same process that discovers a molecule can scale it - less labor, less waste, less of the discovery-to-production hand-off where projects go to die.
Building-block chemistry discovered and patented in Marty Burke's University of Illinois lab; research lineage published in Nature and Science.
Excelsior Sciences founded and spun out of Deerfield Management, with a custom lab at Cure in NYC.
Integrated synthesis-to-testing lab running closed-loop cycles in roughly 40 feet of Manhattan.
$70M Series A (Deerfield, Khosla, Sofinnova, Eli Lilly, MIT and others) plus a $25M Empire State Development grant.
Skeptics are right to ask what's real here, because plenty of "AI chemistry" is a slide deck. Excelsior's evidence is partly the syndicate it pulled together. A $70M Series A co-led by Deerfield, Khosla Ventures, and Sofinnova Partners is one signal. The company you keep is another: Eli Lilly and MIT both put money in - the kind of partners that tend to do their own diligence on whether the chemistry works.
"We view Excelsior's applications as transformational."
- Jim Flynn, Deerfield ManagementThe other proof is structural. New York State doesn't hand out $25M grants for vibes; the money is a bet that Excelsior will build chemistry jobs and capacity in the state. And the founders' track record - Foley and Jensen worked together at Deerfield Discovery & Development before this - means the syndicate isn't backing first-timers. None of this guarantees the drugs work. It does mean serious people think the platform is more than a metaphor.
Strip away the funding and the mission is almost stubbornly simple: make small-molecule chemistry something you can program. If a molecule can be designed by AI, built by a machine, tested in the same loop, and scaled in the same chemistry, then the cost and complexity that gate drug discovery start to fall. The vision extends past pharma - the same approach points at materials science and agtech, anywhere small molecules matter.
"Being able to manufacture small molecules this way opens doors to breakthroughs."
- Nessan Bermingham, Khosla VenturesThere's a national subtext too. Reshoring drug discovery and manufacturing - making more of it in the U.S., where it can be seen and trusted - is part of the argument Excelsior makes to investors and to the state of New York alike. Resilience, in other words, isn't a marketing word here; it's the literal point of building chemistry that doesn't stop when the chemists go home.
It helps that the model has more than one customer in mind. Excelsior runs its own internal therapeutics pipeline while partnering with outside companies across therapeutics and materials science, so the platform earns its keep two ways at once. That dual structure is also a hedge: if one drug program stalls, the chemistry underneath it still has somewhere to go. For a company selling a method rather than a single molecule, that breadth is the whole insurance policy.
Return to that lab inside Cure, the molecule moving down the line without a chemist hovering over it. A year ago that scene was a demo. Now it has $95M behind it and a syndicate of people who make their living being skeptical. The bet is that this stops being a curiosity and becomes the boring default - that "a machine made it" is something nobody bothers to mention.
Excelsior hasn't proven that yet. Its pipeline is early, its team is small, and "we're reinventing chemistry" is the kind of sentence that has buried plenty of companies. But the gap it's aiming at is real, the science underneath it is patented and published, and the people funding it are not easily impressed. Chemistry has been a craft for two centuries. Excelsior is betting the next two hundred years look a lot more like software - and it has built a forty-foot room to prove it.