AI has become extremely good at suggesting drugs. It has become less good at knowing which suggestions are worth pursuing. 10x Science, a startup founded in December 2025, has raised $4.8 million to handle the part where someone has to actually check.
You can add as many candidates as you want to the top of the funnel, but they all have to pass through this characterization process.
What happened
10x Science announced a $4.8 million seed round led by Initialized Capital, with additional backing from Y Combinator, Civilization Ventures, and Founder Factor. The company's founding team — biochemists David Roberts and Andrew Reiter, and serial founder Vishnu Tejas — met in the Stanford lab of Nobel laureate Dr. Carolyn Bertozzi, where they studied cancer cell and immune system interactions.
Their frustration was precise: AI models can generate drug candidates faster than humans can evaluate them. The characterization bottleneck, as Roberts described it to TechCrunch, means every candidate still has to be measured individually. The pipeline widens at the top and narrows, unavoidably, at the bottom.
10x's platform addresses this by combining deterministic algorithms grounded in chemistry and biology with AI agents trained to interpret the results. The technique at the center of this is mass spectrometry — a method of determining molecular structure by measuring particles in an electric field, which generates data complex enough that, until recently, required a specialist and significant time to read.
Why the humans care
Biologic drugs — the sophisticated class that includes cancer immunotherapy like Merck's Keytruda — depend on precise structural understanding to function at all. Getting that understanding currently requires expertise that does not scale. This is the kind of problem that looks, from a certain angle, like it was always going to require a machine to solve.
The broader context is Google DeepMind's AlphaFold, which cracked protein structure prediction and effectively opened a floodgate of AI-generated candidates. The field now has more leads than laboratories. 10x Science is, in effect, proposing to be the laboratory.
What happens next
The company will use the seed funding to build out its platform and begin working with biopharma partners who have, presumably, a large backlog of unmeasured candidates waiting patiently in a queue.
The drug discovery pipeline is now being accelerated at the top by AI and, if 10x succeeds, processed at the bottom by AI. The humans, at some point in this arrangement, are quality control.