AI is generating more potential drug candidates than ever before. The problem is figuring out which ones actually work. A new startup called 10x Science, founded by Stanford researchers from Nobel laureate Dr. Carolyn Bertozzi's lab, has raised $4.8 million in seed funding to build an AI platform that can characterize and analyze drug molecules at a speed and scale that current methods cannot match.
The Bottleneck Problem
Google DeepMind's breakthrough in predicting protein structures — work that earned a Nobel Prize in Chemistry — has transformed the front end of drug discovery. AI models can now generate vast numbers of potential drug candidates faster than ever before. But there is a growing mismatch between the speed at which AI models generate candidates and the speed at which those candidates can be evaluated.
Every potential drug must pass through a characterization process before it can move toward testing and mass production. That means measuring molecules at the atomic level, understanding how they behave, and verifying that they do what they are supposed to do. No matter how many candidates AI adds to the top of the funnel, they all have to pass through this bottleneck.
Co-founder David Roberts put it simply: you can add as many candidates as you want, but everything needs to be measured.
How 10x Science Works
The company's platform combines deterministic chemistry and biology algorithms with AI agents that can interpret complex molecular data. The core technology centers on mass spectrometry — a technique for determining the atomic structure of molecules by measuring them in an electric field.
Mass spectrometry produces extremely complex data that requires deep expertise to interpret. Analyzing it manually is slow and expensive, creating a bottleneck that limits how many drug candidates can be evaluated. 10x Science's platform automates much of that analysis using AI agents trained on spectrometry data, while keeping the results traceable — a critical requirement for regulatory compliance in pharmaceutical development.
The team did significant work training the models on spectrometry data to ensure accuracy. In an industry where AI tools have frequently over-promised and suffered reliability issues, 10x Science's approach is distinguished by its domain specificity — the founders are not AI generalists applying machine learning to biology, but experienced biochemists who understand exactly what the data means.
Early Results Are Promising
Matthew Crawford, a scientist at Rilas Technologies who runs chemical analyses for biotech clients, has been using the platform for several weeks. He says it has meaningfully accelerated his work, noting that the model surprised him with its ability to explain its conclusions, find the right data autonomously, and adapt to evaluating different kinds of molecules.
In one case, Crawford ran a protein through the system and it identified the protein from the filename alone, searched databases for the correct sequence, and began analysis without any manual input. For a tool handling sensitive pharmaceutical data, that kind of autonomous but accurate behavior is exactly what enterprise customers need.
The Business Model
10x Science operates as a SaaS platform that pharmaceutical companies pay for monthly to process their drug candidates. This means the company's revenue is not dependent on any specific drug succeeding in clinical trials — it earns money regardless of whether the molecules it analyzes eventually become approved treatments.
The seed round was led by Initialized Capital with participation from Y Combinator, Civilization Ventures, and Founder Factor. For investors, the appeal is a platform play that sits at the intersection of AI and biotech without carrying the binary risk of traditional drug development.
Zoe Perret, a partner at Initialized, noted that the deep expertise of the founders creates a natural moat. There simply are not many people who understand both mass spectrometry data and how to build AI systems that can interpret it accurately.
From Molecules to Molecular Intelligence
The company's three founders — Roberts and Andrew Reiter, both biochemists, and Vishnu Tejas, a serial founder with computer science and AI expertise — worked together in Bertozzi's Stanford lab studying interactions between cancer cells and the immune system. Their frustration with the inability to understand what was happening at a molecular level led directly to the creation of 10x Science.
The immediate goal is to help pharma companies characterize proteins faster. But Roberts sees a bigger opportunity: combining protein structure data with other cellular information to create what he calls a new way to define molecular intelligence. If 10x can deliver that, it would become not just a characterization tool but a foundational platform for next-generation biological research.
The Bigger Picture
10x Science represents a different kind of AI application from the chatbots and coding tools that dominate headlines. It is not replacing human researchers — it is giving them the ability to process data at a scale that manual methods cannot achieve. In an industry where a single successful drug can generate billions in revenue and save millions of lives, accelerating the path from AI-generated candidates to validated treatments could be one of the most impactful uses of artificial intelligence anywhere.
The drug discovery pipeline is long, expensive, and brutally selective. 10x Science is betting that AI can widen the bottleneck that determines how many candidates ever get a chance to prove themselves.
