The gap is almost absurd when you think about it: artificial intelligence can now dream up new drug candidates in days, sometimes hours. But figuring out whether those molecules will actually work—or are even safe—still takes weeks, sometimes months, of painstaking lab analysis.
It's a bottleneck that David Stephen Roberts watched play out repeatedly during his postdoc in Carolyn Bertozzi's lab at Stanford. He'd seen it before that, too, working at the Broad Institute's Proteomics Platform. Drug discovery was racing ahead. The characterization work? Stuck in what felt like the analog era.
So Roberts did what frustrated scientists with 38 publications and a Rolodex of co-founders tend to do: he started a company. 10x Science, which emerged from Y Combinator's Winter 2026 batch, announced a $4.8 million seed round on April 22. The round was oversubscribed—Initialized Capital led, with Y Combinator, Civilization Ventures, Founder Factor, and a handful of strategic angels coming in behind them.
The pitch is deceptively straightforward. Mass spectrometry, the gold-standard technique for characterizing protein therapeutics at the molecular level, generates mountains of data that someone—usually a highly trained specialist—has to interpret by hand. It's regulatory table stakes for bringing biologics to market, but the manual process can stretch on for months per candidate. 10x Science believes it can automate most of that analysis, delivering results with the explainability and traceability that pharma needs in a fraction of the time.
Whether the industry will trust a four-month-old startup with compliance-grade analysis is another question entirely.
When the Nobel Laureate Says You're Onto Something
Roberts teamed up with Andrew Reiter, a Stanford Biology Ph.D. student and NSF fellow, and Vishnu Teju Sajja, a two-time YC alum with engineering chops honed at Nooks. The technical bet they're making hinges on what they call "deep memory" models—AI architectures designed to learn across datasets and customers over time while maintaining the kind of explainability that regulators demand.
That last part matters. Pharma companies can't just hand the FDA a black-box prediction and shrug. Every analytical result needs a traceable path back to the underlying chemistry and biology. 10x Science claims its platform blends deterministic algorithms with AI in a way that satisfies both the need for speed and the need for auditability.
"The discovery process is getting faster, but we're still stuck with the same old characterization methods," Reiter noted in the company's YC launch materials.
Bertozzi—who won the 2022 Nobel Prize in Chemistry and advised Roberts at Stanford—offered her own endorsement in the company's press release, pointing to the need for "modernized characterization infrastructure" to match the pace of modern drug development. That's not nothing, coming from someone whose career has been built on inventing new ways to manipulate molecular biology.
Early users seem cautiously optimistic. Matthew Crawford of Rilas Technologies, a contract research organization, reported meaningful time savings after a few weeks working with the platform, though the company hasn't disclosed specifics on just how much faster the turnaround is.
The Compliance Tightrope
10x Science is already working with what it describes as "multiple major pharmaceutical companies," according to a March job posting for a founding engineer. The company hasn't named those customers publicly—unsurprising, given how tightly pharma guards information about its pipeline. But the fact that they've secured any pharma partnerships four months in is itself a signal. These aren't companies known for moving fast or taking risks on unproven vendors.
The seed capital will go toward hiring—the company is specifically looking for systems-level engineers fluent in Rust, C++, or Go—and refining the platform for broader enterprise deployment. Incorporated in Delaware, registered to do business in California as of February, the three-person founding team is operating out of San Francisco and building what amounts to mission-critical infrastructure for an industry that doesn't tolerate errors well.
The market opportunity is real, if not exactly explosive. Drug discovery mass spectrometry is projected to hit $1.69 billion by 2031, up from roughly $1.05 billion last year, per MarketsandMarkets data from January. Biologics, meanwhile, accounted for 51% of U.S. drug spending despite making up just 5% of prescriptions as of 2025, according to FDA figures. Twelve biologics won approval last year alone, and the complexity of these engineered therapies keeps climbing.
Each one requires exhaustive analytical work before it can reach patients.
Zoe Perret, a partner at Initialized Capital, framed the investment as a play on recurring revenue in a highly regulated vertical. "This is SaaS-like infrastructure for pharma," she told TechCrunch. "Every new molecule needs this analysis."
That's true enough. Whether 10x Science can scale the technology, earn the trust of notoriously risk-averse pharma incumbents, and navigate the regulatory gauntlet that comes with touching anything related to drug approval—well, that's the $4.8 million question.
For now, the company is hiring, the founders are shipping, and somewhere in a lab, a mass spectrometer is spitting out data that still, for the moment, someone will have to interpret the old-fashioned way.
