Using AI to predict biochemistry for manufacturing and therapeutics
Turn this teardown into a decision-ready prompt for ChatGPT, Claude, or your agent.
20n was a computational synthetic biology company founded in 2014 by computer scientist Saurabh Srivastava and UC Berkeley professor J.Christopher Anderson.
The company built software that predicted DNA insertions into microbes—primarily E. coli and S. cerevisiae—to engineer organisms capable of producing target chemicals through fermentation.Its flagship achievement was predicting and engineering the first known biosynthetic route to acetaminophen (Tylenol), producing it from sugar in a yeast organism. 20n entered YC's Winter 2015 batch, won $1.7M in early contracts from DARPA and a Fortune 500 cosmetics company, and attracted backing from Khosla Ventures.
Despite genuine technical breakthroughs and credible early commercial traction, the company quietly wound down sometime between 2017 and 2018—a victim not of scientific failure, but of a structural mismatch between a software-speed business model and the multi-year validation cycles that define industrial biotech procurement.
20n emerged from a four-year research collaboration at UC Berkeley, not from a garage or a weekend hackathon. Saurabh Srivastava completed his PhD in computer science at the University of Maryland, College Park between 2006 and 2010, specializing in program synthesis—the branch of computer science concerned with automatically generating programs that satisfy a given specification.[1] He then joined UC Berkeley as a postdoctoral researcher, where he spent four years working at the intersection of program synthesis and synthetic biology alongside J. Christopher Anderson, a tenured professor who had spent 17 years in the field.[2]
The pairing was unusual. Anderson brought deep biological domain expertise and academic credibility—the kind that opens doors with DARPA program managers and Fortune 500 R&D teams. Srivastava brought a computer scientist's instinct to treat biology as a search problem: given a target molecule, what sequence of enzymatic reactions, encoded in DNA, would allow a microbe to produce it? The postdoc years appear to have been the period when the core technical thesis was developed and tested, making 20n less a startup built on a hypothesis and more one built on years of pre-commercial research.
The company name encodes the founders' worldview directly. "20n" refers to the 20 standard amino acids specified by the universal genetic code—20^n represents the combinatorial explosion of possible protein sequences, a design space so vast it cannot be manually explored.[3] The name is a mission statement: the only way to navigate this space is computationally.
Srivastava's framing of the market opportunity was deliberately grounded. "Even the most boring chemicals still tend to be billion dollar markets," he told TechCrunch at launch.[4] This was not a pitch about disrupting pharmaceuticals or curing disease—it was a pitch about replacing petrochemical manufacturing with biological fermentation, one molecule at a time, starting with the most commercially obvious targets.
The company entered Y Combinator's Winter 2015 batch as a two-person team based in San Francisco.[5] The precise moment Srivastava and Anderson decided to commercialize their research is not documented publicly, but the timing—immediately following the completion of Srivastava's postdoc in 2014—suggests the decision was made as the research reached a point of commercial readiness, not as a response to external market pressure.
Whether Anderson remained actively involved in day-to-day operations after the YC batch or returned primarily to his UC Berkeley professorship is not confirmed in any public record. His role as CSO suggests a technical advisory function rather than full-time operational leadership—a common structure when an academic co-founds a startup without leaving their institution.
Read the complete post-mortem, the rebuild playbook, and the exact reasons 20n is still worth studying now.