Why they built it
The founders saw that finding biosynthetic routes to industrial chemicals was manual, slow, and limited by what human experts had read. They built 20n/act to treat biology as a search problem, predicting DNA insertions for E. coli and S. cerevisiae so microbes could make target chemicals from sugar by fermentation. They thought it was worth building because specialty chemicals was a $980B market where even the most boring chemicals were billion-dollar markets.
Why it didn't work out
20n quietly wound down between 2017 and 2018 after its $1.7M in early contracts from DARPA and a Fortune 500 cosmetics company did not grow into sustaining revenue, with no Series A announced. Its software licensing model required multi-year customer validation, pilot trials, and manufacturing qualification before production revenue, which was structurally incompatible with a small seed-funded team. The 20n/act platform was open-sourced under the GNU GPL, ending commercial operation.
What went wrong
Most important first.
- 1
Multi-year industrial procurement cycles Distribution
Fortune 500 chemical customers required CRO validation, pilot fermentation trials, and regulatory and safety review taking two to four years before any production revenue, which a two-person seed-stage team could not survive through multiple cycles.
- 2
No wet lab, outsourced validation to CROs Technology
20n owned only DNA design and relied on outside organizations to build DNA, get it into cells, and grow them up, leaving iteration speed measured in weeks to months outside its control.
- 3
Full-stack competitors offered build and test Competition
Ginkgo Bioworks and Zymergen combined computational design with in-house automated wet labs to deliver design-build-test-iterate in one relationship, while 20n offered design only and required customers to manage CRO work separately.
- 4
Per-molecule licenses did not become recurring revenue Monetization
Licenses starting in the hundreds of thousands of dollars per molecule plus milestone-based DARPA contracts produced $1.7M in early contracts and 40+ organisms but did not compound, with no subscription model and no Series A announced.
What could be different today
- AlphaFold2 (July 2021) and ESMFold (October 2022) predict protein 3D structure from amino acid sequence with near-experimental accuracy at near-zero marginal cost, allowing ranking of enzymatic routes by structural compatibility rather than literature co-occurrence alone.
- Emerald Cloud Lab, Strateos, and Arctoris offer API-accessible robotic wet-lab execution, letting a software team submit DNA designs and get fermentation results in days without managing CRO scheduling.
- DNA synthesis cost fell from ~$0.50 per base pair in 2015 to under $0.05 per base pair in 2024, sharply lowering the cost of testing each predicted design.
- The Inflation Reduction Act (2022) and EU Green Deal create financial incentives and procurement mandates for bio-based chemical manufacturing, reducing sales friction for fermentation-based production.