Jason Kelly on Ginkgo Bioworks: GPT-driven autonomous labs beat Stanford benchmark by 40%, democratizing science
Feb 9, 2026 with Jason Kelly
Key Points
- OpenAI's model ran six experimental rounds on Ginkgo Bioworks' autonomous lab platform, beating Stanford's cell-free protein synthesis benchmark by 40% by round six.
- Ginkgo won a $47 million Department of Energy contract to deploy 115 autonomous lab robots across Pacific Northwest National Labs, signaling early government adoption.
- Ginkgo frames autonomous AI-driven labs as a counterweight to drug discovery's geographic shift to China, where large pharma acquisitions of new candidates jumped from under 5% to over 40% in two years.
Summary
Ginkgo Bioworks ran OpenAI's model through six experimental rounds on its autonomous lab platform. By the fourth round, the system beat the scientific state of the art for cell-free protein synthesis. By the sixth round, it had surpassed the benchmark by 40%.
The result reflects a shift in how drug discovery operates. Large pharma acquisitions of new drug candidates from China have jumped from less than 5% two years ago to more than 40% today. Drug discovery startups that traditionally served as the innovation engine for the pharmaceutical industry have been relocating to China over the past two years.
The Ginkgo-OpenAI collaboration demonstrates a model where AI systems can autonomously manage experimental design and iteration, handling variability and optimization faster than human researchers. Autonomous labs that run on AI guidance could distribute advanced scientific capability beyond the handful of well-funded labs that have historically driven drug discovery. This could reverse the geographic concentration of biotech innovation in China.
Ginkgo separately announced a $47 million deal with the Department of Energy to deploy 115 autonomous lab robots across Pacific Northwest National Labs as part of the Genesis mission.