Training Data How Autonomous Labs Will Transform Scientific Research: Ginkgo Bioworks’ Jason Kelly
58 snips
Mar 24, 2026 
Jason Kelly, founder and CEO of Ginkgo Bioworks, who builds automated biological foundries to make biology programmable. He explains how robotic, AI-driven labs run experiments at scale and a collaboration where a reasoning model plus robotics outperformed prior biochemistry results. He describes why automating lab work, cloud labs, and shared data will accelerate discovery and reshape biotech infrastructure.
AI Snips
Chapters
Transcript
Episode notes
Shared Raw Data Multiplies Scientific Learning
- Autonomous labs let many AI scientists share raw experiment data daily, creating cross-hypothesis learning humans rarely get.
- Jason imagines 100 AIs each pursuing different Alzheimer's hypotheses and exchanging daily raw results so failures inform other lines immediately.
Robotics Flips Science Spending To Usage Pricing
- The dominant cost of current experimental science is overhead not reagents, so robotic labs can shift spending to usage-based reagents and increase data per dollar ~10x.
- Jason notes less than 5% of biopharma/NIH spend is reagents while much goes to people, lab space and underutilized equipment.
Solve Integration And Liquid Handling First
- To build autonomous labs solve two engineering problems: integrate diverse benchtop equipment and robust liquid handling for many liquid classes.
- Jason compares lab work to high-mix low-volume automation and stresses integrating thousands of third-party devices and handling viscous fluids reliably.