Ark Invest: AI Progress Is Accelerating, Incentivizing Huge Capital Expenditures (CapEx), & More

Last week, Ginkgo Bioworks and OpenAI made an announcement highlighting that the convergence of AI, biology, and robotics is likely to enhance productivity significantly in biotech.¹³ Together, they intend to close an important loop in the drug discovery process: without human intervention, they are using an autonomous robotic lab to run experiments, generate data, feed results into AI models, and improve the system iteratively.

Their focus is on cell-free protein synthesis (CFPS), a synthetic biology process that underpins many downstream applications. With fully automated AI-guided experimental cycles, the system identifies non-obvious combinations of reagents that human researchers have not explored. After two months and multiple rounds of synthesis, their AI CFPS costs and reagent usage has dropped by ~40% and ~57%, respectively, improving a notoriously complex and expensive system dramatically.¹⁴

This breakthrough is important for several reasons. Multivariate and data intensive, automated CFPS can actually increase, instead of lowering, the use of reagents. AI streamlines the steps involved in the development and manufacturing of novel therapeutics, increasing the speed of experimentation and generating more information at a lower cost than traditional methods.

The Ginkgo collaboration aligns with OpenAI’s interest in backing biotech companies that use AI to advance drug discovery, giving it the opportunity to collect downstream royalties.¹⁵ It also reinforces the importance of data generation and vertical integration in drug discovery. Many AI drug discovery companies maintain proprietary wet labs to create novel datasets and train models. Not everyone wants—or can afford—that vertical integration, which is creating opportunities for platforms like Twist Biosciences to provide proprietary AI-ready biological datasets as a service.¹⁶

Ultimately, the OpenAI–Ginkgo announcement is less about cost reduction and more about a blueprint: AI systems that not only analyze biology, but also run experiments, learn from outcomes, improve processes, and free up resources for further value creation.