Ginkgo Bioworks to develop AI models for drug development

Diving headfirst into the generative artificial intelligence market, Ginkgo Bioworks said on Tuesday it plans to develop its own AI models for drug development and other synthetic biology applications. Creating an AI model like the one underlying ChatGPT, but specialized for drug discovery, will require a vast amount of computing power. So, Boston-based Ginkgo also announced a five-year deal with Google for access to the search giant’s cloud computing and AI modeling resources. Under a five-year agreement, Ginkgo will pay Google escalating minimum amounts for its computing and AI services, starting with $8 million in the first year and rising to $113 million by the fifth year, for a total required expenditure of $289 million, according to a securities filing. Google also agreed to provide Ginkgo with up to $56 million in funding over the next three years, dependent on Ginkgo meeting undisclosed business milestones. Ginkgo said it plans to use Google’s Vertex AI software to create its own generative artificial intelligence model that could be used to suggest ways to combine proteins from its vast database. AI apps could also help Ginkgo and its customers such as Biogen and Merck more easily search through data from prior experiments and academic studies, the company said. The Google deal comes as Ginkgo is trying to expand its biosecurity business, which got started during the COVID pandemic, into monitoring wastewater for pathogens. Last year, Ginkgo brought in $334 million from its biosecurity unit, far more than the $144 million of sales from its original synthetic biology business. But with the demand for COVID testing waning, the biosecurity unit expects a sharp revenue drop this year. — AARON PRESSMAN