Ginkgo Bioworks is getting into the large language model game by inking a five-year artificial intelligence partnership with Google Cloud.
The company hopes to turn the type of technology that underpins programs such as ChatGPT toward synthetic biology by training models on genomic and protein data instead of internet images or the written word.
Under the deal, Ginkgo will make Google Cloud its primary computing services provider, while Google Cloud will provide an unspecified amount of funding “to help Ginkgo achieve certain milestones over the next three years,” the companies said in their announcement.
“Our strategic partnership with Ginkgo is a first-of-its-kind for Google Cloud, underscoring our confidence that Ginkgo will play a critical and pioneering role in the life sciences space, leveraging AI to reshape humanity's understanding of biology,” Google Cloud CEO Thomas Kurian said in a statement.
The scope of Ginkgo's work includes drug discovery, agriculture, industrial manufacturing and biosecurity. The company has set high goals to develop “a number of interconnected models” that it plans to employ both internally, through R&D services provided to its own customers, and to offer others externally via Google Cloud’s AI marketplace.
Ginkgo said it maintains a database of more than 2 billion unique protein chains, which it describes as well suited for training AI models. As such, the company said it expects its first project to focus on proteins—with applications in generative design and sequence optimization, plus bioengineering work aimed at constructing therapeutics or production methods for small molecules and gene therapy vectors.