Bristol Myers Squibb will connect two AI clusters so that data from different studies can help guide the next experiment

Bristol Myers Squibb will connect two AI clusters so that data from different studies can help guide the next experiment

On 20 July, BMS announced the deployment of a second NVIDIA cluster. The company plans to connect it to the first cluster, giving research teams at different sites access to shared data and computing resources.

A cluster is a group of connected servers used to train artificial intelligence models and run computations. BMS has used such a system for about three years. Scientists found it difficult to access because acquisitions had left different sites with their own local rules, while using the system required specialized skills. In a NVIDIA post published on 20 July, BMS described its next step: the company intends to create a single computing environment for all its research sites.

Drug discovery begins with the selection of a biological target, such as a protein or another component of the body that a drug is intended to affect. Scientists then screen molecules, test them in the laboratory, and decide which experiment to conduct next. BMS says it already uses an approach called Predict First: a model prediction helps scientists plan a laboratory experiment before they begin it.

Once BMS connects the systems, data from a research program in Lawrenceville, New Jersey, can serve as input for models used by a team in San Diego. The company also plans to let scientists request complex predictions using ordinary language. The result of one study is intended to provide evidence for the next hypothesis in another study.

The second cluster is still being deployed, and connecting the two systems remains a BMS plan. The company has not published a comparison that would show how a shared environment affects experiment selection, program timelines, or the number of candidates that reach clinical trials.