The pharmaceutical industry’s newest competitive advantage may no longer be larger laboratories—it may be owning the computing power capable of discovering tomorrow’s medicines. (Source: Image by RR)

High-Performance Computing Drives the Next Phase of Biomedical Innovation

Bristol Myers Squibb has partnered with NVIDIA to build what is expected to become the pharmaceutical industry’s largest privately owned AI supercomputer. Scheduled to come online in early 2027 and hosted by Equinix, the new infrastructure will be powered by NVIDIA DGX Vera Rubin NVL72 systems and will support advanced AI research across Bristol Myers Squibb’s drug discovery and development pipeline.

Rather than relying primarily on rented cloud computing, Bristol Myers Squibb concluded that owning dedicated AI infrastructure offered greater long-term value. Company leaders determined that cloud reservations often provided access to older hardware while carrying significantly higher operating costs. By investing directly in its own computing platform, the company also keeps proprietary scientific data and intellectual property within infrastructure it controls.

The supercomputer will provide researchers access to NVIDIA’s BioNeMo Agent Toolkit, allowing scientists to build predictive models, train specialized AI systems, and accelerate workflows involved in identifying and developing new medicines. According to Bristol Myers Squibb, AI is already producing measurable results across its research organization, reinforcing the company’s belief that high-performance computing has become as essential to pharmaceutical innovation as traditional laboratory equipment.

The project also reflects a broader industry trend. Major pharmaceutical companies, including Eli Lilly and Company and Roche, have recently invested in their own AI computing infrastructure rather than relying exclusively on cloud services. As AI becomes increasingly central to biomedical research, computing capacity is rapidly evolving from an outsourced utility into a strategic competitive advantage.

read more at interestingengineering.com