At JPMorgan’s 44th Annual Healthcare Conference, NVIDIA’s Vice President and General Manager of Healthcare, Kimberly Powell, detailed the company’s accelerated computing initiatives across imaging, genomics, life sciences and drug discovery. Speaking for the seventh consecutive year, Powell emphasized that NVIDIA’s AI semiconductor platforms now power over 75% of the world’s large language model training workloads and are being deployed in more than 50 major hospital systems for diagnostic imaging acceleration and real-time analytics. She noted that the company’s end-to-end healthcare stack—spanning its DGX compute servers, Clara imaging framework and BioNeMo molecular modeling library—has driven a 40% reduction in model training times and a 30% increase in throughput for genomic sequencing pipelines over the last year.

In a move to streamline procurement for hyperscale cloud providers and research institutions, NVIDIA announced that customers ordering its next-generation H200 AI processors will no longer be required to submit upfront payments. The policy change, confirmed in a statement to Reuters, allows clients to reserve capacity against their existing credit agreements and defer payment until delivery or deployment. Industry analysts estimate that this shift could unlock an additional $1.2 billion in orders from top cloud operators over the next two quarters, while reducing capital outlay barriers for mid-tier AI startups.

Building on its healthcare focus, NVIDIA revealed terms of a joint $1 billion investment with Eli Lilly to create an AI-driven supercomputing platform for accelerated drug discovery and manufacturing optimization. The collaboration will integrate NVIDIA’s DGX SuperPOD clusters—capable of 4 exaflops of AI performance—with Lilly’s molecular libraries, targeting a reduction in lead candidate identification time from 18 months to under 9 months. Executives forecast that this initiative could address a market now valued at $105 billion for drug discovery and create a recurring revenue stream representing up to 10% of NVIDIA’s data center segment by 2027.