Texas A&M needed to give researchers across seven institutions access to shared GPU infrastructure for AI and simulation, but isolated workstations, slow grant-funded clusters, and long queues were limiting drug-discovery throughput and broader scientific competitiveness.
Reported impact
VISION identified more than 22,000 drug candidates in one week versus roughly 120 from earlier efforts, raised validation hit rates from 1 percent to 10 percent up to 80 percent to 90 percent, runs at 95 percent to 98 percent utilization, and can support 500 to 1,500 concurrent users while avoiding more than $1 million of cloud cost on a single project.
Texas A&M built VISION as a shared DGX SuperPOD environment so frontier-scale compute is available as a university service instead of a lab-by-lab exception. The platform combines nearly 760 NVIDIA Hopper GPUs with DDN storage, Base Command Manager for cluster operations, DCGM telemetry for utilization visibility, Slurm scheduling, an 11-tier quality-of-service model, and dedicated Kubernetes nodes for inference. Research data and jobs from across the Texas A&M system land in the same environment, where labs can scale from interactive debugging to 240-GPU runs lasting up to 72 hours. That architecture let Dr. Reid Powell launch a 10.4 million-compound virtual screen within minutes of receiving access, and it gives the wider university a repeatable way to support drug discovery, genomics, AI training, and classroom workloads at production scale.
University projects and datasets are staged into DDN storage, scheduled through Slurm and the quality-of-service framework, and executed on the DGX SuperPOD across Hopper GPUs. Results flow back to researchers for compound triage, inference, education, and follow-on experiments across the Texas A&M system.
22,000+ drug candidates identified in a single week versus ~120 in all prior efforts combined
Powell's first run on VISION completed 10.4 million molecular simulations in roughly one week-a workload he estimates would have cost over $1 million in rented cloud-based technology.
The DGX SuperPOD operates at 95%-98% utilization with nearly 760 NVIDIA Hopper GPUs.
Validation hit rates jumping from 1%-10% to 80%-90% by running high-precision AI models on NVIDIA DGX SuperPOD
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
NVIDIA Sample Oct 6, 2026 University of Pennsylvania Penn needed shared AI infrastructure that could support researchers across many disciplines without weeks-long wait times, fragmented storage, or the cost barriers that kept frontier-scale training and inference off… 1.5 years of compute time reduced to three days for a 40,000-document project NVIDIA DGX B200 systems · NVIDIA DGX SuperPOD · NVIDIA InfiniBand NVIDIA Sample Aug 3, 2026 StudyFetch StudyFetch needed to keep AI tutoring affordable while scaling live lecture transcription and personalized learning for millions of students, but off-the-shelf cloud transcription created cost and latency ceilings that… 100M+ learning interactions NVIDIA H100 GPUs · NVIDIA L40S GPUs · NVIDIA NIM NVIDIA Sample Jul 30, 2026 Stanford University Stanford needed a campus-wide GPU platform that could support interdisciplinary AI research at production scale without forcing labs to depend on fragmented clusters, complex allocations, or external resources for large… 18 million GPU hours NVIDIA Base Command Manager · NVIDIA DGX SuperPOD · NVIDIA Hopper GPUs 