28‐EQUI‐RFP-1660 GPU High Performance Computing Cluster

Texas A&M Engineering Experiment Station (TEES) is soliciting proposals for the design, delivery, installation, and ongoing support of a state-of-the-art, large-scale GPU High Performance Computing (HPC) cluster to support Texas IGNITE research initiatives. The procurement, identified as RFP #28-EQUI-RFP-1660, requires a fully integrated, US-based system capable of delivering at least 8 PB of NVMe storage sustainably, with aggregate read bandwidth of 2 TB/s and GPU interconnect bandwidth of 1 TB/s, scalable to support up to 1,000 GPUs with minimal architectural redesign. The system must operate within a secure multi-tenant environment using attribute-based access control (ABAC), micro-segmentation, and strict logical isolation between compute, network, and storage domains to prevent any cross-tenant data leakage or interference. All hardware and software must comply with NIST SP 800-171, FIPS 140-2 validated encryption for data at rest and in transit, and mandatory CMMC Level 2 cybersecurity certification, with third-party assessment documentation required prior to acceptance. The solution must support both containerized and traditional HPC workflows on RHEL and Ubuntu Linux, utilize a 400 Gb/s Ethernet fabric, and prioritize direct-to-chip liquid cooling while allowing air-cooled alternatives if performance parity is proven. Proposals must be submitted exclusively through the AggieBid portal by the deadline of March 31, 2026, at 5:00 PM CDT, with no exceptions for emails or other channels. Vendors are responsible for monitoring the portal for all addenda and updates. Acceptance occurs on-site at TEES facilities following a 72-hour stress test validated against NCCL, TOP500, and IO-500 benchmarks, with systems required to demonstrate consistent performance under continuous full-load conditions for one week. The contract includes a three-year base period with an option to extend up to five years, and vendors may propose alternative hosting models—including on-campus, co-location, or third-party data center arrangements under a public-private partnership. Key personnel requirements include a dedicated technical account manager, comprehensive training for internal users, documentation of all systems and processes, and a formal off-ramp strategy. The system must handle Controlled Unclassified Information (CUI), ITAR, and EAR data with appropriate access controls and reporting, and all hardware valued over $5,000 must