LitCoin Server Hardware Cost Variance

This request is for a revision and approval is being requested for additional amount of $195,903.27. Acquisition of a high-performance computing infrastructure package consisting of two Dell PowerEdge XE7745 graphics processing unit servers and four Dell PowerEdge R6725 compute servers, including eight NVIDIA H200 graphics processing units, high-memory configurations, high-speed networking adapters, storage devices, integrated management capabilities, rack mounting components, and three years of Dell manufacturer support and warranty services. The contractor shall provide and deliver a fully configured high-performance computing infrastructure consisting of two graphics processing unit servers, four compute servers, high-speed networking components, storage devices, integrated system management capabilities, rack mounting hardware, and three years of manufacturer support and warranty services. Upon installation, the infrastructure will provide a secure, scalable, and government-managed computing environment capable of supporting large-scale literature analysis, biomedical knowledge extraction, knowledge graph generation, artificial intelligence and machine learning workloads, and data-intensive research activities. The expected outcome is the successful migration and long-term operation of the LitCoin Knowledge Extraction Environment on government-owned infrastructure, ensuring continuity of operations, preservation of prior federal investments, compliance with federal security requirements, and reduced dependence on recurring cloud computing services.

The requirement was previously approved as PCA-NCATS-04186 for a total amount of $565, 452.18. The vendor's total proposed cost is $761,355.45. The vendor quote exceeds the Independent Government Cost Estimate of $565,452.18 and amount originally approved by $195,903.27 (34.65 percent). The variance is attributable to the inclusion of advanced graphics processing units, expanded memory and networking capabilities, enterprise-class infrastructure components, and three years of manufacturer support required to meet current and future artificial intelligence, machine learning, knowledge extraction, and biomedical data processing requirements. Despite the variance, the proposed solution provides significantly greater computing capacity and long-term value than originally estimated and remains a cost-effective alternative to equivalent commercial cloud computing resources. Therefore, approval for additional amount of $195,903.27 is being requested.