National Laboratory of the Rockies Energy Systems Integration Facility High Performance Computing Request for Information

Request for Information (RFI)

This RFI is issued by the National Laboratory of the Rockies (NLR) to gather market information regarding available technologies for a replacement high-performance computing (HPC) system, ESIF-HPC-4. This new system is intended to meet the mission needs of the Department of Energy's Office of Critical Minerals and Energy Innovation (CMEI) for the 2028–2032 timeframe. The system must be compatible with the ESIF data center, balancing energy efficiency and computational performance.

NLR is seeking information on systems with the following characteristics:

• GPU-accelerated compute: Majority of compute cycles, supporting AI/ML and traditional HPC.
• CPU-based compute: Smaller portion (<15%) for non-GPU accelerated workloads.
• Storage: High-bandwidth, low-latency performance tier and high-durability, cost-efficient capacity tier (object store).
• High-speed interconnect: Low-latency, high bandwidth for multi-node applications.
• Consumption-based compute augmentation: Optional capability for on-premise or off-premise resources.

NLR also seeks information on cooling capabilities (warm-water cooled data center), power distribution (prefers 480V three phase), and preferred cooling methods (prefers full Direct Liquid Cooling).

The system will support a mix of AI/ML workloads (training, inference, LLMs, generative AI), modeling and simulation (large-scale, tightly coupled jobs, high throughput ensembles), and hybrid/data-intensive workloads (digital twins, distributed data analytics).

This is a market research effort, not a Request for Proposal (RFP). Responses are due by June 17, 2026, 5:00 PM MST. Specific information requested includes:

• Feedback on draft Technical Specifications and Benchmark Instructions.

• Anticipated subcontract execution date to meet a December 2027 delivery.

• Proposed payment structures (e.g., lease-to-own, milestone-based).

• Description of early access capabilities for system components.

• Information on cooling functionality and waste heat capture.

• Power distribution characteristics.

• Details on hybrid cooling architectures.

• Information on emerging power and cooling technologies.

• Feedback on consumption-based cloud compute models and cost models.

• Solutions for storage needs exceeding dense water-cooled rack capacity.

• Anticipated RFP Release Date: November 2026

• Anticipated RFP Proposal Due Date: January 2027

• Anticipated Contract Award: June 2027

• Anticipated Phase 1 Delivery: December 2027

• Anticipated Phase 1 Acceptance: May 2028

• Potential Phase 2 Delivery: May 2029

• Anticipated Phase 2 Acceptance: December 2029

Not specified in this RFI. NLR reserves the right to set aside in total or in part any resultant award in a future solicitation.

• Responses should be submitted as an Adobe Acrobat PDF or Microsoft Word document via email to ASP.Responses@nlr.gov by the deadline.
• All documents referenced are in draft form and subject to change.
• Costs associated with response development are not reimbursed.
• This RFI does not guarantee a future RFP or award.

Solicitation / RFP / RFQ

This document outlines the technical requirements for the next-generation High Performance Computing (HPC) system, ESIF-HPC-4, to be installed at the National Laboratory of the Rockies (NLR) Energy System Integration Facility (ESIF). The system is intended to support the mission computing needs of the Department of Energy's Office of Critical Minerals and Energy Innovation (CMEI), supporting a diverse workload including traditional modeling and simulations, artificial intelligence (AI), and machine learning (ML).

• System Architecture: The system is logically segmented into four zones: High-Performance Compute Zone, Access Zone, Storage Zone, and Management Zone.

• Compute Nodes: Detailed specifications for processor, memory, network interface, bandwidth, latency, and high-bandwidth memory (HBM) are required. CPU-only partitions should not exceed 10% of total peak FLOP capacity.

• Interconnect: High-speed network requirements include topology, bandwidth, latency, resilience, and quality of service (QoS) mechanisms.

• Storage: A high-performance parallel file system (minimum 50PB capacity, 500Gb/s sustained performance) and an AI-optimized storage tier (minimum 50PB capacity, 100 GB/s sustained performance) are required.

• Access Zone: Must support 400 interactive users, 60 data analysis front-end applications, 150 integrated development environments, and 100 simultaneous compilations.

• Management Zone: Includes tools, networks, and services for system configuration, operation, monitoring, and maintenance.

• Benchmarks: Offerors must provide benchmark results for application, microbenchmark, and I/O tests, categorized as baseline, ported, or optimized.

• Period of Performance: Maintenance and support for a minimum of five (5) years following system acceptance is required, with options for years 6 and 7.

• Guidance for Offerors: Offerors can propose on-premise or off-premise solutions. Responses should address applicable requirements and options. Requirements are designated as Mandatory (MR), Target (TR-1, TR-2, TR-3), or Technical Option (TO-1, TO-2, TO-3).

• Schedule: Tentative schedule includes RFP release in Q4 FY26, proposals due Q1 FY27, early access hardware delivery in FY27, full system delivery in Q1 FY28, and system acceptance in Q2 FY28.

• Not explicitly stated in this document.

• The document includes detailed requirements for facilities and site integration, power, cooling, delivery, installation, maintenance, support, project management, training, and documentation.

• Offerors may propose technical options for scaling the system, processor memory, interconnect, mid-life upgrades, advanced architectures, and off-premise cloud resources.