Innovative Technology Insertion of Advanced HPC Systems

Overview

The U.S. Army Engineer Research and Development Center (ERDC) High Performance Computing Modernization Program (HPCMP) seeks innovative solutions to establish a new high-performance computing (HPC) system acquisition and support services model that leverages industry best practices to transform how the HPCMP procures, operates, and maintains its newest HPC systems for 2027.

Background

Each year the HPCMP exercises its Technology Insertion (TI) process to acquire the most advanced HPC systems based on requirements derived from performance goals, expected workloads, and anticipated classification needs. The TI systems are delivered, installed, accepted, operated, and maintained in the Department of War (DoW) Supercomputing Resource Centers (DSRCs). A rotating schedule among the DSRCs ensures a balance of older and newer systems across the program. A mix of heavy compute-based and artificial intelligence (AI)/machine learning (ML)-intensive architectures is also considered, as well as an appropriate division of unclassified and classified capabilities. Read more about the project background in the CSO Solicitation document.

This project aims to identify innovative solutions to establish a new HPC system acquisition and support services model that leverages industry best practices to transform how the HPCMP procures, operates, and maintains its newest HPC systems for 2027. The HPC systems being considered are a combination of compute-intensive architectures and AI/ML platforms operating in either unclassified or classified (secret and above-secret) environments. The systems would be geographically distributed at up to three of the five HPCMP DoW DSRCs:

• Air Force Research Laboratory (AFRL) DSRC, Wright-Patterson Air Force Base, (WPAFB), OH
• Navy DSRC (under the direction of Commander, Naval Meteorology and Oceanography Command–CNMOC), Stennis Space Center, MS
• Maui High Performance Computing Center (MHPCC) DSRC, Kihei, Maui, HI

Since the systems will reside on-premises in the DSRCs, the services model must also conform to DoW and local service policies, especially regarding security (physical, personnel, and cyber). See Appendix B, C and D in the CSO Solicitation document for facilities’ descriptions.

Project Manager

Information Technology Laboratory (ITL), U.S. Army Engineer Research and Development Center (ERDC)

Requirements

For the 2027 TI cycle (known as TI-27), the HPCMP invites interested participants to propose solutions that address the following HPC system acquisitions and support services to meet the overall objectives:

• Enterprise System Acquisition:
  • Offerings of HPC systems at one to three of the following DSRCs–AFRL, MHPCC, and Navy. Based on total Government investment funds anticipated, the offerings are anticipated to be equal investments for the AFRL and Navy systems and less than 10% of the total investment cost for the MHPCC system. See Appendix A in the CSO Solicitation document for potential funding and capability scenarios.
  • Consider combinations of heavy compute-based systems (most of the nodes containing high-performance CPUs) and AI/ML architectures. Also consider an appropriate addition of special nodes for each system to include cluster management nodes, user login nodes, visualization nodes (hardware accelerated graphics), data transfer nodes, and any other nodes needed to support special functions. Include high-speed interconnects across the HPC clusters and high-performance file systems needed for temporary workspace.
  • Provide a small test and development system that aligns with the architecture of each of the largest production systems for the purpose of testing system updates prior to implementation on the production systems.
  • Support protection of sensitive data or classified environments such as Government-approved encryption solutions.

• Enterprise System Administration:
  • Approaches that recognize the enterprise perspective of performing system administration of multiple HPC systems at multiple DSRCs in multiple classification environments.
  • Interested in solutions that reduce touch labor, reduce system downtime, maximize resource utilization (CPU/GPU/memory), guarantee consistent system performance, and present unified and consistent user experiences for all systems across the DSRCs.
  • The Navy DSRC has very specific interest in solutions that maintain high availability (two-hour response time) in support of unique user requirements.

• Enterprise System Maintenance:
  • Hardware and software maintenance (outside included warranties from the OEM) that results in rapid response and resolution of software patches or hardware repairs, preferably within one business day for routine maintenance occurrences.
  • Strong desire for solutions that focus on updating, replacing, or repairing identified components while not impacting operations of the remainder of the system.
  • Interested in solutions that leverage common hardware and software among several similar/same system architectures purchased from the same OEM. Seeking solutions that perform maintenance for nominally 5-7 years per HPC system at an affordable cost.

Constraints and Considerations

Review CSO Solicitation document for constraints and considerations.

Estimated Government Funding Profile

The Government budgets approximately $57,000,000-$62,000,000 per year for all TI system purchases. Read more about the funding profile in the CSO Solicitation document.

This project does not commit ERDC or ERDCWERX to pay any costs incurred in preparation of a response or guarantee a contract.

Estimated Period of Performance

The period of performance for awarded system acquisitions will be from the date of the award to the agreed-upon acceptance date. The period of performance for any selected system administration and maintenance efforts will start after system acceptance and be for one year with options (starting after system acceptance) to extend up to an additional six years in 1-year increments.

Expected Result

The desired end-state is a new TI system acquisition and support model that adopts a cost-effective, efficient, and reliable enterprise approach covering multiple HPC systems operating at multiple sites within various classification environments. A model that adopts cost-attractive, large quantity purchases and effective partnerships among the administrator/maintainer, OEMs, other third-party supporters, and the government is highly desirable.

Submissions may be shared as appropriate with other stakeholders. The government has the authority to decline all submitted proposals.

Evaluation Criteria

Submissions will be evaluated based on the criteria described in the CSO Solicitation document. Additional criteria specific to this project are listed below:

• Technical requirements will assess how innovative the solution is (as defined in this announcement) and the feasibility of the solution solving the agency’s challenges.
• Importance to agency programs will assess the solution’s potential to enhance the mission effectiveness of the agency.
• Funds availability will assess the availability of funding to procure the solution.

Additional evaluation criteria:

• Applicability: Ability to support traditional and AI/ML workloads in an HPCMP context
• Scalability: Potential to apply across the DSRCs
• Impact: Expected benefits and outcomes for the DoW
• Utility: Potential to minimize downtime, maximize performance of HPC systems

All resultant contracts will be firm-fixed price. All items, technologies, and services (including research and development) procured via this CSO are treated as commercial. Applicants from universities and/or non-profit organizations should be aware that commercial clauses will be integrated into the award and should coordinate proposals with associated legal counsel prior to submission.

Notional Project Schedule

Proposed project milestones include:

September 1, 2026	Project Announced, Submissions Open
October 2, 2026	Question Period Ends
October 23, 2026	Submissions Close

*Dates may vary to accommodate the project team and participant availability. The government may accelerate the pre-proposal review/feedback timeline, and therefore also require earlier delivery of full proposals.

Project Security Classification

Unclassified