Transformative Expandable Supercomputing & AI Infrastructure

Overview

The U.S. Army Engineer Research and Development Center (ERDC) is seeking commercial solutions capable of delivering hybrid architectures that seamlessly combine traditional High-Performance Computing (HPC) systems—optimized for physics-based modeling and simulation—with the specialized tensor and parallel processing power required for modern Artificial Intelligence (AI) and Machine Learning (ML) workloads.

Problem Statement

Lengthy lead times for next-generation accelerators often mean systems are outdated shortly after deployment. The rapid pace of AI and HPC hardware advancement exceeds traditional government acquisition timelines.

Background

The Department of War (DoW) requires the ability to rapidly stand-up mirrored computing environments across different security enclaves without enduring the lengthy design phases of a new system. The government requires an agile supercomputing solution that accelerates delivery of initial compute capacity (IOC), coupled with an engineered pathway for massive scale-out, hardware upgrades, and exact-copy replication on classified networks.

To maintain a decisive technological advantage, DoW research laboratories require continuous access to dedicated state-of-the-art compute. Traditional monolithic supercomputers are rigid; expanding them or upgrading internal components is often cost-prohibitive or engineered out of the initial design. By leveraging an innovative, modular baseline architecture, ERDC intends to establish a flexible ecosystem where compute capacity, storage capacity, and security enclaves can scale dynamically on demand, supported by the latest generation computing hardware.

Project Objective

This Individual Program Requirement (IPR) seeks to identify “baseline” systems designed from the ground up for rapid Initial Operating Capability (IOC), massive scalability (e.g., thousands of CPUs/GPUs), in-place upgradability, and rapid replication across multiple classification levels, including Impact Level 7 (IL7) networks. ERDC will gain comprehensive understanding of industry capabilities that bridge the architectural gap through a unified, dynamically expandable architecture. Solutions should allow for fluid resource orchestration across computing paradigms without degraded performance. The final deliverable must be designed, integrated, and deployed as a highly flexible and upgradable supercomputing architecture that fundamentally breaks away from static deployment models.

Project Manager

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

Requirements

Submissions should consider the design, integration, and deployment highly flexible and upgradable supercomputing architecture that fundamentally breaks away from static deployment models and address the following key areas:

  • Hybrid Compute Architecture & Accelerated IOC: Solutions must deliver a dual-partition architecture integrating traditional HPC (with a preference for x86 CPUs) and AI-optimized, GPU-dense clusters. Anticipating a final scale of thousands of CPUs and GPUs, proposals must detail an innovative deployment strategy that accelerates IOC, providing usable partial compute capacity to end-users during the ongoing system build-out.
  • In-Place Expandability (Scale-Up & Scale-Out): The architecture must be engineered for frictionless, dynamic scaling of hardware and software resources with minimal operational downtime. This requires designed headroom for integrating additional node boards into partially populated racks, as well as seamless topological integration of net-new compute cabinets to support evolving DoW mission requirements.
  • High-Bandwidth Interconnect: Solutions must incorporate a high-bandwidth, low-latency network fabric (e.g., 400Gbps generation technologies or superior) that supports linear scaling across the cluster without degrading parallel workload performance.
  • Upgradability & Future-Proofing: The foundational infrastructure must support modular, “plug-and-play” integration of next-generation accelerators and GPUs as supply chain availability permits. Proposed solutions must actively minimize or eliminate the need for complete “forklift” upgrades over the system’s lifecycle.
  • Expandable Storage Ecosystem: Solutions must integrate a high-performance parallel file system engineered for dynamic expansion in both overall capacity and bandwidth. This storage scaling must align parallel to compute node expansion and be executable without necessitating extended system downtime.
  • Rapid Replicability & Classification Agility: The baseline architecture must function as a standardized, repeatable template. Solutions must demonstrate the capability to rapidly manufacture and deploy identical iterations of this baseline onto isolated, highly classified networks (e.g., Special Access Program [SAP] enclaves). All hardware and firmware must strictly adhere to DoW security baselines and Supply Chain Risk Management (SCRM) protocols to expedite rapid system accreditation.
  • Advanced Liquid Cooling: Proposed systems must capture and remove a minimum of 98% of IT equipment-generated heat via water cooling. While there is no strict maximum per-rack power limit, solutions must exclusively employ Direct Liquid Cooling (DLC) or Rear Door Heat Exchangers (RDHx) to satisfy this critical thermal capture metric.

Constraints and Considerations:

  • Site Deployment and Facility Adaptability: Solutions should consider systems will be fielded across multiple distinct facilities within the ITL campus located in Vicksburg, MS. Proposed solutions must demonstrate modular adaptability to varying facility footprints while strictly sustaining the mandated 98% liquid cooling capture metric.
  • Turnkey Systems Integration: Solutions should consider a fully integrated (physical and logical) system. This encompasses all hardware components, high-bandwidth network fabrics, and parallel file systems, delivering a fully operational, orchestrated environment up to the Operating System (OS) level.
  • Supply Chain Risk Management: All proposed hardware and firmware components must strictly adhere to the Trade Agreements Act (TAA) and DoW supply chain security mandates requisite for systems destined for classified information processing.
  • STIG Compliance and Vulnerability Management: Solutions must implement applicable DISA Security Technical Implementation Guides (STIGs) across all architecture layers (e.g., network devices, internal databases, web servers). For Unix-variant operating systems lacking distribution-specific STIG, the general Unix STIG shall be applied. All deviations from STIG mandates must be formally documented and mitigated via a Plan of Action and Milestones (POAM) utilizing ISSM-approved formats (https://www.cyber.mil/stigs).
  • Data at Rest (DAR) Encryption: All provided nonvolatile storage media within the architecture must natively support full disk encryption utilizing cryptographic ciphers explicitly approved for the protection of National Security Information.
  • Security Clearances: Personnel supporting the integration, deployment, and ongoing maintenance of these systems may be required to be cleared to operate within a Top Secret/Sensitive Compartmented Information (TS/SCI) environment, depending on the facility location where the solutions will be deployed. All personnel requiring administrative system access must comply with the latest versions of DoD Directive 8140.01 and DoD 8570.01-M and must have Information Assurance Technician (IAT) Level II or IAT Level III certification. Per DoD 8570.01-M, DFARS 525.239.7001 and AR 25-2, the Contractor employees supporting IA/IT functions shall be appropriately certified prior to being granted administrative access to systems on the Government’s network.
  • System Acceptance Testing: Solutions should consider Final Government Acceptance of a system is contingent upon the successful execution of comprehensive functionality testing.
  • Lifecycle Maintenance and SLAs: Solutions should consider including comprehensive hardware maintenance and support through Service Level Agreements (SLA).

Applicants must be registered on SAM.gov.

Note: Submissions should NOT include confidential or proprietary details. Submissions may be shared with other ERDC teams if there is an apparent fit with other ERDC projects. 

Estimated Government Funding Profile 

Up to $2,000,000 may be awarded for a solution.

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

Estimated Period of Performance

The period of performance will be determined based on the capability contained in the submission. Submissions should contain timelines detailing the delivery of the accelerated IOC, the completion of the fully scaled baseline deployment, Service Level Agreements (SLAs) and the timelines associated with any future expansion modules that are included.

Desired End State

ERDC gains an understanding of the available industry solutions related to the successful rapid delivery of a highly agile, liquid-cooled supercomputing ecosystem that allows rapid deployment of immediate capability today, scale to massive CPU and GPU density seamlessly, upgrade to next-generation accelerators as supply chains permit and rapidly clone the entire architecture into classified environments.

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.

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:

June 25, 2026Project Announced, Submissions Open
July 13, 2026Question Period Ends
July 24, 2026Submissions Close
July 1-31, 2026ERDC Conducts Reviews
July-Aug 2026Downselected Vendors Notified

*If needed; dates may vary to accommodate 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

How to Participate

Qualified parties may submit by completing a submission form and uploading required documentation as defined in the CSO Solicitation document.

1. Review CSO Solicitation document
2. Review FAQs
3. Complete the submission form

Submission Instructions:
This solicitation is issued consistent with the authority granted to the U.S. Army Engineer Research and Development Center (ERDC) through the establishment of its Commercial Solutions Opening (CSO), W912HZ26SC006. Submissions must follow the requirements as detailed in the CSO Solicitation document.

Questions:
Interested parties may submit questions using this form until July 13, 2026.

Submissions must meet stated requirements and be received no later than 4:00 pm CT on Friday, July 24, 2026.

ERDC ITL is conducting this project announcement on a full and open basis and intends to award contracts in accordance with FAR part 12 and the FAR part that is deemed most appropriate for the solution proposed (i.e. FAR part 13, 15, and/or 35); the government reserves the right to award prototype agreements (e.g. Other Transaction Agreements), in accordance with 10 U.S.C. 4022, if deemed appropriate and in the government’s best interest.

Share This Project