Enclosure No. 2C - Statement of Work.pdf

PDF 375 KB Posted

Attached to
Request for Proposal: ALCF Equinox AI System Federal contract opportunity
Solicitation number
6-B143-P-00251-00
Issued by
Department of Energy Office of Science

About this file

This is a Statement of Work (SOW) for the ALCF Equinox AI System issued by Argonne National Laboratory's Leadership Computing Facility, dated September 1, 2026 (Version 1.0).

The SOW defines requirements for a scalable, rack-scale artificial intelligence computing capability aligned with the NVIDIA AI Factory model to support U.S. Department of Energy science and the Genesis Mission. The Base System consists of sixteen (16) NVIDIA GB300 NVL72 compute racks with all associated equipment, software, installation, integration, documentation, and support necessary to deliver an operational production resource. The acquisition may take the form of either a conventional system acquisition or an alternative commercial structure approach, but the Contractor must provide a compliant Base System in both cases. The Base System shall be capable of operating as a production resource without dependence on future compute racks or expansion infrastructure. The system must be installed in the northeast corner of Building 240 at Argonne's Lemont, Illinois campus, with ALCF preferring a layout of two contiguous rows of compute racks and separate east-west network racks.

Key technical requirements include: use of NVIDIA's rail-optimized network topology for the high-speed east-west network unless alternative topology is justified; integration with ALCF's north-south Ethernet environment and global file system; support for AI training, inference, interactive, service-oriented, agent-based, and batch workloads; compute racks powered by 480V five-wire service with specified cooling and power requirements; and comprehensive software and management capabilities. The Contractor must provide five years of maintenance and support for all hardware, software, firmware, and components. Installation and integration services include delivery coordination, staging, rack placement, cabling, power and cooling connections, software installation, system bring-up, and acceptance testing. The Contractor and ALCF shall develop an Acceptance Test Plan addressing functional, performance, and stability testing no later than fifteen calendar days before delivery. The system must fit within maximum constraints of 10 MW Design Maximum IT Electrical Load, 10 MW Heat-Rejection Requirement, and under 6,000 square feet of occupied data-center floor space. Alternative commercial structures and additional capability expansions are permitted subject to SOW requirements and constraints, with potential staged deployment located 30 to 120 meters from the Base System.

View the file

Other files for this federal contract opportunity

Show all 18

On GovTribe

Work with this file on GovTribe

  • Download the original file
  • Contacts named in this file
  • Similar government files
  • Ask GovTribe AI about this file

Text version

ALCF Equinox AI System SOW v1.0. 1

Statement of Work ALCF Equinox AI System

Organization: Argonne National Laboratory Program/Facility: Leadership Computing Facility Date: September 1, 2026 Version: 1.0

Table of Contents

1 Introduction

1.1 Definitions

1.2 Safety, Security, and Site Access

2 Background and Objectives

3 Base System

3.1 Scope and Location

3.2 Compute Rack and Facilities Requirements

3.3 High-Speed East-West Network

3.4 North-South Ethernet Network and Storage Integration

3.5 Software

3.6 Installation, Project Management, Support, and Acceptance

4 Alternative Commercial Structure Approaches and Additional Capability

5 Documentation and Deliverables

6 References

1 Introduction This Statement of Work (SOW) defines requirements for a scalable, rack-scale artificial intelligence (AI) computing capability aligned with the NVIDIA AI Factory model and

ALCF Equinox AI System SOW v1.0. 2 intended to support U.S. Department of Energy (DOE) science and the Genesis Mission.

The resulting system shall be referred to as Equinox. The capability shall support an initial production deployment based on NVIDIA GB300 NVL72 rack-scale systems.

For purposes of this SOW, the Base System shall mean the required initial production deployment, which, unless otherwise revised by amendment or negotiation prior to award, shall consist of sixteen (16) GB300 NVL72 compute racks and all associated equipment, software, installation, integration, documentation, and support necessary to provide an operational production resource.

The resulting system or acquisition structure may take the form of:

1. a conventional system acquisition for the Base System; or

2. an alternative commercial structure approach that includes the Base System.

Regardless of acquisition structure or solution approach, the Contractor shall provide a compliant Base System. The Base System shall meet the applicable technical, facilities, installation, acceptance, support, security, and operational requirements of this SOW.

The resulting system may also include additional phases, capabilities, technologies, services, or commercial arrangements, subject to the requirements and constraints of this SOW. The Base System shall be capable of operating as a production resource without dependence on later installation of additional compute racks or expansion infrastructure.

The Contractor shall furnish all labor, supervision, transportation, materials, equipment, software, licenses, documentation, project management, installation services, and support necessary to deliver an operational system, except for items explicitly identified as ALCF-furnished in the resulting contract.

This SOW is intentionally performance oriented. The Contractor shall provide a complete, supportable, and operationally practical solution that meets the required outcomes. The Contractor shall clearly identify the architectures, assumptions, dependencies, exclusions, interfaces, risks, and technical tradeoffs applicable to the delivered system.

The Contractor shall also identify any specialized integration, proprietary management component, vendor-operated control element, or other solution-specific dependency that could materially affect interoperability, support, future expansion, technology evolution, continuity of operations, or transition. DOE acquisition guidance and Federal acquisition regulations encourage requirements to be expressed in terms of required results and measurable performance standards when practicable. [4] [5] [6]

ALCF Equinox AI System SOW v1.0. 3

1.1 Definitions

For purposes of this SOW, the following definitions apply:

ALCF means the Argonne Leadership Computing Facility, a DOE User Facility within Argonne National Laboratory and the facility acquiring this system.

Contractor means the entity responsible for performing the resulting contract.

Design Maximum IT Electrical Load means the calculated maximum continuous IT Electrical Load for the installed configuration, based on manufacturer-supported equipment configuration, supported maximum power settings, and the Contractor’s stated design assumptions. It is not limited to nominal, typical, average, or observed workload power.

Heat-Rejection Requirement means the calculated maximum thermal load that must be rejected from the installed IT equipment under the stated design assumptions. Unless otherwise documented by the Contractor, the Heat-Rejection Requirement shall be assumed to be substantially equivalent to the Design Maximum IT Electrical Load.

IT Electrical Load means the electrical input power required by installed information-technology equipment at the equipment or rack power-input boundary. IT Electrical Load includes compute racks, network equipment, management equipment, storage-related client equipment, and other information-technology equipment included in the installed system. It excludes building-side electrical losses, transformer losses, facility power-distribution losses, cooling-system electrical consumption, and facility power usage effectiveness.

Laboratory means Argonne National Laboratory.

Production means the operational state in which the ALCF has accepted a system, pod, rack group, service, or component for authorized scientific or operational use.

Base System has the meaning stated in Section 1 of this SOW. The Base System shall not depend on later delivery or installation of additional compute racks or other expansion equipment to provide its required production capability.

Stability Period means the mutually agreed interval during acceptance testing during which the system is operated under representative sustained load to validate reliability, operational stability, and supportability.

ALCF Equinox AI System SOW v1.0. 4

1.2 Safety, Security, and Site Access

As further described in the contract documents, the Contractor and all Contractor personnel, subcontractors, delivery personnel, installation personnel, and other representatives shall comply with applicable Argonne National Laboratory safety, security, access-control, training, badging, cyber-security, site-access, and work-control requirements, including those of its landlord. The Laboratory will identify applicable requirements in the contract, site-access process, project planning process, or other official written communication.

The Contractor shall identify any security, safety, personnel-access, remote-access, shipping, installation, commissioning, or operational requirement that may affect the schedule, staffing model, support model, or installation plan. The Contractor shall not assume that personnel, systems, diagnostic tools, management platforms, or support services will be granted access to Laboratory networks, facilities, systems, or data without Laboratory approval.

2 Background and Objectives The ALCF operates computing and data services in support of DOE science and requires an AI computing capability that can support a broad and evolving range of scientific workloads. The system will provide compute resources in support of the Genesis Mission, established through Executive Order 14363, “Launching the Genesis Mission.” The Genesis Mission seeks to accelerate scientific discovery and technological advancement through artificial intelligence-enabled approaches and coordinated use of Federal scientific data, computing resources, and research capabilities. [8]

The ALCF’s existing production computing resources are fully allocated. Additional capability is therefore required to meet growing demand for AI-enabled scientific research, development, and deployment without displacing or materially disrupting existing production workloads and commitments.

The Genesis Mission’s AI focus requires a computing environment that supports a user-interaction model that differs in important respects from traditional batch-scheduled high-performance computing (HPC) systems. While the system must support large-scale distributed training and computationally intensive workflows, it must also be capable of supporting interactive and service-oriented AI use cases. Such use cases may include model inference, agent-based scientific workflows, retrieval-augmented generation, tool-

ALCF Equinox AI System SOW v1.0. 5 using AI agents, interactive experimentation, fine-tuning, evaluation, data analysis, and AI-enabled simulation workflows.

The objectives of this SOW are for the resulting system to:

1. Provide additional AI computing capability for DOE science and the Genesis Mission.

2. Deliver a production-ready Base System as rapidly as practicable.

3. Support AI training, inference, interactive, service-oriented, agent-based, and batch workloads.

4. Use a high-performance east-west network and integrate with the ALCF north-south Ethernet environment and global file system.

5. Include five years of maintenance and support.

6. Include the possibility of an alternative commercial structure, additional capability, or phased expansion, subject to the requirements of this SOW.

3 Base System

3.1 Scope and Location

Regardless of acquisition structure or solution approach, the Contractor shall provide a Base System.

The Base System shall be capable of operating as a production resource without dependence on future compute racks or later expansion infrastructure. For a conventional system acquisition, operational responsibility for the Base System shall reside with ALCF.

For an alternative commercial structure approach, operational responsibility for the Base System may reside with the Contractor or be otherwise allocated as specified in the resulting contract.

The Base System shall use the NVIDIA rail-optimized network topology unless the Contractor provides sufficient technical justification for an alternative topology.

To take advantage of existing utilities and speed up initial deployment, the Base System shall be installed in the northeast corner of the data center. ALCF prefers a layout consisting of two contiguous rows of GB300 NVL72 compute racks, with east-west network racks in a separate nearby row. The Contractor shall identify whether this layout is acceptable for the delivered system and shall identify the minimum, maximum, and

ALCF Equinox AI System SOW v1.0. 6 recommended distance between compute racks and east-west network racks; applicable cable and optical-media requirements; and effects on latency, bandwidth, performance, serviceability, cable management, and reliability.

The data center layout below at Building 240 of Argonne’s Lemont, Illinois campus shows approximate locations for the compute racks (blue), non-compute (network, admin, etc.)

racks (orange), and any area designated for additional capability or phased expansion (green). These are general areas. Detailed locations and utility requirements shall be finalized during post-award project planning and coordination between the Contractor, ALCF, and building management.

3.2 Compute Rack and Facilities Requirements

The Base System shall use GB300 NVL72 compute racks. The Contractor shall provide a detailed rack elevation or applicable manufacturer site-planning guide for the installed configuration.

Compute Racks

Non- Compute

Area for expanded alternate business proposals

ALCF Equinox AI System SOW v1.0. 7

The documentation shall identify rack dimensions, weight, service clearances, power-entry locations, cooling-connection locations, network-entry locations, and other information necessary for delivery, installation, operation, and service.

The most readily available electrical service at the required compute-rack location is 480V, five-wire service. The compute racks shall use this service. ALCF will not provide uninterruptible power supply capacity or redundant upstream facility power for the compute racks.

For each compute-rack type, the Contractor shall identify:

1. Minimum number of required power whips.

2. Required voltage, phase, connector type, breaker requirements, and power-entry locations.

3. Typical and maximum IT electrical load.

4. Cooling-connection location, including top-entry, bottom-entry, or either option.

5. Number, type, and size of cooling connections.

6. Required supply and return conditions, including temperature range, flow rate, pressure, pressure drop, water-quality requirements, and fluid requirements.

7. Heat-rejection requirement.

8. Required cooling-distribution equipment, hoses, manifolds, quick-disconnects, leak-detection components, and other ancillary equipment.

9. Required maintenance isolation, service access, and spill-containment provisions, if any.

This facilities information is preliminary planning information. The Contractor, ALCF, and building management shall confirm final facilities characteristics, interfaces, installation requirements, and demarcation points before installation. The Contractor shall identify assumptions, dependencies, exceptions, or additional information required to validate the delivered solution.

3.3 High-Speed East-West Network

The Contractor shall provide and describe the high-speed east-west network for the Base System, including the network technology, topology, and rationale for the design.

The Contractor shall identify:

ALCF Equinox AI System SOW v1.0. 8

1. Network technology, including InfiniBand, Ethernet, or another high-performance technology.

2. Physical and logical network diagrams.

3. Switch, adapter, port-speed, cable, optic, transceiver, and network-rack requirements.

4. Pod size, rail configuration, and rack grouping, where applicable.

5. Bandwidth per compute rack and aggregate bandwidth for the Base System.

6. Network latency, blocking characteristics, oversubscription ratio, congestion-management approach, routing, and traffic-isolation approach.

7. Network-management, telemetry, diagnostics, and fault-isolation approach.

8. Compatibility with applicable AI frameworks, communication libraries, collective-communication libraries, container environments, distributed-training environments, and workload-management systems.

9. Assumptions, dependencies, limitations, and design tradeoffs.

If the NVIDIA rail-optimized topology is used, the Contractor shall identify the pod size, rail configuration, and relationship of the sixteen compute racks to the topology. If another topology is used, the Contractor shall describe the alternative and its effect on performance, latency, bandwidth, scalability, serviceability, and physical infrastructure.

ALCF prefers a layout consisting of two contiguous rows of eight GB300 NVL72 compute racks, with east-west network racks in a separate nearby row. The Contractor shall identify whether this layout is acceptable and identify the minimum, maximum, and recommended distance between compute racks and east-west network racks; applicable cable and optical-media requirements; and effects on latency, bandwidth, performance, serviceability, cable management, and reliability.

The Contractor shall furnish all equipment, software, licenses, cables, optics, documentation, and installation services necessary for the east-west network.

3.4 North-South Ethernet Network and Storage Integration

The Contractor shall describe the north-south Ethernet network connecting the Base System to ALCF services, including the ALCF global file system, user access, management, monitoring, security, and external connectivity as applicable.

ALCF Equinox AI System SOW v1.0. 9

The ALCF global file system will be mounted over the north-south Ethernet network. A separate shared storage system, parallel file system, object-storage system, or storage appliance is not required and shall not be provided unless specifically approved by ALCF.

The Contractor shall identify:

1. Aggregation-switch manufacturer, model, quantity, port speeds, and port counts.

2. Connectivity from compute racks to aggregation switches, including a network diagram.

3. Oversubscription ratio, including calculations and rationale.

4. Redundancy, link aggregation, routing, segmentation, management, and failure-domain approach.

5. Required uplinks to the ALCF Juniper/Apstra Ethernet environment, including uplink speeds, quantities, optics, and cable types.

6. Required connectivity for compute, management, provisioning, baseboard management controllers, telemetry, and other provided services.

7. Assumptions regarding ALCF-furnished addressing, routing, firewalling, DNS, DHCP, NTP, identity, monitoring, logging, and related services.

Juniper switching and Juniper Apstra integration are preferred. If the Contractor uses non- Juniper aggregation equipment or another nonstandard north-south component, the Contractor shall describe the interoperability approach. The Contractor shall provide and support all required integration, configuration, documentation, and troubleshooting necessary for interoperability with the ALCF Juniper/Apstra environment.

The Contractor shall identify any node-local storage and its intended use. Node-local storage shall not be represented as a replacement for the ALCF global file system.

3.5 Software

The Contractor shall identify all software provided for the Base System, including software required for successful operation and any optional software included for resource management, scheduling, orchestration, system management, monitoring, telemetry, networking, security, or support.

For each software component, the Contractor shall identify:

1. Product name and software provider.

2. Whether the software is required or optional.

ALCF Equinox AI System SOW v1.0. 10

3. A brief description of its function when not self-evident.

4. Licensing model, including whether it is included with hardware, separately licensed, subscription-based, or subject to recurring cost.

5. Whether it transmits data outside ALCF or requires external connectivity.

6. Whether it can operate without an ongoing operational dependency on cloud-hosted, vendor-operated, or other external services.

For a conventional system acquisition, the delivered system shall not have an ongoing operational dependence on cloud-hosted management, monitoring, telemetry, scheduling, orchestration, control-plane, or systems-of-record services.

Limited external communication for support purposes, including vendor notification or “phone home” for parts replacement, may be provided. The Contractor shall identify the purpose, data transmitted, destination, frequency, network requirements, and method for enabling or disabling such communications. Any external communication shall require ALCF approval. The Contractor shall provide such additional information regarding any cloud-hosted, software-as-a-service, vendor-operated, or externally hosted component as the Laboratory may require to support cyber-security review and approval.

3.6 Installation, Project Management, Support, and Acceptance The Contractor shall provide all installation and integration services necessary to deliver a functioning Base System, including delivery coordination, staging, rack placement, cabling, power and cooling connections, software installation, system bring-up, and acceptance-test support.

The Contractor shall provide a responsibility matrix identifying Contractor and ALCF roles and responsibilities, major dependencies, demarcation points, and acceptance criteria for power, cooling, rack placement, cabling, network handoff, global-file-system handoff, management integration, security review, remote access, testing, acceptance, and support handoff. The final responsibility matrix shall be agreed by the Contractor and ALCF.

At least fifteen calendar days before delivery, the Contractor shall provide a project plan addressing delivery, site readiness, facilities coordination, installation, integration, safety, security, acceptance, schedule, risks, communications, and production handoff. The Contractor shall provide implementation status reports at least weekly.

The Contractor shall provide five years of maintenance and support for all hardware, software, firmware, licenses, networking components, management components, and Contractor-provided cooling equipment. The Contractor shall describe support coverage, ALCF Equinox AI System SOW v1.0. 11 response and escalation procedures, replacement commitments, on-site and remote support, software and firmware updates, security updates, spare-parts strategy, and support limitations.

The Contractor and ALCF shall develop an Acceptance Test Plan no later than fifteen (15) calendar days before delivery of the first production equipment, unless ALCF approves another date in writing. The Acceptance Test Plan shall address functional, performance, and stability testing, including compute-rack operation, power and cooling integration, east-west and north-south networking, global file-system access, required software, management capabilities, representative AI workloads, support escalation, and delivered documentation.

4 Alternative Commercial Structure Approaches and Additional Capability

The resulting system or acquisition structure may take the form of a conventional system acquisition or an alternative commercial structure approach. In all cases, the Contractor shall provide a compliant Base System meeting Section 3.

If the awarded scope includes an alternative commercial structure approach, the Contractor shall describe, as applicable:

1. Asset ownership and operating responsibilities.

2. User access, allocation, capacity-reservation, and service-delivery model.

3. Security, identity, data-handling, remote-access, and vendor-access model.

4. Facilities responsibilities, dependencies, and proposed demarcation points.

5. Support commitments, acceptance approach, continuity of operations, and transition approach.

6. Contract-term, renewal, termination, and end-of-term equipment, data, software, and service disposition.

In addition to the Base System, the resulting system may include additional compute racks, accelerators, future NVIDIA technologies, other future accelerator-based technologies, networking, management equipment, cooling equipment, or related AI infrastructure. Such additional capability is not required to be limited to the same exact technology generation as the Base System. Additional capability may be proposed either as an expansion of the Base System or as a separate but interoperable system, provided the

ALCF Equinox AI System SOW v1.0. 12

Contractor clearly describes the architecture, compatibility or interoperability implications, operational model, facilities impacts, and any effect on the Base System.

The Contractor is not required to design or fully provision the Base System at initial deployment to accommodate the maximum possible future additional rack count or the full 10 MW / 6,000-square-foot deployment envelope. The Contractor may propose: (a) an initial deployment that includes infrastructure sized to support future expansion; (b) an initial deployment that would require later augmentation, re-cabling, reconfiguration, or other modification to support additional capability; or (c) a separate but interoperable additional system. The proposal shall clearly identify which approach is proposed and shall describe the associated assumptions, dependencies, later-phase infrastructure changes, outage or disruption implications, and any effects on network, cooling, power-distribution, cable-pathway, management, and operational architecture.

The resulting system provided by the Contractor, including the Base System and any additional capability, shall fit within the following collective maximum constraints:

1. A Design Maximum IT Electrical Load of 10 MW.

2. A maximum Heat-Rejection Requirement of 10 MW.

3. An occupied data-center-floor footprint of under 6,000 square feet.

The 6,000-square-foot footprint shall include all equipment and required operational space on the data-center floor, including compute, network, management, cooling-distribution, and power-distribution equipment; aisles; service clearances; and operational access space. Equipment located in adjacent mechanical, electrical, or plumbing spaces is excluded from the floor-space calculation but shall be explicitly identified, together with its requirements and dependencies.

Additional capability may require staged deployment and may be located nearer the center of the data center, approximately 30 meters to 120 meters from the Base System. If additional capability is included in the awarded scope, the Contractor shall describe:

1. Equipment, quantity, deployment stages, and expected availability.

2. Required facilities work and dependencies.

3. Whether the additional capability will operate as part of the Base System, as a separate system, or through another architecture.

4. The effect on latency, bandwidth, distributed training, collective communications, data movement, inference, scheduling, and other scientific workloads.

ALCF Equinox AI System SOW v1.0. 13

5. Required network connectivity, optics, cable pathways, management components, and operational changes.

6. Expected impact on the Base System, including required outages, re-cabling, reconfiguration, or disruption.

Operating the Base System and additional capability as separate systems is acceptable. If that approach is used, the Contractor shall describe the operational benefits and limitations of that approach.

The maximum facilities envelope is an upper constraint, not a target deployment quantity.

The Contractor shall justify the delivered quantity, footprint, deployment sequence, infrastructure design, and operational model.

5 Documentation and Deliverables The Contractor shall provide complete, current, and editable documentation applicable to the awarded scope, including:

1. Final bill of materials and rack elevations.

2. Floor plans, power and cooling requirements, and applicable site-planning documentation.

3. East-west, north-south, management, and storage-integration diagrams.

4. Cable schedules, network configurations, and Juniper Apstra artifacts, if applicable.

5. Installation, as-built, operating, maintenance, and hardware-replacement documentation.

6. Software, licensing, monitoring, telemetry, resource-management, and support documentation, as applicable.

7. Security, remote-access, facilities-interface, and demarcation documentation.

8. Acceptance Test Plan and Acceptance Test Report.

9. Documentation for any awarded expansion, capability, staged deployment, or alternative commercial structure.

ALCF Equinox AI System SOW v1.0. 14

6 References [1] NVIDIA, “NVIDIA GB300 NVL72,” NVIDIA Data Center Products, https://www.nvidia.com/en-us/data-center/gb300-nvl72/.

[2] NVIDIA, “NVIDIA Announces Blackwell Ultra AI Factory Platform To Power Age of AI Reasoning,” March 18, 2025, https://nvidianews.nvidia.com/news/nvidia-announces-blackwell-ultra-ai-factory-platform-to-power-age-of-ai-reasoning.

[3] NVIDIA, “NVIDIA AI Factory,” NVIDIA Enterprise, https://www.nvidia.com/en-us/data-center/ai-factory/.

[4] U.S. Department of Energy, “Department of Energy Acquisition Guide,” https://www.energy.gov/management/articles/department-energy-acquisition-guide.

[5] Acquisition.gov, “Federal Acquisition Regulation, Subpart 37.6—Performance-Based Acquisition,” https://www.acquisition.gov/far/subpart-37.6.

[6] Acquisition.gov, “FAR 37.602—Performance Work Statement,” https://www.acquisition.gov/far/37.602.

[7] Acquisition.gov, “DEAR Subpart 970.11—Describing Agency Needs,” https://www.acquisition.gov/dears/subpart-970.11%E2%80%94describing-agency-needs.

[8] The White House, “Launching the Genesis Mission,” Presidential Actions, November 2025, https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/.

https://www.nvidia.com/en-us/data-center/gb300-nvl72/ https://nvidianews.nvidia.com/news/nvidia-announces-blackwell-ultra-ai-factory-platform-to-power-age-of-ai-reasoning https://nvidianews.nvidia.com/news/nvidia-announces-blackwell-ultra-ai-factory-platform-to-power-age-of-ai-reasoning https://www.nvidia.com/en-us/data-center/ai-factory/ https://www.nvidia.com/en-us/data-center/ai-factory/ https://www.energy.gov/management/articles/department-energy-acquisition-guide https://www.energy.gov/management/articles/department-energy-acquisition-guide https://www.acquisition.gov/far/subpart-37.6 https://www.acquisition.gov/far/37.602 https://www.acquisition.gov/dears/subpart-970.11%E2%80%94describing-agency-needs https://www.acquisition.gov/dears/subpart-970.11%E2%80%94describing-agency-needs https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/ https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/

1 Introduction
1.1 Definitions
1.2 Safety, Security, and Site Access
2 Background and Objectives
3 Base System
3.1 Scope and Location
3.2 Compute Rack and Facilities Requirements
3.3 High-Speed East-West Network
3.4 North-South Ethernet Network and Storage Integration
3.5 Software
3.6 Installation, Project Management, Support, and Acceptance
4 Alternative Commercial Structure Approaches and Additional Capability
5 Documentation and Deliverables
6 References

File details come from the government source that posted it. Updated .