TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 1 of 31

REQUEST FOR PROPOSAL

GPU High Performance Computing Cluster

RFP # 28-EQUI-RFP-1660

PROPOSAL MUST BE RECEIVED BEFORE:

2:00 p.m. CST on Thursday, March 19, 2025

NOTE: PROPOSALS must be submitted through Aggiebid.

https://bids.sciquest.com/apps/Router/PublicEvent?CustomerOrg=TAMU

Please plan ahead and submit your proposal early just in case you have technical issues.

REFER INQUIRIES TO:

Danielle Toler, CTCD

Buyer III

Texas A&M Engineering Experiment Station

Email: danielle.toler@tamu.edu mailto:danielle.toler@tamu.edu

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 2 of 31

Table of Contents

SECTION 1. INTRODUCTION 5

1.1 Introduction

1.2 Brief History and Current Organization

1.3 Scope of Work

1.4 Award

1.5 Contract Term

1.6 Schedule of Events

SECTION 2. REQUIREMENTS 8

2.1 Technical Proposal

2.2 References

2.3 Implementation…

2.4 Acceptance of Work ………………………………

2.5 Progress Reports ……………………………………………………………

SECTION 3. GENERAL INFORMATION 13

3.1 Submittal Deadline and Location

3.2 Submittal Instructions

3.3 TEES Contacts

3.4 Inquiries and Interpretations

3.5 Open Records

3.6 Terms and Conditions

3.7 Proposal Components

SECTION 4. GENERAL TERMS AND CONDITIONS 15

4.1 General

4.2 Final Review and Approval

4.3 Definitions

4.4 Time of Performance

4.5 Default

4.6 Termination

4.7 Agreement Amendments

4.8 Independent Contractor

4.9 Right to Audit

4.10 Sales and Use Tax

4.11 Observance of TEES Rules and Regulations

4.12 Non-Disclosure

4.13 Publicity

4.14 Severability

4.15 Non-Waiver of Defaults

4.16 Governing Law and Venue

4.17 Ownership of Created Works ……………………………………………………………………………17

4.18 Non-Assignment ……

4.19 Representations and Warranties…

4.20 Non-Waiver Privileges and Immunities…

SECTION 5. CONTRACTUAL REQUIREMENTS 18

5.1 Texas Public Information Act

5.2 I0nsurance Requirements

5.3 Indemnification

5.4 Other Benefits

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 3 of 31

5.5 Dispute Resolution

5.6 Prohibition On Contracts Related to Persons Involved in Human Trafficking

5.7 Not Eligible for Rehire

5.8 Delinquent Child Support Obligations…………………………………………………………………...19

5.9 Payment of Debt or Delinquency to the State……………………………………………………………19

5.10 Loss of Funding ……

5.11 State Auditor’s Office Right to Audit…

5.12 Products and Materials Produced in Texas…

5.13 Certification Regarding Boycotting Israel

5.14 Certification Regarding Business with Certain Countries and Organizations

5.15 Verification Regarding Boycotting Energy Companies…

5.16 Verification Regarding Discrimination Against Firearm Entities and Trade Associations…

5.17 Billing Resolutions

5.18 Payment Terms/Prompt Payment

5.19 Force Majeure…

5.20 VetHUB Subcontracting Plan…

5.21 HIPAA…

5.22 Export Compliance

5.23 Compliance with Laws

5.24 Use of Name…

5.25 Prior Employment…

5.26 Access by Individuals with Disabilities…

5.27 Access to Agency Data…

5.28 Cloud Computing Services…

5.29 Cybersecurity Training Program

5.30 Data Privacy…

5.31 Executive Order GA-48…

5.32 Executive Order GA-43……

5.33 Survival…

5.34 Suspension and Debarment…

5.35 Excluded Parties…

SECTION 6. EXECUTION OF OFFER 23

6.1 Proposer Affirmation

6.2 Texas Family Code Section 231.006

6.3 Substitute W-9

6.4 Direct Deposit

6.5 Signature

SECTION 7. RESPONDENT’S QUESTIONAIRE/COMPANY EXPERIENCE 25

SECTION 8. PRICING 26

SECTION 9. EVALUATION CRITERIA FOR AWARD 27

9.1 Evaluation Information

9.2 Demonstration/Presentation

9.3 Evaluation Criteria and Weights

SECTION 10. NON-COLLUSION AFFIDAVIT 28

SECTION 11. INSURANCE REQUIREMENTS 29

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 4 of 31

SECTION 12. VetHUB SUBCONTRACTING PLAN 31

ATTACHMENT A – VETHUB SUBCONTRACTING PLAN (HSP) FORM

ATTACHMENT B – SERVICES AGREEMENT (DRAFT)

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 5 of 31

SECTION 1:

INTRODUCTION

1.1 Introduction

The Texas A&M Engineering Experiment Station (“TEES”) is seeking proposals from qualified vendors (also referred to as

“Respondents”) to design, deliver, install, and support a state-of the art, large-scale GPU High Performance Computing (“HPC”) cluster for Texas IGNITE. This cluster will be a secure, multi-tenant, and highly performant ecosystem designed to accelerate research across a wide spectrum of disciplines and use cases from conducting large-scale, high-precision scientific computations to training large language models. The Texas IGNITE cluster will serve as the cornerstone of TEES’ research and innovation initiatives, enabling breakthroughs in hypersonics, energy, national security, healthcare, robotics, artificial intelligence (“AI”), biotechnology, materials, manufacturing, and other computationally intensive fields.

Successful implementation of the Texas IGNITE cluster will transform the agency’s research capabilities. By providing a secure, scalable, and accessible resource, the platform will enable existing research to be done faster and at scale, will open doors to entirely new, research paradigms with high impact, and will help maintain Texas’ competitive advantage in the global AI race, driving economic growth, and enhancing national security through technological advancements.

Proposed solution must be a turn-key cluster, fully integrated, tested, and workload-ready upon delivery.

By submitting proposals, each respondent certifies that they understand this Request for Proposal (“RFP”) and has full knowledge of the scope, nature, quality, and quantity of work to be performed, the detailed requirements of the services to be provided, and the conditions under which the services are to be performed. Each respondent also certifies that it understands that all costs relating to preparing and responding to this RFP will be the sole responsibility of the Respondent.

RESPONDENTS ARE CAUTIONED TO READ THE INFORMATION CONTAINED IN THIS RFP CAREFULLY

AND TO SUBMIT A COMPLETE RESPONSE TO ALL REQUIREMENTS AND QUESTIONS AS DIRECTED.

1.2 Brief History and Current Organization

About Texas IGNITE:

Texas IGNITE: Innovative Growth in Next-gen AI Technology Ecosystems is a comprehensive initiative designed to position

Texas as a leader in artificial intelligence (“AI”), machine learning (“ML”), and high-performance GPU computation driven innovations. This project aims to establish a robust high performance AI technology infrastructure at TEES, including a high-performance computational facility with next-generation Tensor Core GPUs . The project will also feature two primary research initiatives: (1) the AI Foundry for Rapid Materials Discovery; and (2) the AI Connected-Intelligence Center for secure connected robotics and autonomous systems. These initiatives will enable breakthroughs in computational materials science, autonomous robotics, and other cyber-physical systems critical to the Texas economy and to national security. The initiative will also drive workforce development through AI/ML courses and hands-on training, as well as foster interdisciplinary collaboration and increasing AI adoption in research. The project is crucial for maintaining Texas' competitiveness in the global AI race, driving economic growth and enhancing national security through technological advancements.

About Texas A&M Engineering Experiment Station:

For more than 100 years, TEES has improved lives through basic and applied engineering research and development for the State of Texas and the nation. As one of eight state agencies within The Texas A&M University System (“A&M System”), TEES was established in 1914 as part of the Texas landgrant university system (under state statute, chapter 88, subchapter E, Texas Education

Code). TEES performs cutting-edge research, advances workforce through education and technology transfer, and offers solutions to improve quality of life and fosters economic development.

TEES’ structure maximizes research and educational partnerships across the state and nation by forming networks that bring together subject matter expertise from universities, national laboratories, state and federal agencies, and industry. TEES serves as a catalyst for collaborations that position Texas to be especially competitive for federal projects and plays a major role in strengthening research leadership across the state. Headquartered in College Station, TEES has a close relationship with Texas

A&M University (“Texas A&M”), as well as regional divisions at 22 other institutions of higher education in Texas and affiliations with community colleges.

About Texas A&M

Established in 1876, Texas A&M has since grown to become a world leader in teaching, research, and service, with an international reputation for excellence in both academics and athletics. With an enrollment of nearly 80,000 students, Texas A&M is the fourth largest among U.S. universities. Located in the heart of Texas just hours from five of the nation's 20 largest cities

(Houston, San Antonio, Dallas, Fort Worth and Austin) and home to the George Bush Presidential Library and Museum, Texas

A&M is a popular and convenient destination for visitors from throughout the state and around the country. Additionally, the

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 6 of 31

"Aggie Spirit” and its many proud traditions, from the friendly "Howdy" that serves as the official campus greeting, to the majesty of the Fightin’ Texas Aggie Band and the solemnity of Silver Taps and Aggie Muster, make Texas A&M a unique academic and cultural experience. From world-class research to a broad range of academic and extracurricular programs, Texas A&M offers a myriad of opportunities for intellectual and personal development in an environment that fosters achievement at the highest level.

About The A&M System

The Texas A&M University System is one of the largest and most impactful higher education systems in the country, with an annual budget of $8.1 billion. Its statewide network includes 12 universities, a comprehensive health science center, eight state agencies, Texas A&M–Fort Worth and the Texas A&M–RELLIS. The Texas A&M System serves approximately 175,000 students and reaches millions more through service, research and outreach programs each year. With nearly $1.6 billion in annual research expenditures, the A&M System fuels innovation, supports communities and drives Texas’ economy forward.

More information about the A&M System and all its Members can be found in the A&M System About page

(https://www.tamus.edu/system/about/).

1.3 Scope of Work

The goal of this project is to acquire a world-class supercomputer designed specifically for the demands of modern AI workloads and large-scale scientific computing projects. The proposed architecture must be optimized for broad use cases, such as multi-node, high-precision simulation and computation, large-model training, and high-throughput inference capability.

The solution will be built on a scalable, high-speed, low latency fabric that eliminates network bottlenecks and allows for GPU-to-GPU communication efficiency. The system will have at least 500 next-generation GPUs, interconnected by dedicated east-west compute fabric. The system will also be designed for high-availability, manageability, and security, leveraging north-south

Data Processing Units (“DPU”) to isolate infrastructure and application planes. Storage will consist of a fast, scalable performance tier for direct use of the cluster and a multi-target capacity tier for local and cloud long-term and archival data storage.

Workloads that IGNITE will enable include, but are not limited to:

• Hypersonic & national security research

• Computational fluid dynamics and thermal simulation

• High-precision scientific computing

• Validating scientific AI models

• Rapid materials discovery

• Laboratory automation

• Digital twins

• Autonomous robot training models

• Large language model (“LLM”) development

• Scientific AI & ML models

• 3D models

Security-first architecture: To support a wide range of workloads with varied security requirements on shared infrastructure, the platform will be built on policy-driven multi-tenant architecture. This architecture eliminates the need for inefficient, physically isolated secure enclaves that often sit idle. Instead, it will use dynamic, attribute-based access controls (“ABAC”) and network micro-segmentation to securely separate workloads. Research teams will be able to work on a Controlled Unclassified

Information (“CUI”) level project requiring a secondary identity one day and a non-CUI project the next, or at the same time, with security policies applied in real-time based on project context, not physical hardware.

The selected Respondent will be responsible for complete, end-to-end delivery of the proposed solution. Responsibilities include, but are not limited to:

• System Design: Propose a fully integrated system architecture that meets all technical requirements specified in

Sections 2.

• Procurement and Logistics: Sourcing, staging, and managing the delivery of all hardware and software components.

Respondent will also manage all up-stream, data center, and supply-chain vendor relationships.

• Physical Installation: Full turn-key installation, including racking, cabling (power, data, and management), and power-on of all components in the target data center.

• System Configuration: Configuration of all hardware, firmware, networking fabrics (compute, storage, and management), operating systems, and the core software stack.

https://www.tamus.edu/system/about/

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 7 of 31

• Security: Enable multi-tenant capabilities to enable workloads governed by varied security framework(s).

• Acceptance Testing: Executing a mutually agreed-upon Acceptance Test Plan (ATP) to validate system functionality, peak performance (e.g., HPL, MLPerf benchmarks), and reliability.

• Project Management: Providing dedicated project manager(s) and detailed project plan with key milestones.

• Training and Documentation: Providing comprehensive system administrator and user training, as well as all "as-built" documentation.

• Support: Provide a multi-year, comprehensive warranty and support plan for all system components.

• Data center: Provide co-location options and articulate target data center requirements as specified in Section 2.

• Services: Propose options for workload management, cluster operations, and hardware support (see Section 2).

• Timeline: While multi-year options will be considered for services and data center hosting (see Section 2), cluster needs to be deployed and workload-ready within 12 – 17 months after contract award.

• Workload-ready: Upon cluster deployment, software and hardware will be stable, reliable, tested, and documented, security and multi-tenancy requirements will be enabled, and system will be ready for use by real end-users in a live environment.

1.4 Award

The intent of this RFP is to make a single award. With exception to section 2.1.5, Option 4, which may require a separate agreement with the Texas A&M System.

1.5 Contract Term

Should TEES, in its sole discretion, enter into an Agreement with the successful Respondent as a result of this RFP, the resultant

Agreement shall be effective upon execution for a suggested initial period of three (3) years. Agreement may be extended for additional terms providing all parties mutually agree in writing on the extension. Any extensions shall be at the same terms and conditions plus any approved changes to be determined by TEES and negotiated in writing with the successful Respondent.

1.6 Schedule of Events

The review and approval of RFPs is a multi-step process that requires variable amounts of time.

Responders are advised that the projected “Formal Execution of Agreement” date as listed in the “Tentative Timetable” may require extension.

Tentative Timetable

Release of RFP February 19, 2026 @ 2:00 pm CST

Deadline for Questions March 5, 2026 @ 5:00 pm CST

Posting Addendum and Response to Questions (if applicable) March 12, 2026 @ 2:00 pm CST

Deadline for Receipt of Proposals March 19, 2026 @ 2:00 pm CST

Evaluation of Proposals Completed March-April 2026

Formal Execution of Agreement (Projected) April 19, 2026

TEES will make every effort to adhere to the above schedule. The schedule, however, is subject to change. This may be in the event that further clarification of responses or terms of contract are in the best interest of TEES and/or in the event TEES requires more time to ensure that the selection of the Respondent is in accordance with its policies, rules and regulations, as well as actual timing needs.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 8 of 31

SECTION 2:

REQUIREMENTS

2.1 Technical Proposal

Provide a detailed plan on how you propose to provide all aspects of the Texas IGNITE cluster and services based on your understanding of the current environment and the scope of work described in Section 1.3.

Respondents must provide a comprehensive risk matrix identifying potential technical, operational, and supply-chain risks associated with proposed solution(s), especially for selected chipset architecture(s). For each identified risk, provide a mitigation strategy and a residual risk level (e.g., low, medium, high).

Response shall clearly detail the following system hardware, software, security, services, storage, and data center requirements:

2.1.1. Hardware

Recognizing the rapidly evolving landscape of hardware acceleration, Respondents are encouraged to propose multiple homogeneous chipset options determined to be the most effective for enabling both high and low precision workloads as well as multi-tenancy requirements for consideration. Proposals should clearly justify the selection of chosen architecture(s) by demonstrating superiority in handling varied workloads described in Section 1.3, including a balance of raw computational power, memory bandwidth, energy efficiency, floating point precision, and software ecosystem maturity.

Node Count: Respondents will recommend the most appropriate node count to meet the minimum GPU count of at least 500. All nodes in the cluster must be the same chipset type.

Cooling: Each compute node will include Direct-to-Chip (“DTC”) water cooling technologies with in-rack Coolant

Distribution Units (“CDU”).

GPU Configuration: Each node will contain eight (8) GPUs.

GPU Specifications: Each GPU must meet or exceed the following specifications:

• HBM3e (or technologically superior) high-bandwidth memory per GPU.

• Support for new-generation tensor formats, including FP4, FP6, and FP8 precision as well as FP32 and

FP64 support for traditional HPC and scientific computing workloads.

• Next-generation high-speed interconnect providing bidirectional GPU-to-GPU bandwidth within the node.

CPU Host: Each node will be equipped with dual, high-performance x86-64 core CPUs necessary to support anticipated workloads.

System Memory: Each node will be configured with high-speed DDR5 system memory. Respondents should include maximum node memory recommended per chipset.

Monitoring: Respondents will include real-time and environmental monitoring capabilities for temperature, humidity, and water leaks, with automated alerts.

2.1.2. Networking

The system will include high-speed compute East-West Fabric connecting all GPUs with the following requirements:

o Per-GPU Networking: The fabric will provide one dedicated network adapter for each individual GPU.

o Protocol: Proposals should specify per-port bandwidth and total fabric bisection bandwidth.

o Core Switches: The recommended fabric must be capable of supporting the full bandwidth of all 500+

GPU-attached adapters.

o Capabilities: The recommended fabric must support advanced networking features to accelerate collective communication operations.

The system will also include North-South fabric for infrastructure and storage connectivity with the following requirements:

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 9 of 31 o DPU Requirement: Each compute node will be equipped with two (2) dual-port Data Processing Units

(DPUs) o Fabric Isolation: These DPUs will create a logically (and/or physically) separate infrastructure fabric for storage, cluster management, and security, isolating this traffic from the primary compute fabric.

o Storage Fabric Switches: The storage fabric will be built on high-performance, low-latency switches designed for HPC and AI storage workloads.

o DPU Functionality: DPUs will provide accelerated networking for storage and support for secure, multi-tenancy security infrastructure management.

Management fabric (Out-of-Band or “OOB”):

o Topology: A dedicated, physically separate 100GbE Out-of-Band (OOB) management network will be provided.

o Switches: The OOB fabric will utilize resilient 100GbE switches for managing all compute nodes, switches, PDUs, and storage controllers.

o Management: Include required nodes for cluster management, running operations software, scheduling services, enhanced AI services, and other core software needed for the cluster to operate.

2.1.3. Storage

The system will include storage needed for cluster operations with the following requirements:

o Local Storage: Each node will include high-speed, local storage for operating systems (“OS”) and scratch space.

o Performance: The Respondent will propose a single, high-performance, parallel file system storage solution.

The solution must be capable of saturating ethernet storage fabric and providing low-latency access for all

GPUs.

o Direct-to-Accelerator Storage: The Respondent will propose a storage solution that supports direct data paths between storage and accelerator memory. This should enable high-throughput, low-latency data transfers directly between storage and GPU memory, bypassing the CPU.

o Certification: The proposed storage solution will be from storage partners certified and validated to work with recommended GPU chipset(s).

o Capacity & Bandwidth: The proposal will clearly state the aggregate read/write bandwidth (TB/s) and total usable capacity (PB) of the solution (e.g., minimum 8 PB usable, minimum 2 TB/s aggregate read).

o Connectivity: Vendor will provide data management software to securely and easily transfer data to and from various locations including, but not limited to, on-premises enterprise storage platforms and cloud file sharing platforms, like Microsoft 365.

o Cluster specific scratch and semi-persistent storage required for cluster operations will be physically located with the cluster.

o Respondent will include system design drawings with details about storage connectivity and protocols.

To facilitate anticipated growth in data storage and management needs beyond the cluster, vendor will also propose options for long-term, on-premises enterprise file storage solutions:

o Enterprise Storage: Propose reliable, available, and secure enterprise grade file storage solution that includes fast, always available storage tiers and cost-effective archive storage tiers for long-term data storage. This ecosystem is for data storage needs beyond the cluster.

o Include the ability to connect to cloud storage options (e.g., Microsoft 365 Commercial & GCC-High, Amazon S3, etc.) as destination targets for data entering and exiting the storage ecosystem.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 10 of 31

2.1.4. Software

Operating System: A supported Linux distribution (e.g., Ubuntu) must be installed and configured on all nodes.

Enterprise AI Software Stack & Support: Proposal must include a comprehensive, enterprise-grade AI software suite with 3-year, 24x7 technical support for all accelerators. This stack must provide a fully integrated and optimized environment for large-scale distributed workloads, including: optimized frameworks, performance libraries, inference engines, and containerization & orchestration.

Cluster Management: The solution will include a full suite of cluster management, monitoring, and provisioning tools. Vendor may include additional software to fine tune or provide extended functionality for the cluster (e.g., for security).

Scheduler: The solution will include 3-year, enterprise-support licenses for a GPU-aware and topology-aware workload scheduler for all GPUs.

Functionality: The scheduler will provide advanced GPU orchestration, including fractional GPU sharing, job queuing, prioritization, and detailed real-time and historical analytics on GPU utilization.

2.1.5. Data Center

Racks: Respondent will provide all necessary racks for the compute nodes, switches, management, and utility/storage components.

PDUs: All racks will be populated with redundant (A/B) intelligent Power Distribution Units (PDUs). Integrated racks with built-in power and cooling inter-connects are also acceptable.

Cabling: All necessary optical and copper cabling (e.g. ethernet, storage, management) must be provided, cut to length, and managed.

Respondent will propose four options for cluster hosting:

o Option 1: Provide detailed specifications (e.g., power, cooling, dimensions, number of racks, human capital, etc.) as required to successfully host IGNITE in a Texas A&M System owned data center.

o Option 2: Provide co-location options within the State of Texas that meet the power, cooling, security, and space requirements for the cluster for a minimum of two (2) years with the option to extend for one (1) year for a total of three (3) years. Then propose logistics, timeline, considerations, and cost to physically relocate the cluster to a Texas A&M System owned data center. Specify third-party vendor(s) who may be able to conduct the move.

o Option 3: Provide co-location options within the State of Texas that meet the power, cooling, security, and space requirements for the cluster for a minimum of three (3) years with the option to extend in one-year increments as needed for the life of the cluster.

o Option 4: Provide Public–Private Partnership (“P3”) co-location facility on A&M System property. Respondent will propose a P3 model in which a third-party developer finances, designs, constructs, owns, and operates a privately owned, multi-tenant data center facility located on A&M System property. The facility will function as a commercial co-location environment serving multiple users, with the A&M System retaining strategic tenancy rights, including for example, rights of first refusal (“ROFR”) on available capacity.

Connectivity: Co-location options must include the ability to connect cluster resources into the Lonestar Education and Research Network (“LEARN”). Co-location providers will work with the Trans-Texas Digital Network

(“TTDN”) to ensure connectivity back to the Texas A&M network and include options to enable express routes to major cloud providers (e.g., Microsoft, Amazon, and Google).

Capacity: For co-location options, include the ability to reserve up to 30% additional capacity for future expansion.

2.1.6. Services & Support

Warranty: Respondent will provide a minimum 3-year, 24/7/365, Next Business Day (“NBD”) on-site support plan for all hardware components. Articulate what spare parts may be on-site vs. those requiring shipment as well as where support teams will be located. Specify service level agreements (“SLA”) for hardware remediation activities for consideration and agreement.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 11 of 31

Software Support: Includes 3-year enterprise support for software components as specified above.

Installation & Deployment: Full, turn-key installation and configuration services as defined in sections 1.3 and 2.

Training: Respondent will include training credits to enable rapid adoption, knowledge transfer, and staff readiness.

Respondent will propose professional service tiers for Texas IGNITE, including, but not limited to the following:

o Workload management: Recommend the number of personnel and targeted expertise needed to facilitate end user engagement, consulting, financial planning, reporting, and use case and workload management.

This team will be internal. Propose vendor options to assist the internal team with day one workload management for up to one (1) year.

o Cluster operations: Propose vendor options to fully manage the cluster. Include server, storage, security, and operations management for a minimum of one (1) year with the option to either transition to an internal team or if needed, continue vendor cluster management services in one-year increments.

o Hardware Support: Propose vendor and co-location provider options to fully manage cluster hardware.

Include servers, cooling, switching, etc. This service may or may not be leveraged depending on the target data center chosen in the final award.

2.1.7. Security

Multi-tenancy: This cluster must support multiple customers (e.g., tenants) who share resources but have isolated, secure, and private access to their own data, configurations, and user management.

Logical partitioning and performance isolation: The proposed solution must support robust, verifiable multi-tenancy that ensures strict logical compartmentalization of computer, memory, and network resources. This must include the ability to create “Secure Enclaves” where specific projects are isolated to prevent tenant data leakage and interference.

Fabric-level segmentation: The interconnect fabric must support native isolation to ensure that inter-node traffic for one tenant is invisible to others.

Frameworks: Enforce an identity-centric, Attribute-Based Access Control (ABAC) policy framework, ensuring that

Controlled Unclassified Information (“CUI”) and non-CUI workloads can securely coexist on shared infrastructure.

These requirements are guided by key frameworks such as NIST SP 800-171 and NIST SP 800-53 respectively.

Racks: All individual racks must be lockable.

Encryption: All data for cluster specific storage must be encrypted using FIPS 140-2 validated cryptographic modules.

Certification: The solution must provide a path for meeting Cybersecurity Maturity Model Certification (“CMMC”)

Level 2 requirements for protecting the confidentiality of research data within a multi-tenant environment.

Respondent will be prepared to support and assist the agency certification efforts.

Recommended co-location providers will have Service Level Agreements (SLA) that explicitly state they meet NIST

SP 800-171 physical standards including, but not limited to, biometric and multi-factor entry, visitor management, output device privacy, escort policy, and video surveillance.

2.1.8. Governance & Data Management

Research Governance: Respondent will propose a recommended governance strategy to manage strategy, risk and provide oversight of the Texas IGNITE cluster.

Data Management: Vendor will propose a digital data lifecycle management framework for the IGNITE platform, enabling automated tiering, secure archiving, and policy-driven retention and deletion.

2.1.9. Incubation Lab

Respondents must include the design and implementation of a small-scale incubation lab that mirrors the production

GPU cluster environment. This lab should replicate key hardware and software elements of the full GPU cluster to support training, experimentation, and onboarding for staff, faculty, and students, ensuring they gain hands-on

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 12 of 31 experience before transitioning to the workload-ready cluster.

2.1.10. Vendor Qualifications

Partner Status: Vendor must have demonstrated, verifiable experience in deploying multi-node, large-scale GPU supercomputers (500+ GPUs) within the last 24 months.

Storage Certification: Vendor must demonstrate partnership and certification status with the proposed storage provider.

Summarize why the Respondent is the best firm to undertake the scope of work including relevant business experience.

Provide resume of key individual(s) that would be associated or committed to the project.

2.1.11. General Requirements

Provide a detailed overview of Respondent system integration process for large scale GPU and AI-enabled clusters.

Provide an overview of Respondent complies with NIST 800.53 and NIST 800.171 frameworks.

Provide a detailed overview of what system monitoring and reporting features are offered. Submit samples of key reports in response.

Provide details on any other functionality or requirements that would be available to TEES.

Provide a detailed timeline with key milestones and target completion date.

2.2 References

Respondents shall provide a list of references, minimum three (3), where services comparable in size and scope have been performed in the last two (2) years. References shall include Company Name, Contact Person, Company Address, Telephone & Fax Number and

Email Address. TEES reserves the right to contact these references to verify the vendor’s ability to perform these services. A negative reference may be grounds for the disqualification of your RFP response

2.3 Implementation

See section 2.1.

2.4 Acceptance of Work

Final acceptance of the Texas IGNITE cluster shall be contingent upon successful completion of a formal acceptance testing period, lasting no fewer than 30 consecutive days following the vendor’s notice of installation complete. During this period, TEES will evaluate the system’s performance against the technical specifications and benchmarks outlined in this RFP, including a 72-hour continuous “stress test” to ensure hardware stability and thermal reliability under peak load. The cluster will be deemed accepted after it demonstrates adherence to the technical specifications in Section 2, passes all agreed-upon performance metrics, and provides a fully functional, CMMC-ready multi-tenant environment. Any deficiencies identified during the acceptance testing period must be remediated by the Respondent at no additional cost to TEES, at which point the 30-day acceptance clock will be reset.

After formal acceptance of installation complete, day one services agreed to and opted into by TEES shall begin.

2.5 Progress Reports

Respondents will be required to submit monthly written status reports detailing progress against the project baseline, including completed milestones, upcoming tasks for the next reporting period, and an updated log of technical issues or blockers. Additionally, the Respondent must maintain a dynamic risk and mitigation register, to be reviewed during monthly briefings, which provide real-time visibility into supply chain lead times, installation dependencies, co-location readiness, and hardware or software integration stability. These reports may be shared with internal stakeholders, System Internal Audit, and external auditors.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 13 of 31

SECTION 3:

GENERAL INFORMATION

3.1 Submittal Deadline and Location

3.1.1 All responses must be received by TEES no later than Thursday, March 19, 2026 @ 2:00 pm CST

It is each Respondent’s responsibility to ensure that the entire submittal response is received by TEES by the deadline stated above. Late responses will not be considered under any circumstances. TEES shall not be responsible for failure of electronic equipment or operator error.

3.2 Submittal Instructions

3.2.1 Proposals, including Execution of Offer (Section 6), Non-Collusion Affidavit (Section 10), and the VetHUB

Subcontracting Plan (HSP – Section 12), must be signed by Respondent’s company official authorized to commit such proposals. Failure to sign the Execution of Offer may be basis for proposal disqualification.

3.2.2 Complete proposals must be submitted thorough the Aggiebid portal. It is the responsibility of interested vendors to regularly check the Aggiebid site for any possible addenda to this project. The RFP is inclusive of all addenda issued.

NOTE: Respondents must be a registered vendor. If you need assistance with vendor registration, please reach out to vendorhelp@tamu.edu or Jaggaer’s Supplier Support Team at 1-800-233-1121.

All electronic copies must be in Adobe Portable Document Format (PDF). All image files must be in one of the following formats: .jpg, .gif, .bmp, or .tif. We prefer image files to already be inserted as part of a document such as a

PDF.

Submittal File Format: Submission shall be saved as two (2) separate files in Adobe Portable Document Format (PDF) according to the Proposal Components listed in Section 3.7 below and titled as “Proposal” and “Forms.”

3.3 TEES Contacts

TEES requires that respondents submit any questions related to the scope, requirements, or submissions of the RFP response through the Aggiebid portal specified in Section 3.2.2. TEES specifically requests that Respondents restrict all contact and questions regarding this RFP through the Aggiebid portal. Questions submitted via email or phone call will not be accepted.

Exception to the above instructions:

For any issues accessing the portal, please contact the Buyer listed below:

Danielle Toler, CTCD

Buyer III danielle.toler@tamu.edu

Deadline for questions is Thursday, March 5, 2026 @ 5:00 pm CST.

3.4 Inquiries and Interpretations

Responses to inquiries which directly affect an interpretation or change to this RFP will be issued in writing by addendum

(amendment) and posted to the Aggiebid portal. All such addenda issued by TEES prior to the time that proposals are received shall be considered part of the RFP, and the Respondent shall consider and acknowledge receipt of such in their proposal.

Only those TEES replied to inquiries which are made by formal written addenda shall be binding. Oral and other interpretations or clarification will be without legal effect.

3.5 Open Records

TEES considers all information, documentation and other materials requested to be submitted in response to this solicitation to be of a non-confidential and/or non-proprietary nature and therefore shall be subject to public disclosure under the Texas Public

Information Act (Texas Government code, Chapter 552) after an agreement is awarded.

Respondents are hereby notified that TEES strictly adheres to all Statutes, court decisions and the opinions of the Texas Attorney

General regarding the disclosure of RFP information.

mailto:vendorhelp@tamu.edu mailto:danielle.toler@tamu.edu

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 14 of 31

3.6 Terms and Conditions

The Terms and Conditions (ref. Section 4) shall govern any Agreement issued as a result of this solicitation RFP.

Additional or attached terms and conditions which are determined to be unacceptable to TEES may result in the disqualification of your proposal. Examples include, but are not limited to, liability for payment of taxes, subjugation to the laws of another State, and limitations on remedies.

3.7 Proposal Components

The following documents are to be returned as part of your proposal submittal:

• Proposal:

✓ Cover letter to include a brief introduction with interest and capability of Respondent.

✓ Technical Proposal (Section 2 – Technical Proposal must provide a detailed response to all requirements outlined in

Sections 2.1 through 2.1.11)

✓ References (Section 2.2)

✓ Respondent’s Questionnaire (Section 7)

✓ Pricing Proposal (Section 8)

• Forms:

✓ Signed Execution of Offer (Section 6)

✓ Non-Collusion Affidavit (Section 10)

✓ Certificate of Insurance (Section 11)

✓ VetHUB Subcontracting Plan (Section 12)

✓ Any Addendums, if applicable

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 15 of 31

SECTION 4:

GENERAL TERMS AND CONDITIONS

4.1 General

These General Terms and Conditions shall be made a part of and govern any Agreement/Purchase Orders resulting from this Request for Proposal.

Each response should be prepared simply and economically, providing a straightforward and concise description of Respondent’s ability to meet the requirements of this RFP. Emphasis should be on completeness, clarity of content and responsiveness to the offer requirements.

In accordance with Texas Education Code 51.9335, TEES shall make the award based on, but not limited to, the following best value criteria: The purchase price; The reputation of the vendor and of the vendor's goods or services;

The quality of the vendor's goods or services; The extent to which the goods or services meet the institution's needs;

The vendor's past relationship with the institution; The impact on the ability of the institution to comply with laws and rules relating to historically underutilized businesses and to the procurement of goods and services from persons with disabilities; The total long-term cost to the institution of acquiring the vendor's goods or services; Any other relevant factor that a private business entity would consider in selecting a vendor; and The use of material in construction or repair to real property that is not proprietary to a single vendor unless the institution provides written justification in the request for bids for use of the unique material specified. Additionally, all respondents are hereby notified that TEES shall consider all factors it believes to be relevant in the determination of the “Best Value” including, but not limited to: past experience, references, proposal, and price. TEES’s decision is final.

Responses are to be valid for a minimum of 180 days from the submittal deadline date to allow time for evaluation, selection, and any unforeseen delays.

4.2 Final Review and Approval

Failure to comply with the requirements contained in this Request for Proposal may result in the rejection of the proposal.

The vendor agrees to protect the State from claims involving infringement of patents or copyrights.

The vendor hereby assigns to purchaser, any and all claims for overcharges associated with any contract resulting from this RFP which arise under the antitrust laws of the United States 15 U.S.C.A. Section 1, et seq. (1973) and which arise under the antitrust laws of the

State of Texas, Texas Business and Commercial Code Ann. Sec. 15.01, et seq. (1967).

Questions should be directed to the TEES official identified in Section 3.3 of this Request for Proposal.

Proposals and any other information submitted by Respondent in response to this Request for Proposal shall become the property of

TEES.

TEES will not provide compensation to Respondents for any expenses incurred by the Respondent(s) for proposal preparation, product evaluations or demonstrations that may be made, unless otherwise expressly indicated.

Proposals which are qualified with conditional clauses, alterations, items not called for in the RFP documents, or irregularities of any kind are subject to disqualification by TEES at its option.

4.3 Definitions

Whenever the following terms are used in these General Terms and Conditions or in other documents the intent and meaning shall be interpreted as follows:

RFP shall mean Request for Proposal.

Proposal shall mean Respondents offer

TEES shall mean Texas A&M Engineering Experiment Station (TEES) and other system parts.

Respondent shall mean the individual, partnership, corporation, or other entity responding to this RFP.

Vendor shall mean the individual, partnership, corporation, or other entity awarded an agreement for labor or for equipment & supplies under this RFP in accordance with the terms, conditions, and requirements herein.

Agreement shall mean an agreement, documented by written instrument, between TEES and the successful respondent to provide products/services as requested to TEES in College Station, Texas.

4.4 Time of Performance

Time is of the essence in the rendering of services. Seller agrees to perform all obligations and render services set forth per this proposal.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 16 of 31

4.5 Default

In the event of substantial failure by Vendor to perform in accordance with the terms of this Agreement, TEES may terminate this

Agreement upon fifteen (15) days written notice of termination setting forth the nature of the failure (the termination shall not be effective if the failure is fully cured prior to the end of the fifteen-day period), provided that said failure is through no fault of TEES.

The cancellation of the Agreement, under any circumstances whatsoever, shall not affect or relieve Vendor from any obligation or liability that may have been incurred or will be incurred pursuant to the agreement and such cancellation by TEES shall not limit any other right or remedy available to TEES at law or in equity.

4.6 Termination

4.6.1. For Convenience:

TEES reserves the right to terminate this Agreement at any time, in whole or in part and without cost or penalty, by providing thirty (30) calendar days’ advance written notice to Vendor, if TEES determines that such termination is in the best interest of the Agency. In the event of such termination, the Vendor shall, unless otherwise mutually agreed upon in writing, cease all work immediately upon the effective date of termination. TEES shall be liable only for payments limited to the portion of work

TEES authorized in writing and which Vendor has completed, delivered to TEES, and which has been accepted by TEES. All such work shall have been completed, in accordance with contract requirements, prior to the effective date of termination.

TEES shall have no other liability, including no liability for any costs associated with the termination.

4.6.2. In no event shall such termination by TEES as provided for under this Section give rise to any liability on the part of TEES including, but not limited to, claims of Vendor for compensation for anticipated profits, unabsorbed overhead, or interest on borrowing. TEES’s sole obligation hereunder is to pay Vendor for products and/or services ordered and received prior to the date of termination.

4.6.3. TEES reserves the right to immediate cancellation, without penalty, due to non-performance. In the event of a breach a material term of this Agreement by a Party, the non-defaulting Party may terminate this Agreement upon thirty (30) days’ prior written notice to the other Party detailing the nature of the breach and the other Party fails to fully cure the breach within such 30-day period. In the event that Agency terminates this Agreement pursuant to this Section 4, Agency shall receive a pro-rata refund of any pre-paid.

4.7 Agreement Amendments

No modification or amendment to the agreement shall become valid unless in writing and signed by both parties. All correspondence regarding modifications or amendments to the agreement must be forwarded to the TEES Purchasing Department for prior review and approval. Only the contract administrator within Strategic Sourcing & Purchasing Services or his/her designee will be authorized to sign changes or amendments.

4.8 Independent Contractor

Notwithstanding any provision of this Agreement to the contrary, the Parties hereto are independent contractors. Vendor agrees that no employer-employee, partnership, agency, or joint venture relationship is created by this Agreement or by Vendor’s service to TEES.

Except as specifically required under the terms of this Agreement, Vendor (and its representatives, agents, employees and subcontractors) will not represent themselves to be an agent or representative of TEES or the A&M SYSTEM. As an independent contractor, Vendor is solely responsible for all taxes, withholdings, and other statutory or contractual obligations of any sort, including but not limited to workers’ compensation insurance. Vendor and its employees shall observe and abide by all applicable TEES policies, regulations, rules and procedures, including those applicable to conduct on its premises.

4.9 Right to Audit

At any time during the term of this agreement and for a period of four (4) years thereafter TEES or duly authorized audit representative of TEES, or the Texas A&M University System, at its expense and at reasonable times, reserves the right to incrementally audit Vendor’s records and manufacturer’s pricing relevant to all pricing provided under this agreement. In the event such an audit by TEES reveals any errors/overpayments by TEES, Vendor’s shall refund TEES the full amount of such overpayments within thirty (30) days of such audit findings, or TEES at its option, reserves the right to deduct such amounts owing TEES from any payments due Vendor.

4.10 Sales and Use Tax

TEES, as an agency of the State of Texas, qualifies for exemption from State and Local Sales and Use Taxes pursuant to the provisions of the Texas Limited Sales, Excise, and Use Tax Act. The Seller may claim exemption from payment of applicable State taxes by complying with such procedures as may be prescribed by the State Comptroller of Public Accounts.

4.11 Observance of TEES Rules and Regulations

Vendor agrees that at all times its employees will observe and comply with all regulations of TEES, including but not limited to parking and security regulations.

4.12 Non-Disclosure

Vendor and TEES acknowledge that they or their employees may, in the performance of the resultant agreement come into the possession of proprietary or confidential information owned by or in the possession of the other. Neither party shall use any such information for its own benefit or make such information available to any person, firm, corporation, or other organizations, whether or not directly or indirectly affiliated with Seller or TEES unless required by law.

TEES RFP #28-EQUI-RFP-1660; GPU High Performance Computing Cluster Page 17 of 31

4.13 Publicity

Vendor agrees that it shall not publicize this agreement or disclose, confirm or deny any details thereof to third parties or use any photographs or video recordings of TEES’s name in connection with any sales promotion or publicity event without the prior express written approval of TEES.

4.14 Severability

In case any one or more of the provisions contained in this Agreement shall, for any reason, be held to be invalid, illegal, or unenforceable in any respect, such invalidity, illegality, or unenforceability shall not affect any other provisions hereof, and this Agreement shall be construed as if such invalid, illegal, and unenforceable provision had never been contained herein.