Attachment A - Statement of Work (SOW).pdf

Statement of Work (SOW) Title: Computer for Modeling, Simulation, and Artificial Intelligence in Additive

Manufacturing

I. Background Information

The National Institute of Standards and Technology (NIST) Engineering Laboratory’s Measurement Science for Additive Manufacturing (MSAM) program consists of teams from the Intelligent Systems Division (ISD) and Systems Integration Division (SID). Where ISD has long focused on the machines and processes, SID has focused on the design-to-product transformation. As AM is a digital manufacturing process, advances systems modeling, systems integration, and data science can transform the process, and thus the fabricated part. The role of modeling and simulation is increasingly playing a role in the fabrication of all AM parts, and the amount of data collected, processed, and analyzed has quickly moved from gigabytes to terabytes. This evolution is spurring activity in real time control, and together these activities are driving computation costs. The Advanced Informatics and Artificial Intelligence project in MSAM aims to reduce the burden and upfront costs of industry to realize the benefits of advanced data science techniques, including digital twins, machine learning, and artificial intelligence. The procurement of a new computer is necessary to test and validate new models, methods, algorithms and AI agents so that the results can be benchmarked, scaled, and disseminated to SMMs (small and medium manufacturers) and LSIs (Large Systems Integrators) alike.

Additive manufacturing refers to a class of emerging manufacturing technologies for producing highly-complex, customized components by joining materials to make objects based on three-dimensional (3D) model data, typically built layer upon layer. Parts are fabricated directly from a 3D computer-aided design (CAD) file that is virtually sliced into many thin layers and sent to an additive manufacturing (AM) system where each layer is incrementally formed. Several AM processes and systems have been developed over the past two decades and their capabilities have grown significantly. Generally, metal-based additive processes form parts by melting or sintering material in powder form until all layers are completed.

Additive manufacturing is capable of manufacturing parts that are not possible to make with conventional machining. The parameter variations that can be used to fabricate a part greatly exceed those of conventional manufacturing. The in-process monitoring capabilities and ex-situ part measurements create large data sets with underlying complex relationships that require substantial data processing to reveal. Modeling and simulation in AM have become important tools for understanding and facilitating the fabrication of an AM part. The emergence of generative AI, agentic AI, and foundation models is changing how practitioners interact with the data and value and utility of the correlations unearthed from the data. As modeling, simulation, and AI capabilities continue to advance, new hardware support is necessary to conduct meaningful, impactful research in these areas.

II. Purpose

The purpose of this requirement is to procure a computer system to support the extensive data science needs of additive manufacturing and the demanding computational needs of follow-on artificial intelligence applications.

Scope/Objective: The computer will be used to develop, test, and deploy 1) advanced analytics, including AM digital twins, microstructure modeling, and process simulation and 2) artificial intelligence, including agentic AI for AM, multi-modal foundation model development, and AM tensor frameworks. The GPUs capabilities will allow for tailoring approaches to align with the latest in AI application deployment.

III. Minimum Requirements

The Contractor shall provide a system that meets all technical specifications identified below.

Hardware and Processing

• The hardware shall support GPU processing capabilities in line with current state-of-the-art AI deployment and include (1) NVIDIA H200 NVL Tensor Core GPU 141GB HBM3e PCIe Gen 5.0 with capability and support for adding at least one more in the future. An NVIDIA B300 GPU is an acceptable alternative.

• The motherboard provided shall be capable of supporting a minimum of two (2) of the provide GPU

• The power supply shall support 120V and 240V configurations capable of running a minimum of two (2) of the provided GPU.

• The hardware shall support computing loads equivalent or to exceed the capabilities of 64-Core 3.20 GHz AMD Threadripper Pro 7985WX

• The system shall support at least (6) display monitors using GPU capacity equivalent to an NVIDIA RTX 5090 32GB or greater

• The system shall have a minimum of 128 GB of RAM

• The operating system shall be Windows 11 pro or later

• The primary storage hardware of the computer shall be a 7.68 TB PCI-E 4.0 NVMe SSD or greater

• The system shall come in a stand-alone tower configuration and not require additional hardware/ rack to install

• The system shall support 100 GbE networking

IV. Deliverables

Deliverable Quantity or Format Due Date Computer System 1 Within 30 Days of

Award Hardware Validation Report 1 Prior to Delivery Product Training and Material Up to 5 people Within 30 Days After

Installation

V. Place of Performance

The equipment shall be located and operated in in Building 304, Additive Manufacturing Research Center.

VI. Period of Performance/Lead Time

Delivery shall be made within thirty (30) days or less after receipt of order (ARO).

VII. Hardware Validation and Burn-in Testing

The contractor shall perform a minimum 48-hour continuous stress test (Burn-in) of the fully assembled system prior to delivery. This testing must verify system stability, thermal management, and power delivery under maximum load.

VIII. Delivery Terms

Delivery shall be F.O.B Destination and shall occur in accordance with the delivery due dates provided in the above table.

FOB Destination means: The contractor shall pack and mark the shipment in conformance with carrier requirements, deliver the shipment in good order and condition to the point of delivery specified in the purchase order, be responsible for any loss of and/or damage to the goods occurring before receipt and acceptance of the shipment by the consignee at the delivery point specified in the purchase order; and pay all charges to the specified point of delivery. The contractor shall deliver all Line Items to:

National Institute of Standards and Technology Shipping and Receiving 100 Bureau Drive, Building 301 Gaithersburg, MD 20899 POC: Paul Witherell (Building 220 A367 x3385)

***Due to ongoing limited site access of the NIST facilities, the Awardee SHALL confirm with the Designated Government Official and the Contracting Officer prior to ANY shipment to NIST***

IX. Training

The contractor shall provide a minimum of 2 hours of training for up to five (5) people. The training shall provide a thorough demonstration of all hardware functions. The training shall include hands-on exercises as part of the instruction. The training shall include electronic copies of training materials and related information, including training handouts and operator manuals.

The training shall be completed virtually and shall be completed no later than 30 calendar days after notification of installation.

X. Inspection & Acceptance

In addition to the inspection and acceptance terms articulated in 52.212-4, the Government reserves the right to perform such performance tests and evaluations as defined below to verify specified system performance. Such tests and evaluations, if performed, shall be conducted within the environment that the system is to be operated. The Contractor has the right to be present during the tests and evaluations, if performed, at the Contractor’s expense.

Performance Tests:

The performance test will consist of testing the ability of the system to run at least (3) different large language model configurations on the dedicated GPU. NIST reserves the right to independently test the system before acceptance for a period not exceeding 14 calendar days.

XI. Warranty

The contractor shall warranty the entire system for a period of a minimum of 3 years labor and 1 year part replacement to resolve any hardware or software problems discovered after receipt of the equipment and shall be in accordance with terms in FAR 52.212-4. Warranty shall commence upon acceptance of the system by the Government and at a minimum shall include the following:

1) Full one year part warranty of all parts, including GPU, Processor, RAM, power supply, etc.

2) Full one year software warranty to the extent that delivered functionality can be recovered in case of failure

3) Full three-year labor warranty to address any hardware or software deficiencies.

XII. Payment Schedule

Advance payment is not authorized. The Contractor shall be paid, in accordance with Net 30-day payment terms, upon receipt and acceptance of a proper invoice, in accordance with the following schedule:

1. 100% after acceptance by the TPOC of fully installed system,

2. After the successful completion of the testing requirements set forth in this document under section set forth in this document,

3. After successful demonstration by the instrumentation that it performs IAW the technical requirements set forth in this document, and

4. After receiving an invoice submitted properly, IAW the purchase order terms and conditions.

NOTE: Partial shipments and partial invoices will not be accepted, unless otherwise requested and accepted by the Contracting Officer prior to award offer.

***Due to ongoing limited site access of the NIST facilities, the Awardee SHALL confirm with the Designated Government Official and the Contracting Officer prior to ANY shipment to NIST***