Mapping Machine Learning to Physics (ML2P)

Key Dates
Due Date

-

Posted Date

-

Agency & Value
Agency

DEF ADVANCED RESEARCH PROJECTS AGCY

Contract Value

Not provided

Codes & Classification
Solicitation Number

DARPA-PS-25-32

NAICS

541715 - Research and Development in the Physical, Engineering, and Life Sciences (except Nanotechnology and Biotechnology)

Classification Code

AC12

Additional Details
Set Aside

None

Place of Performance

-

SourceSAM
The Defense Advanced Research Projects Agency (DARPA) has issued a solicitation titled "Mapping Machine Learning to Physics (ML2P)" under procurement identifier DARPA-PS-25-32. This initiative seeks to address a critical challenge in the deployment of machine learning (ML) technologies within Department of Defense (DoD) operations, particularly in resource-constrained battlefield environments where power availability is limited. The ML2P program is designed to fundamentally reimagine how power efficiency is integrated into the development and deployment of ML systems. Rather than treating power consumption as an afterthought, ML2P aims to embed power efficiency considerations from the outset of ML design. The program will achieve this by mapping ML efficiency directly to physical principles, utilizing precise Joule measurements to enable accurate predictions of power usage and performance across a variety of hardware architectures. A central objective of ML2P is the development of multi-objective optimization functions that balance power consumption with key performance metrics. The program will also explore the interactions of local optimizations through the concept of Energy Semantics of ML (ES-ML), with the goal of solving the broader energy-aware ML optimization problem. The intended outcome is to provide the DoD with ML solutions that are not only high-performing but also optimized for energy efficiency, thereby enhancing operational capabilities in environments where power is a limiting factor. This solicitation falls under NAICS code 541715, which covers Research and Development in the Physical, Engineering, and Life Sciences (except Nanotechnology and Biotechnology), as per the 2022 NAICS version. The associated Product Service Code is AC12. The contracting entity is the Defense Advanced Research Projects Agency (DARPA), a federal office dedicated to advancing cutting-edge research for national defense. Interested parties should note that the solicitation is currently active, with a posting date of June 10, 2025, and a response deadline of December 17, 2025. The primary point of contact for this opportunity is the Solicitation Coordinator, who can be reached at [email protected]. Additional details, including attachments relevant to the solicitation, are available via the official notice on SAM.gov at https://sam.gov/opp/c4624b87b5e043bfb0949975fbb274dd/view. Contractors considering this opportunity should possess expertise in machine learning, power efficiency optimization, and hardware architecture analysis, as well as a strong background in research and development within the physical and engineering sciences. The program’s focus on energy-aware ML optimization and multi-objective function development will require advanced capabilities in both theoretical and applied research. No specific set-aside requirements or special considerations are indicated in the notice. Overall, ML2P represents a significant opportunity for organizations capable of innovating at the intersection of machine learning and energy efficiency, with the potential to deliver transformative solutions for defense applications in power-constrained environments.
Documents8