NSF AI institute for edge computing and systems (ATHENA)

The National Science Foundation Division of Computer and Network Systems awarded Duke University a $3 million cooperative agreement on November 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to support the NSF AI Institute for Edge Computing and Systems (ATHENA).

ATHENA conducts research toward deploying advanced artificial intelligence on edge devices—everyday-use devices and small systems—while ensuring availability, efficiency, functionality, and trustworthiness. The institute builds an AI-native full-stack design integrating AI functions across front-end and back-end systems that interact with classical and quantum computing in the physical world. The research transitions AI applications from assistive and analytical uses toward coordinated autonomous systems and integrated intelligent systems embedded in and directly interacting with the physical environment.

The institute's education and workforce development program engages learners across K-12, undergraduate, graduate, and industry levels through cross-campus curricula, industry internships, graduate bootcamps, seminars, and community-partnered outreach. ATHENA disseminates research through open-source release of code, designs, benchmarks, and testbed; position papers and workshops; and industry collaboration via a consortium supporting licensing, joint development, and entrepreneurship.

Performance occurs at Duke University's Office of Research Administration in Durham, North Carolina, with a period of performance from November 1, 2026, through October 31, 2031. The obligated amount is $3 million.

Abstract

The NSF artificial intelligence (AI) institute for edge computing and systems (ATHENA) conducts research toward ?big AI for small devices?. More specifically, the institute will create and deploy advanced AI on devices used in everyday applications while ensuring availability, efficiency, functionality, and trustworthiness of those systems. ATHENA?S research extends the rapid advances in AI, transitioning from assistive and analytical use toward coordinated autonomous systems and edge applications improving from standalone smart devices into integrated intelligent systems embedded in, and interacting directly with, the physical world. ATHENA?S mission is to build an AI-native full stack design (i.e., one in which AI functions are central to how a website or system appears to users on the front-end all the way to behind-the-scenes and into the back-end) that tightly interacts with computation ? Both classical and quantum computing ? In the physical world. The broader impact of ATHENA is in shaping the nation?S technology roadmap for future edge AI, a rapidly growing sector of the AI industry, by advancing critical technologies, promoting open communities, and partnering closely with industry. ATHENA?S education and workforce development program will continue to engage learners at all levels through cross-campus curricula, industry internships, graduate bootcamps, seminars, and K-12 outreach CO-designed with community partners. ATHENA continues its impact as a nexus point for collaborative efforts and leadership in the sector through open-source release of code, designs, benchmarks, and testbed; dissemination of results through position papers and workshops; and deepening of collaboration with industry and investors through a consortium that supports licensing, joint development, and entrepreneurship. By strengthening cross-institutional collaboration, ATHENA sustains an open research and education community that advances U.S. Leadership in edge AI and informs national policy. ATHENA seeks to realize the vision of ?immersive AI in the physical world? By developing next-generation autonomous AI technologies and advancing long-lived, self-adaptive, and self-evolving edge AI that enables unprecedented levels of interaction and coordination across edge systems, human users, and the physical world. The four research thrusts of the institute include: (I) computing systems to make long-lived edge AI programs first-class runtime objects, stateful, interactive, deadline-constrained, and accelerator-aware; (II) foundational AI to build the theoretical and algorithmic substrate for next-generation edge AI that is multimodal, agentic, trustworthy, and quantum-aware; (III) physical AI to develop AI deeply woven into the physical world, perceiving heterogeneous environments, reasoning under strict latency and resource budgets, and acting autonomously without cloud connectivity; and (IV) applications?use cases enabled by these systems, with emphasis on embodied AI and edge AI-assisted quantum computing, that both validate and drive the other thrusts. Institute research traverses the full cycle from theoretical design and modeling to implementation and prototyping, with quantitative evaluation on shared testbeds and benchmarks. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are planned for this award.