Three wins out of 5,000 applications
New York University announced August 3 that its faculty were selected for three projects in the Department of Energy's Genesis Mission, a new federal initiative aimed at accelerating scientific discovery through AI. The inaugural round drew more than 5,000 applications nationwide and funded 278 projects, putting NYU's three selections in a competitive pool with an acceptance rate under six percent, a ratio that signals just how much appetite there currently is across US research universities for federal AI research dollars.
The announcement was framed by NYU leadership as validation of a cross-disciplinary AI research push. Juan J. de Pablo, the university's executive vice president for global science and technology and executive dean of the Tandon School of Engineering, said: 'Science is entering an era where discovery itself can be accelerated by the tools we build to pursue it.' Gerard Ben Arous, dean of the Courant Institute, added that 'leveraging AI from the ground up is vital for both scientific discovery and technological advancement.'
Project one: controlling quantum hardware directly with AI
The first project, led by physics assistant professor Andrei Vrajitoarea and chemistry and physics assistant professor Norah Hoffman, focuses on AI-driven pulse-level optimal control of many-body quantum eigensolvers. In plain terms, the team is using AI to control quantum computing hardware at the lowest possible level, bypassing intermediate translation layers that currently slow computation and introduce errors when running quantum algorithms, an approach that could meaningfully shorten the path from theoretical quantum algorithms to hardware that runs them reliably.
That is a narrow, deeply technical target, but it sits at the center of one of the most consequential infrastructure bets in computing right now. Any institution running or partnering on quantum research programs should note that DOE is explicitly funding AI-native approaches to quantum control rather than treating AI and quantum as separate research tracks, a combination that is likely to define the next wave of national lab and university partnerships in the field.
Project two and three: multi-agent science, built with industry partners
The second project, MARS, standing for Multi-Agent Reinforcement Learning for Scientific Hypothesis Generation, is led by Brookhaven National Laboratory with NYU's Mengye Ren of the Courant Institute as the university lead, alongside Carnegie Mellon University and Microsoft as partners. MARS is building teams of specialized AI agents designed to generate evidence-based scientific hypotheses, with initial applications targeting analog circuit design and interpreting X-ray material structure data, two problem domains chosen because they generate large volumes of data that historically require significant human expert time to interpret.
The third project, DAISY, or Decentralized Agentic Intelligence System for Scientific Inquiry, is led by Iowa State University with NYU's Chinmay Hegde and Juliana Freire of the Tandon School as university leads, partnered with Argonne National Laboratory. DAISY is building decentralized AI agents that divide labor and share knowledge tokens across a research problem, with early target applications in materials science, crop breeding, and power grid research. Both projects reflect a consistent structural choice DOE appears to be enforcing across the entire program: a national lab as lead institution, university faculty embedded as named contributors, and at least one named industry or additional university partner rounding out the team.
What the funding model itself signals
The structure of these awards is arguably more newsworthy than any individual project. Genesis Mission's inaugural round did not fund single-PI university grants in the traditional NSF or NIH mold. Every NYU project sits inside a team that spans a national laboratory, at least one additional university, and in MARS's case a named commercial partner in Microsoft. That is a template federal AI research funding appears to be converging on across multiple agencies this year, and NYU's selections happen to illustrate it cleanly across all three of its awarded projects.
For university research offices and technology transfer teams, that has direct operational implications. Competing for this class of funding now requires having multi-institution, multi-sector partnerships assembled before a proposal is written, not after an award is announced. Universities without existing relationships with national labs or major AI industry players are starting from a structural disadvantage in this funding environment, regardless of the strength of their individual faculty's research, which raises the stakes for building those partnerships proactively rather than waiting for the next funding cycle to force the issue.
The transparency gap worth tracking
Notably absent from NYU's announcement, and from Genesis Mission's public disclosures more broadly at this stage, are funding amounts and project timelines. For a program that selected 278 projects from over 5,000 applications, the lack of disclosed award sizes makes it difficult for other institutions to benchmark their own funding strategy or estimate the total federal capital being deployed through the initiative, and it leaves university leadership guessing at how much of their own research budget planning should assume Genesis Mission dollars will offset, a real gap for anyone trying to model out multi-year research spending right now.
As Genesis Mission moves from selection announcements to active funding, research administrators at other universities should watch for whether DOE publishes award sizes and full project rosters, information that will matter for anyone trying to map the emerging landscape of federally funded, agent-driven scientific AI research and identify where the next round of partnership opportunities will open up before competitors lock in their own national lab relationships first.
