DARPA's Young Faculty Award 2027 Is Open, and the Real Deadline Is November 5

DARPA's Young Faculty Award is one of the few defense research programs built specifically for people who have never worked with the Department of War. It is also one of the most unforgiving on timing. The calendar on the listing says January 13, 2027. The date that actually decides who competes is November 5, 2026, fifteen days from today.

The 2027 Research Announcement, DARPARA2701, was posted to Grants.gov on October 20, 2026. It lists 18 topic areas, funds a 24-month base period of up to $500,000, and adds two options that can push a single award to $1,225,000. Every applicant must first clear a mandatory, blind-reviewed, one-page executive summary. If you miss that window there is no path to a full proposal, and the FAQ says so plainly.

This post covers what changed from the 2026 cycle, who is actually eligible under the funding-history test, how the 18 topics divide, and how to write a one-page summary that gets an "encourage" letter.

The numbers

The announcement says DARPA anticipates "multiple awards" and gives no award count or total budget. Its June 2026 retrospective said the program has supported more than 500 early-career researchers at more than 60 institutions over 20 years, which works out to roughly 25 awards a year on average. Treat any specific success-rate figure you hear secondhand as folklore.

What changed from 2026

Three changes matter.

The graduate student option got smaller and narrower. The 2026 call introduced an option to fund a "U.S. person" graduate student for up to an additional $350,000. The 2027 version caps it at $225,000, limits it to 12 months running concurrently with the base period, and restricts it to U.S. citizens. Permanent residents no longer qualify for the option. DARPA calls the option "an experiment" to build "a clearable pipeline of talent," and the switch from "person" to "citizen" fits that goal: citizenship, not residency, is what a security clearance requires.

The option is pre-priced but not automatic. It is exercised once the PI names an eligible student, the institution certifies citizenship on Attachment F, and the DARPA program manager concurs. If the student leaves and no eligible citizen replacement is found, the option is de-scoped. Proposers can structure it to start later in the base period, as long as it finishes inside the 24 months.

The total ceiling is now explicit at $1,225,000. That is $500,000 for the base, $500,000 for the Director's Fellowship and $225,000 for the student option. In practice most awardees will see the $500,000 base. In the 2024 class DARPA selected seven performers for Director's Fellowships, which gives a sense of how selective the third year is.

The topic list was rewritten around AI theory and numerical foundations. At least five of the 18 topics are about the mathematics of machine learning or computation rather than any particular application. More on that below.

The eligibility test most people get wrong

The PI must be, by the full proposal deadline, one of:

• A current tenure-track assistant or associate professor
• Tenured faculty within three years of their tenure date
• An equivalent at a nonprofit research institution within 12 years of the PhD

That part is simple. The funding test is where applicants disqualify themselves without realizing it. A PI who has received formally obligated funding of more than $500,000 from DARPA or more than $1,250,000 from all Department of War sources combined, as prime or subawardee PI or co-PI, is ineligible to lead. The announcement includes a table on attribution: if you were sole PI on a prime-only effort, count the entire funded amount; if you were PI on a subaward or one of several co-PIs, count only the funds allocated to your portion.

Two details matter here. First, the test is measured "as of the full proposal submission deadline," and the RA notes that if more funds are obligated after you submit an executive summary, you must recount. An incremental funding action on an existing Army or ONR grant in December could push you over. Second, the test counts obligated money, not award ceilings. A $2 million award that has obligated $900,000 counts as $900,000.

The hard exclusions:

• Prior YFA PIs can never be PI again, though they can serve as Senior Personnel.
• Current and former DARPA PMs cannot participate in any capacity.
• Staff at FFRDCs, UARCs, national laboratories, and other government entities cannot participate in any capacity. That rules out lab-joint appointees as Senior Personnel.
• Every team member must be employed at a U.S. institution.
• Citizens of China, Cuba, Iran, North Korea, Russia, and Venezuela cannot be Key or Senior Personnel, and the institution must certify this.

DARPA also says outright that it "encourages proposals from Principal Investigators who have not previously proposed to a Government solicitation." That is unusual language for a federal program and worth taking at face value. A short federal track record is not a weakness here.

The 18 topics

Proposals must align with exactly one topic. Here they are, grouped by our own reading of them:

Mathematics of AI and computation (5)

1. The Numerical Analysis of Iterative Prompt Engineering and Iterative Alignment
2. Numerical Stability Theory for Extreme Low-Precision Computing
3. Frugal Intelligence
4. Causal Unfolding of Stability Parameters
5. Towards a Science of Deep Learning

Information, networks and hardware assurance (5) 6. Multimodal Knowledge Completion Under Sparse Mission Data 12. Communicating Only Novel, Consequential Information via Semantic Encodings 13. Preserving Programmatic Intent in Hardware Synthesis 17. Photonic Activation Function 18. Emergent Communication Protocols for Distributed Systems

Physical sciences and materials (4) 2. Plasma as a Dynamic Heterostructure Layer for Electron Emission 3. Fundamental Mechanisms to Drive Scalable Thermomechanical Resilience 5. Algorithms for Mixed-Precision Quantum Information Processing 9. Biosynthetic and Biomimetic Pathways to Advanced High-Performance Material Formulations

Biology and human performance (3) 7. Pioneering Non-Invasive Deep Tissue Modulation 8. Programmable Biological Time 10. Continuous Critical Care

Social science (1) 11. A Pre-Columbian Approach to Understanding Sociopolitical Organization

The AI theory topics are notable for what they leave out. Topic 1 asks for frameworks that treat prompt engineering "as formal numerical algorithms," describing current practice as "brittle, practitioner-dependent empirical heuristics." Topic 16 asks for theory built on "well-established subfields of pure mathematics, such as geometry, dynamical systems, spectral theory, and harmonic" analysis. Topic 14 draws explicitly on Pólya's How to Solve It and the game of 20 Questions to build reasoning systems that minimize tokens, energy, cost and latency. None of them asks for a bigger model or a new benchmark.

That has a practical consequence for applied mathematicians, numerical analysts and theoretical computer scientists who have never thought of themselves as DARPA performers: five of 18 topics are written for you. Meanwhile, ML researchers whose work is primarily empirical will find the fit harder than the "AI" labels suggest.

Topic 11 is the outlier and deserves a mention. DARPA wants historical case studies of Pre-Columbian societies as a test bed for modeling sociopolitical evolution, valued for "falsifiable forecasts" and multigenerational data. Computational archaeologists and anthropologists rarely see a DoW solicitation written for their methods. This is one.

Writing the one-page executive summary

The executive summary step is where most of the competition happens, and it has three features that change how you should write it.

It is blind. Your name and institution may appear only on the cover sheet, which is separated before review. If identifying information shows up in the body, the summary "may be deemed non-conforming and removed from consideration." The FAQ acknowledges that this is awkward when your argument depends on your own prior results. Its answer is that the idea must "stand alone." Cite your preliminary data in the third person, or describe the result without attributing it.

It is one page, with no bibliography. Figures and tables count toward the limit. References cannot be added. So the summary has to make its case through clarity and specificity, not citation density.

The response is a signal, not a gate, but treat it as a gate. DARPA replies within about 45 days with an "encourage" or "discourage" letter. You may submit a full proposal either way, but DARPA says a discourage letter "should be viewed as a strong indicator that DARPA is not interested." Full proposals must include that letter as Attachment E. Submitted around November 5, a response should arrive near December 20, which leaves about three weeks, spanning the holidays, to write the full proposal. Start drafting the full proposal in parallel now rather than waiting for the letter.

What a strong one-pager looks like:

1. Quote the topic's own problem statement back, then name the gap precisely. Reviewers are checking alignment to a single topic first. A non-aligned summary can be declared non-conforming.
2. State a falsifiable 24-month outcome. The first evaluation criterion rewards proposals whose schedule "aggressively pursues performance metrics." Say what will be true at month 24 that is not true today, and how you will measure it.
3. Name the biggest technical risk and how you will retire it early. The criteria explicitly require risks and mitigations.
4. Tie the work to national security without overreaching. The second criterion is "potential contribution and relevance to the DARPA mission," meaning creating or preventing technological surprise. One sentence connecting your result to a plausible capability is better than a paragraph of speculation about weapons.
5. Do not propose finished work. Under the alignment rules, concepts already funded or receiving a "positive funding decision" from any agency are out of scope. If an NSF grant already covers the core idea, YFA will not pay for it again.

Budget and logistics people miss

• Travel is prescribed. Budget two 2-day meetings in the Washington, D.C. area and one 2-day meeting in San Diego across the base period, plus one more D.C. meeting if you receive the Director's Fellowship. Include travel in the graduate student option too. The FAQ discourages travel to non-DoW events.
• Each 12-month interval is capped at $250,000. You cannot front-load year one.
• Submit through BAAT, not Grants.gov. Each topic has its own listing in DARPA's Broad Agency Announcement Tool, and DARPA warns about heavy traffic on deadline day.
• Expect a research security review. DARPA runs a risk assessment of senior personnel for undue foreign influence before award, and all personnel supporting the project go through a pre-award registration process if selected.
• Equipment over $5,000 needs explicit justification in the cost volume.
• The mentorship is the program. Every awardee is assigned a DARPA PM mentor with related interests. Director's Fellowship nominations come from those PMs, based on "measurable progress" and "well-managed expenditure levels." Spend predictably.

Who should apply this cycle

• Applied mathematicians and theorists whose work fits Topics 1, 4, 14, 15 or 16. The fit is unusually direct.
• Junior faculty with Army, Navy or Air Force basic research grants that are still under the thresholds. Check your obligated totals now, and again in January.
• Faculty at institutions without large defense portfolios. The 2024 Director's Fellows came from places like North Carolina A&T, the University of Rhode Island and Rochester Institute of Technology, not just the usual top recipients.
• Researchers with U.S. citizen graduate students who can use the $225,000 option, which is effectively a funded clearable student position.

Who should skip: anyone whose only real fit is a stretch to one of the topics. Out-of-alignment summaries are non-conforming, and since you are allowed only one summary and one proposal, the first time-stamped submission is the only one DARPA reviews.

Related reading

For DARPA's other open doors this fall, see our analysis of the Defense Sciences Office office-wide BAA and the I2O office-wide BAA closing November 1. Early-career researchers comparing options should also look at the DOE Early Career Research Program results and the NSF CAREER award.

Granted can help you test your topic alignment and draft a blind, one-page executive summary against DARPARA2701's evaluation criteria before November 5. Start drafting with Granted.

Sources: DARPA Research Announcement DARPARA2701, Young Faculty Award (YFA) 2027, posted October 20, 2026, and DARPARA2701 Administrative and Technical FAQs dated October 19, 2026 (Grants.gov opportunity 364129); DARPA, "DARPA celebrates 20 years of Young Faculty Awards, announces Director's Fellows," June 2026.