ARPA-H Is About to Open a Funding Lane Where Winning an ARPA Award Before Disqualifies You

Federal agencies almost never write eligibility rules that exclude their own best customers. ARPA-H just did.

The Emerging Health Innovators (EHI) Initiative — an annual funding pathway ARPA-H has been building since 2024 and is now running as a recurring cycle — releases its 2026 solicitation on SAM.gov by the end of September 2026. A virtual Proposers' Day follows on October 13, 2026, 11:00 a.m.–2:00 p.m. ET, with registration closing October 12 at 11:59 p.m. ET.

The headline is not the money. It is the screen. To lead an EHI proposal you must be within eight years of completing your first terminal degree, you must not be a postdoc or trainee, your institution cannot be a national laboratory, FFRDC, UARC, or government agency, and — the clause that does the real work — you are ineligible if you have ever served as a prime project lead on an award from any ARPA agency.

Read that last one again. ARPA-H, DARPA, ARPA-E, IARPA. If you have run one as prime, you are out.

What ARPA-H is actually buying

ARPA-H's normal operating model concentrates risk in a small number of program managers who assemble teams from a repeat-player bench: large academic medical centers, established device and biotech firms, and the same handful of PIs who know how to write a solution summary and survive an oral pitch. That bench is efficient. It is also closed, and ARPA-H knows it.

EHI is the deliberate counterweight. The initiative was launched to widen access to federal health R&D funding for early-career investigators and community-based innovators, and the 2026 cycle is the first to run as a structured annual pathway rather than a one-off pilot. The prior EHI Innovative Solutions Opening, ARPA-H-SOL-25-118, carried a $700,000 minimum award, anticipated multiple awards and multiple award types, and specified Other Transaction agreements and cooperative agreements rather than standard research grants — with eligibility open to educational institutions, nonprofits, small businesses, for-profits, and individuals.

That instrument choice matters more than most applicants realize, and we will come back to it.

Sixteen topics, not an open call

This is the single biggest misread waiting to happen. EHI is not a "submit your best early-career idea" program. The 2026 cycle carries 16 named topic areas, and your proposal claims one of them:

1. Precision Control of Skin Inflammation without Systemic Exposure
2. Internal Sensing and Targeted Control of Inflammation
3. Deep Brain Stimulation for Mechanistic Sleep Exploration
4. CLEAR-MIND: Restoring the Brain's Drainage and Immune Balance to Enable Healing After Injury
5. Host-Pathway Interventions for Neurotropic Disease
6. Protecting Patient Data in Life Science AI
7. Engineering DRT for Programmable DNA Synthesis as an Emerging Medical Countermeasures Strategy
8. Prevention of Lethal Alpha-gal Allergic Responses Caused by Tick Bites
9. SMILE: Smart Microbiome-Informed Lesion and Enamel Restoration
10. Ending Preventable Deaths from Ectopic Pregnancy Through First-Visit Diagnosis
11. Transforming the Well-Woman Exam Through Objective Data
12. Innovations in Biotransport to Support Long-Term Preservation of Tissues and Organs
13. Organs-in-Action: Next-Generation Human Tissue Platforms for Longevity, Disease, and Therapy Discovery
14. Near Real-Time, Non-Contact Determination of Sterility in Distributed Manufacturing Systems
15. ICAT: Implantable Systems for Local Cancer Control at the Resection Margin
16. Causally-Grounded Alignment for Clinical Generative Artificial Intelligence Tools

Look at the shape of that list. It is not a survey of biomedicine. It is four clusters and a few singletons.

Inflammation control appears twice (topics 1 and 2), split by route — topical without systemic exposure, and internal sensing with targeted actuation. Neuro appears three times (3, 4, 5), and two of those are explicitly mechanistic rather than therapeutic. Women's health appears twice (10 and 11), and both are diagnostic-first framings: ectopic pregnancy caught at the first visit, and the well-woman exam converted from subjective to objective data. Tissue and organ preservation appears twice (12 and 13). Manufacturing, data security, medical countermeasures, dentistry, oncology, and clinical AI each get one.

The strategic reading: ARPA-H picked topics where a small team with one clever mechanism can produce a demonstrable result in a short program. Nothing on this list requires a Phase III trial. Several of them — sterility determination in distributed manufacturing, causal grounding for clinical generative AI, first-visit ectopic diagnosis — are fundamentally measurement problems, which is precisely the class of problem an eight-years-post-degree investigator can actually own.

The eight-year clock tightened

When EHI was first announced, the technology-driven track described early career as within 10 years of the final degree. The 2026 cycle describes the PI as within eight years of completing their first terminal degree.

Two years came off the window. If you are seven or eight years out, this may be your last or only EHI cycle — and the initiative is explicitly annual, so there is a next one, but you may not be in it. If you are two years out, you have several bites, which changes how aggressively you should chase a topic that is a poor fit.

"First terminal degree" is also doing quiet work. For an MD/PhD, the clock generally starts at the first of the two, not the more recent one — a detail worth confirming against the solicitation text the moment it posts, because it can move a borderline applicant to the wrong side of the line.

The exclusions are the opportunity

Most applicants read an exclusion list defensively: am I disqualified? The productive read is the opposite: who else is disqualified, and what does that do to my odds?

• No national labs, FFRDCs, UARCs, or government agencies. That removes a large fraction of the institutional capacity that normally competes for ARPA-style health money, along with the proposal-development infrastructure those institutions bring.
• No prior ARPA prime project leads. This is the load-bearing exclusion. Every investigator who has learned the ARPA idiom — the quad chart, the milestone table, the go/no-go structure, the "what would change if this worked" framing — is barred from leading. The people who are best at winning this money cannot compete for it.
• No postdocs or trainees leading. You need an independent position.

Put together, the eligible pool is: independent investigators at U.S. universities, nonprofits, and companies, within eight years of a terminal degree, who have never run an ARPA prime. That is a real population, and it is one that has mostly been writing R01s and K awards — a genre with almost nothing in common with an ARPA pitch deck.

Which is the actual competitive edge available here. The pool is large but the fluency is scarce. If you spend the next three weeks learning to write like an ARPA proposer instead of an NIH applicant, you are competing against a field that mostly will not.

Other Transactions are not grants, and the difference costs money

If the 2026 cycle follows the prior EHI opening, awards will come as Other Transaction agreements or cooperative agreements. Three consequences your research office needs to hear before, not after, you win:

Milestone-based payment. OT agreements typically pay against negotiated technical milestones rather than incurred cost. Miss a milestone and the money stops. This is a different cash-flow profile than a grant drawdown, and small institutions occasionally discover that fact in month four.

Indirect costs are negotiated, not applied. Your federally negotiated rate is a starting position in an OT, not an entitlement. Budget accordingly and get your business office in the room early.

Termination is structurally easier. ARPA agencies exit programs that stop producing. That is the model working as designed, but it means an EHI award is not the multi-year stability an R01 provides. This sits alongside a broader federal award environment where termination authority has been expanding — see our analysis of the 2 CFR Part 200 overhaul taking effect October 1, 2026.

Cooperative agreements behave more conventionally, but they carry substantial federal involvement — a program manager who is genuinely in your project, redirecting it. If that sounds intolerable, EHI is the wrong program.

What to do in the next 21 days

The solicitation is not out yet. That is an advantage, and it expires.

Confirm the clock. Pull the date of your first terminal degree. If you are inside eight years but close, get the exact conferral date documented now.

Audit your ARPA history — and your collaborators'. The prime-lead exclusion applies to the person leading the proposal. If your intended lead has an old DARPA prime in their history, restructure roles before you write, not after.

Pick exactly one topic. Not two. ARPA topics are written by a specific program manager with a specific hypothesis about why the field is stuck. A proposal that hedges across two topics reads as a proposal that owns neither.

Register for Proposers' Day by October 12. Attendance is not mandatory. It is also the only pre-solicitation venue where you hear the topic authors describe scope in their own words, and where you find the collaborator you are missing. For a first-time ARPA proposer, skipping it is a self-inflicted wound.

Draft the milestone table before the abstract. ARPA-style review rewards a credible sequence of falsifiable checkpoints far more than it rewards background. Write the go/no-go structure first and let the narrative follow it.

Early-career federal funding has spent 2026 getting narrower — K awards are losing their clinical trial authority, and high-risk NIH mechanisms are running on multi-year forecast horizons. EHI is one of the few lanes moving the other direction, and it was built specifically to be winnable by people who have never won federal R&D money before.

The catch is that it opens in about three weeks and closes on a schedule ARPA-H sets, not one you can negotiate. Track the SAM.gov posting, and have your topic chosen before it lands.