NIH Will Pay $650,000 for Autoimmune Nutrition Apps. Phase 1 Opens October 1 — and the Money Arrives After You Build the Thing.

Phase 1 of the NIH NOURISH Autoimmunity Digital Health Challenge opens on October 1, 2026 and closes on January 8, 2027 at 8:00 PM ET. The total purse across three phases running to August 2028 is $650,000.

That number will attract a lot of people who should not enter, and will be dismissed by some who should. The reason is that NOURISH is not a grant, and almost every instinct a grant-writing team has about federal money is wrong here.

The structure, precisely

NOURISH — Nutrition for OUR Immune System Health — asks for digital health applications that help people living with or at risk for autoimmune disease understand how diet and nutrition affect their condition. The sponsors are the NIH Office of Autoimmune Disease Research within the Office of Research on Women's Health (OADR-ORWH), the NIH Coordinating Committee for Autoimmune Disease Research, the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), the Office of Nutrition Research, and the Office of Dietary Supplements.

The prize ladder:

Phase 1 judging runs January–February 2027, with winners announced in March 2027. Note the gaps: four months between the Phase 1 deadline and the Phase 1 announcement, and another four months between that announcement and the Phase 2 window opening. A team that wins all three phases collects $150,000 across roughly twenty months, with the large majority of it arriving in the summer of 2028.

Phase 1 submission is three pieces, all due January 8:

1. The electronic Challenge Registration form.
2. A PDF narrative of no more than six pages — 11-point Calibri, one-inch margins, entity or team-leader name and idea title in the header. References, appendices, figures, and tables all count inside the six pages.
3. A video demonstration of three to five minutes showing the prototype actually working. NIH says explicitly it need not be highly produced, but it must be clear and legible.

Both the narrative and the video are judged against five criteria, all weighted equally: understanding of the problem and scientific rationale; engagement; potential impact; feasibility; and innovation.

The word doing the most work in that list is "prototype"

Read the Phase 1 requirements again. A three-to-five-minute video demonstrating prototype functionality is not a mockup, a wireframe deck, or a narrated Figma file. To submit a competitive Phase 1 entry on January 8, you need working software by roughly mid-December.

This inverts the cash-flow logic of every grant mechanism you are used to. An R01, an R21, an SBIR Phase I — all of them pay you to do the work. NOURISH pays you after you have done a meaningful slice of it, and pays only ten of you. The $20,000 Phase 1 prize is not development capital. It is, for most winners, a partial reimbursement of sunk engineering cost.

The honest comparison is to an NIH SBIR Phase I, which at NIH commonly runs in the low-to-mid six figures and funds the build. If your goal is to finance the construction of an autoimmune nutrition app, SBIR is the mechanism and NOURISH is not. If you already have a prototype — because you built it on private capital, on an accelerator, on institutional discretionary funds, or as a graduate project — NOURISH is an unusually efficient way to convert it into federal validation.

The rule that disqualifies the most obvious applicant

Buried in the terms is a provision that matters enormously and reads as boilerplate: participants may not use federal grant or contract funds to develop their submission.

Think about who the natural applicant is. A digital health researcher in a university lab studying diet and autoimmunity. Someone with the domain expertise, the clinical collaborators, and possibly an existing NIH award in an adjacent area. That person's salary, their postdocs' salaries, and their software engineer's effort are very likely charged to federal awards.

If so, the development cannot be done on that effort. It has to come from institutional funds, foundation money, startup packages, departmental discretionary accounts, or genuinely uncompensated time. This is not a technicality your sponsored programs office will wave through — it is a term of participation, and the winner-verification step requires submitting all requested documentation within 20 business days of notification.

The practical consequence: talk to your research administration office before you start building, not before you submit. Establishing a clean, non-federal funding line for the development work is a weeks-long conversation at most institutions, and you have until roughly mid-December to have the software working.

For an independent developer or a pre-revenue company with no federal awards, this rule is a non-issue and quietly a competitive advantage.

What you actually keep

NIH is clear on intellectual property, and the terms are better than most people assume:

• You do not transfer IP to NIH to receive an award. You keep what you build.
• You must grant NIH an irrevocable, paid-up, royalty-free, non-exclusive worldwide license to publish, post, link to, share, and display the submission publicly. That is a publicity license over the submission, not over the product.
• Source code may be requested for validation but will not be publicly released.
• Winners consent to NIH using name, likeness, photograph, voice, and submission details promotionally, unless they opt out in writing.
• AI tools are permitted, with full disclosure of every tool used.

Compare that to the IP posture of a cooperative agreement or a contract and the prize mechanism looks generous. The trade is that you get no cost reimbursement, no indirect cost recovery, no negotiated rate, no facilities-and-administrative dollars at all. There is no F&A line in a prize. For a university team accustomed to recovering 50-plus percent on top of direct costs, a $20,000 prize is $20,000 — and the institution's finance office will treat it very differently from a grant.

It is also, generally, taxable income rather than a cost-reimbursable award. Budget accordingly; the net to a small company after tax is meaningfully less than the headline.

Eligibility is narrower than it looks

Participants must be U.S. citizens or permanent residents, and entities must be incorporated in and primarily doing business in the United States. Federal employees acting in their official capacity, NIH staff, and Challenge judges are excluded. Participants under 18 need parental or guardian consent unless emancipated. Each person is limited to one submission role — you cannot hedge by appearing on three different teams.

The citizenship requirement rules out a lot of otherwise ideal postdoctoral talent and a lot of international digital health teams. Read it before you assemble a team, not after.

Who should enter, and who should not

Enter if:

• You already have working software in the autoimmune, nutrition-tracking, symptom-logging, or continuous-glucose-monitoring space and can adapt it. Phase 1 rewards demonstrated function, and adaptation is cheaper than construction.
• You have a clean non-federal funding path for the development work.
• The three-phase arc is strategically useful to you independent of the money — NIH validation, a named Office of Autoimmune Disease Research association, and a public demonstration path through August 2028 are worth real money in a seed round, and worth more than $150,000 to a company whose next raise depends on credibility with clinicians.
• You want a non-dilutive entry point into the NIH ecosystem without a UEI, a SAM.gov registration cycle, and a Uniform Guidance compliance posture.

Do not enter if:

• You need the money to build the thing. The sequencing does not work.
• Your development effort is unavoidably on federal awards and your institution cannot carve out a non-federal line in time.
• You are treating it as a lottery ticket. Ten Phase 1 slots against a purse this visible, in a disease area with an extremely engaged patient community, is not a soft field — and a six-page narrative plus a live video demo is a real submission, not a form.

The strategic read

Prize competitions under the America COMPETES authority have become NIH's preferred instrument for problems where the agency knows the outcome it wants but does not want to specify the method, and where the likely solvers are not traditional grantees. NOURISH fits that exactly: the sponsors are offices, not institutes with large extramural budgets, and the target builder is a small digital health team that would find an R01 application unrecognizable.

Read it that way and the $650,000 stops looking small. NIH is not trying to fund the development of an autoimmune nutrition platform. It is trying to find out whether one exists, put three of them in front of judges in 2028, and create a credentialed on-ramp for the winners into the much larger funding streams that follow. The prize is the audition, not the job.

Phase 1 opens October 1. The binding constraint is not the January 8 deadline — it is having working software in December.