NSF's $83M Data Backbone: Inside the IDSS Awards (NSF 26-509), the Six Winners, and How to Position for the July 2027 Cycle

Most of the money and most of the headlines in AI-for-science go to two things: the compute and the models. NSF's Integrated Data Systems and Services (IDSS) program is funding the third thing — the part nobody puts on a slide but everybody trips over. On July 22, 2026, NSF announced $83 million in awards to six national-scale projects whose entire job is to make scientific data findable, accessible, movable, and usable by AI systems at scale. This is the plumbing. And after a decade in which "the data isn't ready" quietly killed more AI-for-science projects than any modeling failure, NSF is treating the plumbing as national infrastructure.

For research computing teams, data-repository operators, and institutions with serious cyberinfrastructure ambitions, IDSS is one of the most strategically important — and most misunderstood — programs in the current NSF portfolio. Here is how it is built, who just won, and how to be in the running for the next cycle.

What IDSS actually funds

IDSS (solicitation NSF 26-509) funds systems and services that connect the fragmented landscape of scientific data repositories to computing resources, AI tools, and cyberinfrastructure. The goal is not another dataset or another model. It is the connective tissue: the fabrics, platforms, and services that let a researcher discover data across institutional silos, move it to where the compute lives, and run reproducible, AI-driven workflows over it without reinventing the pipeline every time.

Brian Stone, performing the duties of NSF director, framed the bet plainly in the announcement: "America's leadership in artificial intelligence depends not only on advanced computing resources but also on the data infrastructure that enables researchers to discover, access, share and analyze scientific data at scale." That sentence is the whole thesis of the program. Compute without accessible data is a very expensive idle GPU.

The awards sit inside NSF's Computer and Information Science and Engineering (CISE) directorate and are explicitly tied to the White House's Genesis Mission — the multi-agency, multi-billion-dollar push to harness AI for scientific discovery. IDSS is the data layer of that effort. If Genesis is the strategy, IDSS is one of the load-bearing walls.

The six winners, and what the split tells you

NSF organized the awards into two categories, and reading the split is instructive.

Category I — National-Scale Integrated Data Systems (the anchor tier, for systems ready to operate at national scale):

• Fabric for AI-Driven Science (FabAID) — Morgridge Institute for Research, Madison, Wisconsin. A 5-year, $24.5 million project building a national data fabric that connects repositories, computing resources, and cyberinfrastructure — including the National AI Research Resource (NAIRR) — to streamline access to and curation of large-scale scientific datasets.
• National Data Platform (NDP) — UC San Diego. A platform for discovering, accessing, and working with distributed scientific data across communities.

Category II — Transition to National-Scale Operations (for systems maturing from a proven prototype toward national service):

• Interactive Discovery Laboratory (iDLab) — UCLA
• BRIDGE National Center — UC Irvine
• National Science Data Fabric (NSDF) — University of Tennessee, Knoxville
• Multidisciplinary Environment for Scientific Advancement (MESA) — University of Arizona

Two things jump out. First, this is a small, elite cohort — six awards out of a national field. IDSS is not a broad-participation program; it funds a handful of teams to build shared infrastructure that everyone else uses. Second, the FabAID number — $24.5M over five years — tells you the real scale of a Category I award. These are not typical single-investigator grants. They are multi-million-dollar, multi-year commitments to stand up systems that NSF expects to become durable national resources. That has direct implications for who can realistically compete, which we will get to.

The category structure is a readiness ladder — read it that way

The most important thing to understand about IDSS is that its three categories are not size buckets. They are maturity stages, and treating them as a ladder is the key to a viable strategy.

Category I (National-Scale Integrated Data Systems) is for systems that already work and are ready to operate at national scale. You do not propose here with an idea. You propose with a running, proven system and a credible plan to serve the national research community.

Category II (Transition to National-Scale Operations) is the on-ramp above the prototype line — for efforts that have demonstrated value in a narrower context and are ready to harden, scale, and transition into sustained national operation. Four of the six awards landed here, which signals that NSF sees the healthiest pipeline in the "proven-but-not-yet-national" band.

Category III (Planning Grants) is the entry point almost every new team should be looking at, and it is easy to miss in the award-announcement coverage because no planning grants made the July 22 headline. NSF 26-509 explicitly funds 1 to 5 Category III planning grants intended to help individuals or groups "improve their readiness for future submission of an IDSS Category I or Category II proposal." Translation: NSF will pay you to get ready to compete. If you have a promising data system but you are not yet at national scale, a planning grant is how you build the partnerships, governance model, sustainability plan, and technical roadmap that a Category I/II proposal is judged on.

This is the single most actionable insight in the program. Most teams that eventually win a Category I or II award do not spring fully formed into a $24M proposal. They climb: planning grant → transition → national-scale. If you are early, do not swing for the anchor tier and get discouraged. Aim at Category III, use it to close your readiness gaps, and set up the next rung.

The deadline structure — and why "the deadline just passed" is not the setback it sounds like

IDSS runs on a recurring annual cycle: full proposals are due the fourth Tuesday in July, annually. The most recent deadline was July 28, 2026 — the mechanics of that solicitation, including the award ceilings and cyberinfrastructure requirements, are broken down in our companion piece on the NSF 26-509 solicitation. The next full-proposal deadline falls on the fourth Tuesday in July 2027.

A full year sounds like a long runway, and for the wrong reasons teams treat it as one — then start the proposal in spring and discover that the hard parts of an IDSS proposal cannot be written in a quarter. The competitive parts of an IDSS submission are not prose. They are:

• A working system with evidence of use. For Category I/II you need real users, real throughput, real reliability data. You cannot manufacture that in the two months before a deadline.
• A multi-institution partnership structure. National-scale data fabrics are consortia. Lining up repository operators, computing centers, and domain-science partners — with real letters and real commitments — takes months.
• A sustainability and governance plan. NSF is funding infrastructure it expects to outlive the award. Reviewers want a credible answer to "what happens in year six?" That answer requires institutional buy-in you have to negotiate now.
• A NAIRR / national-cyberinfrastructure integration story. FabAID's explicit NAIRR connection is not incidental. Alignment with the national AI research resource and the broader Genesis data agenda is a differentiator, and it requires you to understand where you fit in an architecture that is still being drawn.

So the year is not slack. If you intend to compete in July 2027, the work — partnership formation, usage evidence, governance design — starts in the second half of 2026. And if that readiness work is exactly what you lack, that is the argument for a Category III planning grant.

Who should actually pursue this

Be honest about the tier that fits you.

Pursue Category I only if you operate a data system that is already serving a research community at meaningful scale and you can point to usage, reliability, and a multi-institution coalition ready to go national. This is a short list of teams nationwide.

Pursue Category II if you have a proven prototype — demonstrated value in a specific domain or region — and a realistic path to hardening it into sustained national service. This is where NSF sees the most opportunity, based on the award split.

Pursue Category III (planning) if you have a promising system or a serious concept but gaps in partnerships, governance, sustainability, or national-scale technical design. This is the right target for most teams, and it is deliberately structured as an investment in your future competitiveness rather than a consolation prize.

And if none of those fit — if your asset is a dataset rather than a system for connecting datasets — IDSS is the wrong door. The right one is likely NSF's AI Datasets program (NSF 26-512), which funds making individual scientific datasets AI-ready. IDSS funds the fabric; AI Datasets funds the threads. Knowing which program matches your asset is half the battle; teams routinely burn a cycle proposing to the wrong one.

The bigger signal

Step back and IDSS reads as a statement about where federal AI-for-science strategy has matured to. Two years ago the money chased models. Now NSF is spending $83 million on the unglamorous middle layer — the fabrics and platforms that determine whether all that compute and all those models ever touch usable data. That is what a field looks like when it stops being a demo and starts being infrastructure.

For institutions with cyberinfrastructure ambitions, the message is clear. The programs funding the data backbone — IDSS, AI Datasets, NAIRR, and the wider Genesis Mission apparatus — are becoming a coherent, well-funded layer of the research economy. The teams that win in it will be the ones that started building partnerships and usage evidence a full cycle before the deadline. The next fourth Tuesday in July is closer than it looks.

Granted helps research teams, small businesses, and institutions find and win federal funding. Search live grant opportunities or explore our funder profiles to build your pipeline.