NSF Rebuilt Its Scholarship-for-Service Program Around AI. The July 21 Deadline Funds $2.5M Per Institution — but the Award Goes to Schools That Can Prove a Government Hiring Pipeline, Not Just a Curriculum.

The federal government has a recruiting problem that money alone has never solved: it cannot compete with the private sector on salary for the AI and cybersecurity talent it most needs. For more than two decades, its answer has been the CyberCorps Scholarship for Service (SFS) — a program that pays for students' education in exchange for a binding commitment to work for the government after graduation. In 2026, NSF rebuilt that program around the technology now reshaping both fields, releasing NSF 26-503, the CyberAICorps Scholarship for Service (CyberAI SFS).

The Scholarship Track carries a hard deadline of July 21, 2026 — and NSF has stated plainly that "Scholarship Track proposals submitted after July 21 will not be accepted." For universities weighing whether to compete, the question is not whether the program is generous. It plainly is. The question is whether your institution can credibly deliver the one thing the program is actually buying: graduates who go into government service.

What the money covers

CyberAI SFS is unusually rich on a per-student basis, which is what makes it a genuine recruiting tool rather than a token subsidy. The awards work on two levels.

At the institutional level, a project can receive up to $2,500,000. A typical award supports two cohorts of students, layered so that the program runs continuously: a first-year cohort that may include students receiving two or three years of support, and a second-year cohort of students receiving two years.

At the student level, each scholarship recipient receives:

• An annual stipend of $27,000 for undergraduates or $37,000 for graduate students.
• Full tuition and education-related fees.
• A $6,000 annual professional allowance covering job-fair and conference travel, certifications, books, supplies, professional training, and a one-time laptop purchase.

Support for any individual student is capped at three years. The combination — full tuition, a livable stipend, and a real professional-development budget — is designed to make a government-service track financially competitive with the private-sector offers these students would otherwise chase.

The obligation is the product

Here is the part that distinguishes SFS from an ordinary scholarship, and the part that shapes everything about a competitive proposal: the scholarship is a service contract, not a gift.

Recipients incur a post-graduation service obligation in a qualified government organization — federal, state, local, or tribal — for a period at least equal to the length of the scholarship. A student funded for two years owes at least two years of government service. During the scholarship period itself, recipients must also complete qualified summer internships related to AI or cybersecurity, typically in government settings, which serve as the on-ramp to permanent placement.

This structure is why the program's success is measured in placements, not graduations. And it is why NSF evaluates proposals less on the elegance of a curriculum and more on whether an institution has the machinery to route students into qualifying government jobs. A university can teach an excellent AI-security course and still fail the program's actual purpose if its graduates can't — or don't — convert into government service.

Why placement infrastructure wins the award

If you read the program through the lens of "we'll design great courses," you will write a losing proposal. The institutions that win CyberAI SFS awards are the ones that can demonstrate, concretely, the infrastructure that turns scholarships into government hires. That infrastructure has four components.

1. Active relationships with hiring agencies. Winning programs maintain standing relationships with federal agencies (think DHS/CISA, NSA, the national labs, civilian CIO shops) and increasingly with state and local government security offices. The proposal should name partners and describe the pipeline — internships that lead to offers, hiring managers who return each year, alumni placed across the past cohorts.

2. A demonstrated internship-to-placement conversion. Because summer internships are mandatory and are the primary feeder into permanent roles, reviewers want evidence that your internships actually convert. Existing SFS programs with track records cite their placement rates; new applicants must show a credible plan and the partnerships to back it.

3. Designated coordination capacity. SFS programs live or die on a dedicated coordinator who manages the relationship with NSF and the federal SFS program office, tracks each student's service obligation, runs the placement process, and ensures compliance. A proposal that buries this role in faculty release time signals it doesn't grasp the operational lift.

4. A recruiting pipeline that reaches the right students. The program's diversity and breadth goals mean reviewers look favorably on recruitment strategies that draw from community colleges, regional institutions, and populations underrepresented in the federal cyber and AI workforce. The AI dimension of CyberAI SFS specifically broadens the relevant student pool beyond traditional cybersecurity majors into data science, machine learning, and AI-assurance fields.

What's new about the "AI" in CyberAI SFS

This is not merely a rebranding of the cybersecurity SFS program. The addition of AI reflects a genuine shift in what the federal government needs to defend and to build: AI systems that are themselves attack surfaces, AI tools that augment cyber defense, and the assurance, governance, and security of AI deployed across government missions.

Practically, that widens both the curriculum and the eligible student population. A competitive CyberAI SFS proposal should articulate where AI and cybersecurity genuinely intersect in its program — secure machine learning, adversarial robustness, AI for threat detection, the security of AI supply chains — rather than bolting an AI course onto an existing cyber track. Institutions that built AI capacity over the past few years, and can fuse it with an established cybersecurity program, are positioned to make the strongest case.

Where this sits in the July NSF calendar

CyberAI SFS lands in an exceptionally crowded NSF window. The July 21 Scholarship Track deadline shares the calendar with a dense run of opportunities, including:

• July 1 — Archaeology Program Senior Research Awards
• July 15 — Arctic Research Opportunities; IUSE
• July 21 — CyberAI SFS and EPSCoR E-CORE
• July 22 — CAREER, NSF's flagship early-career award
• July 27 — SBIR/STTR Phase I/II

For research-development offices, that clustering is a resource-allocation problem: the same grants staff supporting a CAREER submission on July 22 may be needed for a CyberAI SFS package on July 21. Institutions serious about CyberAI SFS should be staffing it as a distinct effort now, not assuming it can ride the same bandwidth as the rest of the July pipeline. (For the companion infrastructure program on the same date, see our deep dive on NSF EPSCoR E-CORE.)

Who should apply

The honest eligibility picture: CyberAI SFS is open to accredited institutions of higher education, but it rewards a specific institutional profile.

• Strongest fit: institutions with an existing SFS award or a designated cybersecurity program (e.g., NSA/DHS Centers of Academic Excellence) that have now built real AI capacity, and that already place graduates into government roles.
• Viable with effort: institutions with strong AI or cybersecurity programs and emerging government partnerships, provided they can stand up the coordination and placement infrastructure the program demands.
• Premature: institutions with classroom strength but no government-hiring relationships and no placement track record. The service obligation is the product; without a pipeline to honor it, the proposal can't deliver.

The bottom line

CyberAI SFS is one of the best-funded workforce programs NSF runs, and the per-student economics genuinely make government service competitive with private offers. But the award is not a teaching grant. It is a contract to supply the federal, state, local, and tribal governments with AI and cybersecurity talent — and NSF funds the institutions that can prove they will deliver that talent into actual jobs.

With a July 21 deadline that NSF has explicitly said it will not extend, the institutions positioned to win are the ones that already have the agency relationships, the internship-to-hire conversion, and the coordination capacity in place. If you're building those now, you're building for the next cycle. If you have them already, the next three weeks are about translating that infrastructure into a proposal that makes the pipeline visible.

Granted tracks NSF education and workforce programs alongside the full federal funding calendar. For more on AI funding at NSF, see our complete guide to NSF AI funding programs and Granted News.