Earlier this year, the Secretary of War declared a goal to transform America’s armed forces into an “AI first” military.
That transformation is essential to maintaining US military dominance, but the effort is already running into a major obstacle: there’s a shortage of staff at the Pentagon with the expertise needed to integrate and wield the technology effectively.
Successfully building and adopting AI in crucial national security functions depends on the government possessing enough experts who can understand, test, validate, and build this emerging technology.
Right now, America is struggling to fill this gap.
The challenge is not just filling current vacant roles. The Pentagon needs personnel who can test commercial AI systems and judge their reliability under unique military settings.
Specifically, this talent gap impacts the ability to assess and validate commercial AI systems being sold to the government.
As a result, the government will struggle to evaluate performance, manage risk, and retain control over mission-critical capabilities.
Additionally, our military crucially needs the ability to build mission-tailored models. Without that specialized competency inside government, the US risks becoming dependent on private companies to operate and build this technology.
Washington Is Moving, But Not Fast Enough
Two branches of government are already moving on this, but not far enough.
On June 2, the White House issued an executive order on AI cybersecurity and innovation, and three days later, President Donald Trump signed National Security Presidential Memorandum 11 (NSPM-11), directing the national security enterprise to accelerate AI adoption.
NSPM-11’s Section 5 does more than gesture at the workforce problem: it directs agencies to use special hiring and pay authorities where they already exist, and it gives the Office of Personnel Management 120 days to stand up an AI National Security Strategic Reserve of vetted, non-governmental experts.
That’s a meaningful step, but the memorandum sets up the reserve as a directive, not a durable program. It carries no dedicated appropriation and no statutory footing, which means it can be slow-walked, underfunded, or unwound by a future administration with the stroke of a pen.
Congress has its own vehicle. The House-passed version of the FY2027 National Defense Authorization Act contains personnel pilots aimed at technical talent recruitment.
But they’re narrow: one leans on the size and durability of the existing US Digital Corps-style Tech Force program, and another authorizes only a temporary pilot rather than the broader hiring and pay reforms the Pentagon actually needs.
Cleared Reserve for Crisis Moments
Closing this gap will take AI-specific measures, not AI folded into talent initiatives meant to broadly serve the whole department.
Congress needs to enact policies that identify the talent that already exists but cannot yet be effectively used, onboard new talent to support the government during a crisis, and construct the permanent scaffolding needed to recruit and retain experts.
The most viable and needed effort is exactly what NSPM-11 gestures at but doesn’t lock in: a cleared standing reserve of AI experts from industry and academia who could be activated during a crisis involving AI.
Congress should codify the reserve NSPM-11 proposes, fund it, and give it the durability an executive memorandum can’t. A congressionally established reserve would also circumvent the clearance bottleneck by pre-processing clearances before skills are urgently needed.
Permanent Career Track
A reserve is essential to address a crisis, but more permanent talent is also needed.
The easiest path to build it is to find people already in service who are not easily identifiable as AI experts. Each military service should stand up a dedicated AI and machine learning career track spanning both uniformed and civilian personnel, defining training pipelines and ensuring people can move into it without losing career progression.
To recruit new talent, there needs to be a special AI incentive and pay system.
With pay for frontier AI talent far outpacing federal ceilings, no career track or recruiting pilot will reach its full potential while the government is capped below market for this specific skill set.
Congress should create national security-specific pay authority for AI and data science roles, exempt from current GS and SES caps, designed to make government compensation more competitive with the private sector.
Congress has decoupled pay from the GS scale before when markets demanded it — the Cyber Excepted Service is the template — and it’s time to do so again for AI.
Building the Pipeline Early
Another way to obtain talent is to target students entering higher education. A scholarship-for-service program, similar to CyberCorps, would fund graduate-level AI education in exchange for a post-graduation service obligation.
This shifts the competition with industry to an earlier stage, when the compensation gap is narrower and the pitch doesn’t rely exclusively on matching a private salary.
While the pilots in the current NDAA are time-limited experiments, a fellowship pipeline builds a renewable, long-term supply that can endure administrations and budget cycles alike.
The Stakes
The country that wins the AI race will be the one that can recruit and surge the people who understand and build the technology.
Congress and the White House already agree that AI talent is a national security priority, but the current measures still gesture at requirements the Pentagon and the intelligence community already understand well, without the funding or permanence to meet them.
Permanent, targeted solutions are what will build a workforce able to operate a technology that can transform the nature of war itself.
Congress should codify four measures into durable law: a cleared reserve, a joint career track, competitive pay authority, and a fellowship pipeline.
Jacqueline Viteznik is a policy analyst at Americans for Responsible Innovation, a nonprofit organization dedicated to policy advocacy in the public interest, focused on emerging technologies like artificial intelligence.
She is also a former graduate consultant and data analytics intern at Department of State and former program management intern at UNDRR.
The views and opinions expressed here are those of the author and do not necessarily reflect the editorial position of Military AI.
Have a perspective to add? See our Write for Us page.
