Booz Allen's Civil Revenue Fell 15% While Its Peers Grew: Federal IT Dollars Are Being Re-Scoped, Not 'AI-ified'

Two numbers in the same release point in opposite directions, and the market mostly saw one of them. Of the $112 million year-over-year revenue decline, roughly $97 million came from billable expenses — pass-through subcontract and equipment costs — which fell 10.6% to about $821 million from about $918 million. The remaining $15 million, or 0.8%, is the actual fee pool: revenue excluding billable expenses of $1,957 million. The fee pool, not the headline, is the honest measure of the work, and it was essentially flat.

The civil problem is not agency demand, it is scope. Management attributed the 15% decline in Civil and Commercial revenue — $702 million in the quarter — to recompetes that transitioned to follow-on contracts with reduced scope, and guided that segment to be roughly flat sequentially in the third and fourth quarters. The decline reflects contract structure, not a collapse in the civilian budget, which is why the same quarter produced a 15% civil decline and a 6% National Security gain excluding billable expenses.

Scale matters here: Booz Allen's civil and commercial book fell $922 million in fiscal 2026, against a civilian agency IT budget that fell $7.2 billion, from $75.1 billion in 2025 to $67.9 billion for fiscal 2026. The segment also carries commercial clients, so the comparison is indicative rather than exact, but the order of magnitude tells you the loss was not a straight read-through of the budget line. The civil damage is concentrated in recompete scope, not in the size of the civilian wallet.

It would be easy to stop at 'federal IT budgets are being cut.' Civilian agency IT spending fell about 9.6% to $67.9 billion for fiscal 2026, and the fiscal 2027 proposal only asks to return to the 2025 level. But in overlapping quarters CACI grew revenue 8.2% to $2,475 million in the identical September quarter, SAIC grew 6.3% to $1,880 million with civilian revenue up 9.1%, and Leidos grew 7.2% to $4,558 million. The contraction is not uniform, and one integrator is losing share while its peers grow.

The 'AI replaces services' thesis has a measurable signature, and it is not visible in the services companies' revenue. Federal agencies have obligated $4.1 billion across 2,255 AI-related awards since fiscal 2019, with $1.162 billion obligated in fiscal 2025 alone, and 83% of those dollars went to the Defense Department. Over the same window Palantir grew US government revenue 90% to $809 million in a single quarter, on a 62% adjusted operating margin. The AI dollar is funding platforms and defense capacity, not civilian advisory hours.

Federal services is a chain, and this quarter moved money at both ends of it. The upstream end is the prime-contract layer and the platform layer; the downstream end is the agencies and the commercial market that absorbs capacity freed up when civilian budgets shrink. The transmission runs through recompete vehicles upstream and through civilian agency demand downstream, and the two ends are moving in opposite directions.

Booz Allen kept FY2027 revenue guidance at $11.3 billion to $11.6 billion while the first half fell 4.1%, from $5,814 million to $5,578 million. That implies second-half revenue of $5.72 billion to $6.02 billion against $5,403 million a year earlier — growth of 5.9% to 11.4% after a first half that shrank. Management's own framing confirms the strain: FY2027 guidance assumes mid-single-digit growth in National Security and a low double-digit decline in Civil, with second-half margins stepping down on seasonality, lower civil volumes and the timing of strategic investments.

The first hard denominator on the federal-AI thesis says something narrower and more investable than 'services are dead.' Federal IT budgets are contracting in nominal terms, but the integrators holding recompete vehicles and mission technology are growing, and the AI money is consolidating into defense platforms and software rather than advisory hours. Booz Allen is the clearest expression of the wrong side of that trade: a shrinking civil book, a national security business doing the heavy lifting, and a second half that must accelerate to stay in range. The stock is a bet on national security execution, not on a federal IT recovery.