S18.4 · Government, Defense & Public Sector
Technology supporting intelligence collection and analysis missions, shifting from cleared-analyst staffing to AI-native mission software.
This segment covers technology supporting intelligence collection and analysis missions for US and allied intelligence agencies. There is no defensible public figure for total segment size — the underlying budgets are classified — so the working proxy is the National Intelligence Program budget of roughly $70-72B plus the Military Intelligence Program at roughly $28B, a combined US intelligence-community budget near $100B (ODNI; Congressional Research Service R44381). The commercial story inside that opaque envelope is specific: budget is migrating from cleared-analyst staffing, billed by the hour, toward AI-native mission software, licensed as a platform, as large-language-model triage begins to absorb first-pass intelligence analysis. That is a reallocation, not new money — and the margin difference between the two revenue forms is enormous.
The National Intelligence Program has grown at roughly a low-single-digit annual rate since 2015 (ODNI, 2024); AI and data-analytics line items are growing considerably faster than the overall budget, which means they are taking share from everything else in it. The contractor base clusters around Washington DC and Northern Virginia, serving the CIA, NSA, National Geospatial-Intelligence Agency, National Reconnaissance Office and Defense Intelligence Agency; the UK and other Five Eyes partners maintain equivalent domestic bases, including BAE Systems Applied Intelligence and QinetiQ. Concentration is moderate: cleared primes — Leidos, Booz Allen Hamilton, Peraton, CACI and Palantir — sit above a deep tail of specialist boutiques.
The segment sits downstream of collection hardware built by defense-electronics and space-systems suppliers, and upstream of the analyst-facing mission software that consumes that collection. Contract type sets the economics. Collection contracts historically run cost-plus or time-and-materials but are pushing toward fixed-price software, where margins can be very high — Palantir has disclosed gross margins above 80%. Collection-hardware contracts tend to carry multi-year contracted backlog protection; analyst-support labor contracts are annually appropriated, rebid competitively and priced heavily on cost — the weakest revenue quality in the segment, attached to its largest historical revenue pool. The entry barrier is Top Secret/Sensitive Compartmented Information clearance with polygraph, plus SCIF facility accreditation — the highest clearance bar anywhere in the government and defense sector.
Adjacent markets include the collection platforms feeding this segment, the commercial cloud infrastructure the intelligence community increasingly runs on, and cybersecurity for government and defense — mapped separately — where signals-intelligence and cyber missions overlap. Established primes are defending cleared-labor recompete incumbency. Software-native entrants, notably Palantir and Anduril's Lattice platform, arrive from commercial data-platform roots rather than cleared-staffing businesses, and they are not competing for the billable hour — they are competing to make it unnecessary.
Large-language-model-based triage is absorbing first-pass all-source analysis, and with it the classic cleared-analyst billable-hour model that has been the segment's dominant historical revenue form. Headcount-based staffing incumbency gives way to platform licensing revenue, and the split between the two — AI-native mission software sold as a licensed platform versus traditional analyst-staffing contracting billed by the cleared hour — looks durable. The shift has been underway since roughly 2022, tracing to the Maven program and its successors; full re-platforming of the analyst workforce is a five-to-ten-year proposition given the pace of security accreditation.