The Growth Engine

The Growth Engine

Booz Allen's growth engine — AI, cyber, and defense technology — is real: AI revenue grew about 30% to roughly $800 million in FY2025, and the firm is the federal government's largest AI provider [1]. But at around 7% of revenue it is still too small to outrun a $922 million Civil decline, its precise size is no longer disclosed, and management is funding it partly by redirecting capital away from buybacks into $500 million of illiquid venture commitments [2]. Whether this engine scales fast enough to carry the National Security core remains unproven.

What the engine is

The clearest evidence that Booz Allen is more than a labor-arbitrage staffing house is its artificial-intelligence business. Management sized it at nearly $600 million in FY2024 and about $800 million in FY2025, a gain of over 30% in a single year, and cited Deltek's ranking of the firm as the leading AI provider to the federal government [3] [4]. The FY2026 10-K describes roughly 400 active AI projects spanning agentic AI, physical AI, and AI-Radio Access Networks, alongside a cyber business protecting federal, defense, and Fortune 500 customers [5].

Sources: AI revenue as disclosed on the Q4 FY2025 call [6]; total revenue per the FY2026 10-K, as reported.

The engine is broader than AI. In FY2026 the firm reported a nearly 90% rise in Other Transaction Authority (OTA) proposal submissions and about a 50% rise in OTA awards — the faster, commercial-style contracting the administration is pushing — plus new product lines in its Vellox cyber suite and EdgeXtend edge-computing family, and partnerships with Nvidia, AWS, and a16z portfolio companies [7]. Management frames these as its "growth vectors" and, for FY2027, organizes the entire strategy around them: accelerating cyber and defense tech, injecting AI into all work, seeding the next wave (physical AI, quantum, 6G, AI-RAN), and monetizing its venture network [8]. The single largest near-term prize the firm names is Golden Dome missile defense, funded at $25 billion in the reconciliation package with roughly $24 billion in adjacent missile-defense money, where Booz Allen is positioning its "Brilliant Swarms" space-based interception concept [9].

The engine is real, but it is small

The arithmetic is where the case has to be honest. AI at roughly $800 million is about 6.7% of FY2025 revenue; growing it 30% adds on the order of $240 million a year. Civil revenue fell $922 million in FY2026 alone, and National Security — 71% of the company and the home of most of these growth vectors — grew only about 2% that year despite the backdrop [10]. A high-growth business that is one-fourteenth of the whole cannot yet offset a collapse in the piece that broke. That is the mechanical reason a company with a genuine AI franchise still guided total revenue roughly flat for FY2027, at $11.2 billion to $11.7 billion [11].

Source: AI growth-rate applied to the ~$800M FY2025 base [12]; Civil decline per the FY2026 10-K revenue-by-customer table [13].

There is also a disclosure tell. Management volunteered a precise AI number for FY2024 ($600 million) and FY2025 ($800 million, over 30% growth), the years the figure flattered the story [14] [15]. For FY2026, the year total revenue fell 6%, the CEO said only that the AI business was "still growing and expected to grow significantly this year," with no dollar figure [16]. The original ambition was to pass $1 billion "in the next couple of years" from the FY2024 base [17]. A metric that a company stops quantifying is not proof the metric turned down — the AI dollars are increasingly embedded across contracts and hard to isolate, which is the benign reading — but a reader tracking whether the engine is scaling has lost the cleanest gauge management used to provide.

The capital pivot

The more consequential move is what Booz Allen is doing with its cash. During FY2026 the firm committed, through its operating subsidiary, up to $400 million to Andreessen Horowitz's late-stage venture fund and up to $100 million to a second fund — $500 million in total, structured as limited-partner interests accounted for on the equity method and disclosed as related parties, subject to capital calls over the funds' 10-year terms [18]. It funded $178 million of that in the fourth quarter, leaving $322 million of unfunded commitments outstanding at year-end [19]. The a16z tie-up made Booz Allen a16z's first-ever "Technology Acceleration Partner for Governments" [20].

That commitment reordered the capital-allocation mix inside a single year. In the first quarter of FY2026, Booz Allen deployed $154 million to buybacks and $70 million to dividends — $224 million to shareholders — against just $9 million into ventures [21]. By the fourth quarter, the ratio had flipped: $219 million into strategic and venture investments versus $147 million returned to shareholders [22]. Total investing cash outflow rose to $300 million for the year from $218 million, which the 10-K attributes primarily to these venture and fund investments [23].

Source: quarterly capital-deployment disclosures, Q1 and Q4 FY2026 earnings calls [24] [25].

The pivot deserves a professional investor's scrutiny because of its timing. Booz Allen chose to lock up half a billion dollars in illiquid, related-party late-stage venture funds — a 10-year term, front-loaded capital calls, marks rather than realizations — in the same year its own equity fell to a multi-year low and traded near an 11% free-cash-flow yield (Priced-In Pessimism). The case for it is strategic rather than financial: the funds buy pipeline access, co-creation rights, and early sight of dual-use technology the firm can carry into classified missions, which a buyback cannot [26]. The case against it is that management is deploying capital into someone else's fund at venture-market valuations while declining to maximize repurchases of a business it describes as advantaged and the market prices for decline. Both can be true; the test is whether the venture network converts into contract revenue that a passive buyback never could.

Venture Marks and the Margin Promise

The venture book is already touching reported earnings in a way that connects to the quality questions raised earlier (Financials and Estimates). Fourth-quarter FY2026 adjusted results included $12 million of pretax unrealized gains on the ventures portfolio, and management quantified the full-year contribution at $0.11 of adjusted EPS [27]. Notably, the FY2027 guidance explicitly assumes no contribution from strategic venture investments [28]. Management is, in effect, telling investors these mark-to-market gains are real enough to flow through adjusted EPS when positive but too volatile to guide on — which is the correct treatment, and also a reminder that a slice of FY2026's adjusted earnings came from non-operating marks that can reverse.

The other claim to test is margin. Management calls the growth vectors "fast-growing, highly profitable" and says that as they outperform it expects "margin expansion across the portfolio" [29]. The FY2027 guidance does not yet reflect it: adjusted EBITDA of $1.24 billion to $1.29 billion on $11.2–11.7 billion of revenue implies a margin of about 11%, in line with recent years rather than above them [30]. The higher-value, product- and outcome-based economics management describes are a plausible source of future margin, but for now they are an assertion the guidance has not cashed.

What would decide it

The bull reading is that Booz Allen has spent a decade positioning for exactly this moment — largest federal AI provider, a cyber and defense-tech book aligned to the administration's priorities, and a venture network that feeds it commercial technology at speed — so the ~2% FY2026 National Security growth understates a backdrop about to accelerate as procurement reforms and defense appropriations land [31]. The bear reading is that the engine is still too small and too slowly disclosed to prove itself, the margin uplift has not appeared, and the capital funding it is being diverted from a cheap own-stock into illiquid third-party funds. The evidence available today supports a measured read: the growth vectors are genuine and strategically coherent, but they have not yet reached the scale or the margin at which they visibly carry the company — and the clearest confirmation would be National Security growth reaccelerating in FY2027 while the AI figure returns to the disclosure it once had.