Booz Allen CEO warns Chinese AI models pose 3 national security threats

Booz Allen's CEO said Chinese AI models threaten US national security as the defense contractor pivots toward autonomous systems.

Chinese AI models pose three distinct threats to US national security — from embedded vulnerabilities to agentic warfare — Horacio Rozanski, Booz Allen's chairman and CEO, said, as the defense contractor reported $11.2 billion in fiscal 2026 revenue.

"Chinese AI models present risks across the software supply chain, and we are seeing agentic AI threats emerge in real time," Rozanski said on Fox Business's Mornings with Maria on May 31.

Booz Allen posted $11.2 billion in revenue for fiscal 2026, down 6.4% year over year, while adjusted EBITDA reached $1.23 billion. The company closed the year with a record backlog of $38 billion and a trailing 12-month book-to-bill ratio of 1.1x. Rozanski described fiscal 2026 as "the most challenging year" Booz Allen has faced as a public company, citing turbulence in federal procurement and pressure on civil sector contracts.

The warning comes as the Pentagon accelerates its pivot from legacy manned platforms toward AI-driven autonomous systems, a shift that Booz Allen is capturing through its partnership with Anduril Industries and investments in cyber and defense technology. The outcome of the US-China AI competition will determine which nation controls the software-defined battlefields of the next decade.

Three Threat Vectors From Chinese AI

Rozanski identified three specific risk categories during his appearance. First, Chinese AI models may contain embedded vulnerabilities or backdoors that could compromise US systems that integrate them. Second, the software supply chain — where open-weight models from Chinese developers are increasingly used by Western companies — creates an attack surface that adversaries can exploit. Third, agentic AI systems capable of autonomous decision-making could be weaponized for cyber operations targeting critical infrastructure.

The Pentagon has responded by tightening export controls on advanced semiconductors and GPUs, restricting China's ability to train frontier models on high-end hardware. The US also designated Anthropic as a "supply chain risk" after a dispute over AI safety guardrails, illustrating the growing tension between commercial AI development and national security requirements.

Defense Tech Pivot Accelerates

Booz Allen is betting that its pivot toward autonomous systems will offset headwinds in its civil business. The company recently finalized a partnership with Anduril Industries to integrate command-and-control, cyber, and zero-trust capabilities onto Anduril's Menace tactical edge hardware. Anduril, which raised $5 billion in a Series H round at a $61 billion valuation in May 2026, is building Arsenal-1, a 5-million-square-meter manufacturing facility designed to produce tens of thousands of autonomous weapons annually.

Booz Allen also saw a nearly 90% increase in OTA proposal submissions and approximately 50% growth in OTA awards compared with the prior year, reflecting the Pentagon's shift toward faster, nontraditional procurement methods. The company's Vellox cyber platform and EdgeXtend tactical edge product line, alongside partnerships with NVIDIA and Amazon Web Services, form the technology backbone of its defense strategy.

The last time the US faced a comparable technology-driven national security challenge was during the Cold War's space race, when the Soviet Union's Sputnik launch triggered a surge in defense R&D spending that reshaped the US industrial base for decades. The current AI race carries similar structural implications, with defense budgets increasingly allocated to software-defined systems rather than traditional hardware platforms.

This article is for informational purposes only and does not constitute investment advice.