AI Arms Race: US, China & Europe’s Defense Strategies

War of Wits

Artificial intelligence is rapidly emerging as a cornerstone of military strategy across the United States, China, and Europe. Each power is investing heavily in AI-driven defense capabilities, but their approaches and priorities reveal as many differences as similarities in this new era of high-tech geopolitics.

The New Battlefield: AI as a Strategic Priority

In the past year, an AI arms race in the military arena has accelerated. All three regions – the U.S., China, and Europe – see AI as critical to future warfare and are ramping up investments accordingly.

Global venture funding for defense tech startups hit a record $49.1 billion in 2025, nearly doubling from the year before. This surge is driven in part by real-world validation: the use of drones and AI-enabled systems on the Ukrainian battlefield has demonstrated their effectiveness, fundamentally shifting how investors view defense.

Military budgets are reflecting this trend as well, prioritizing projects like autonomous drones, AI-enabled decision support, and “collaborative combat” platforms that team manned and unmanned systems.

The result is a global race to harness AI for strategic advantage. Yet the means and motives differ.

The U.S. leverages its vibrant private tech sector and open innovation ecosystem; China pursues a state-directed push for “intelligentized” warfare and self-reliance in AI; Europe, spurred by security threats, seeks to boost capability while aligning with ethical norms and multilateral cooperation.

Despite disparate methods, all acknowledge that mastering military AI will be pivotal to deterring adversaries and winning future conflicts.

United States: Private-Sector Innovation and Defense Contracts

The United States has placed AI at the heart of its defense modernization. In mid-2025, the Pentagon’s Chief Digital and AI Office awarded contracts of up to $200 million each to four leading AI firms – Anthropic, Google, OpenAI, and xAI – to accelerate adoption of cutting-edge “frontier” AI models.

This reflects the U.S. strategy of tapping private-sector innovation to maintain a technological edge. Agentic AI systems are being pursued to enhance battlefield planning, threat detection, and command decision-making, with the Department of Defense aiming to support troops and maintain a strategic advantage through AI.

American defense startups have accordingly seen a flood of capital. Venture funding for U.S. defense-tech companies nearly tripled to $14.2 billion in 2025, far outpacing Europe’s ~$2.5 billion.

Heavyweight investors are shedding previous taboos around defense and reframing it as backing technologies that safeguard democracies. The poster child is Anduril Industries, a California-based firm building AI-enabled military systems, which raised a massive $2.5 billion round in mid-2025, valuing it at $30+ billion.

Anduril has secured billions in U.S. government contracts, from counter-drone networks to surveillance towers. In March 2025, the Marine Corps awarded Anduril a $642 million deal to deploy a family of AI-enabled systems protecting bases from small drones. At the core of this solution is Anduril’s “Lattice” AI software, which fuses sensor data and provides autonomous threat detection and tracking across its network of towers, cameras, interceptors and drones.

The U.S. military’s embrace of such platforms underscores a “software-first” approach to warfighting technology, where adaptable AI and autonomy are just as important as the hardware.

Beyond startups, traditional defense players and the Pentagon’s own labs are deeply involved. DARPA projects in recent years showcased AI-piloted fighter jets and helicopters, proving that autonomous systems can handle complex tasks like dogfighting and flight without humans.

The Air Force has experimented with human–AI teaming for rapid targeting decisions – in one 2025 exercise, an AI “Smart System” suggested real-time courses of action along the “kill chain,” with human operators validating targets and ultimately making the kill decisions. The Defense Innovation Unit similarly launched Project “Thunderforge” with industry partners to integrate intel feeds and sensor data into AI-driven operational plans, again keeping humans “in the loop” for final calls.

American policymakers acknowledge the risks alongside the rewards of military AI. Experts have cautioned that increasingly autonomous weapons could accelerate conflicts or trigger unintended escalation, so the DoD has adopted ethical AI guidelines and testing protocols.

The prevailing U.S. view is that AI should augment human decision-makers, not replace them, ensuring commanders retain control over the use of force. This careful stance, however, goes hand-in-hand with a resolve that the U.S. must lead in military AI lest rivals gain an upper hand. As such, from an investor lens, the U.S. defense sector now offers enormous opportunity – but also high expectations to deliver AI innovations that can be fielded at scale and with reliability.

See my previous work on AI in military applications:

The Geopolitics of AI: Large Language Models

Geopolitical Power Play: AI Unleashed

Battle of the Bytes: Too Little Too Late for China in AI US Face-Off

China: State-Driven “Intelligentization” and Algorithmic Sovereignty

China likewise sees AI as transformative for its military, but its approach is centrally orchestrated and shrouded in secrecy. The People’s Liberation Army (PLA) has officially embraced the doctrine of “intelligentized warfare,” viewing AI, autonomy, and data as the keys to surpassing conventional military strength.

In practice, Beijing is methodically integrating AI across its defense industry under a policy of “military–civil fusion.” A Reuters investigation in late 2025 revealed a systematic effort by the PLA and China’s defense giants to harness AI for battlefield advantage, focusing on capabilities like autonomous target recognition and real-time decision support. Many of these efforts mirror U.S. initiatives in concept, underlining that China is determined not to fall behind.

However, China’s strategy also reflects its unique constraints and goals. Faced with U.S. export controls on high-end chips and tech, the PLA has been **pushing for “algorithmic sovereignty” – reducing dependence on Western technology and using Chinese alternatives wherever possible.

In 2025, Chinese military procurement shows a growing preference for domestic AI models and hardware. One standout is “DeepSeek,” an AI model developed in China that has become the pride of China’s tech sector and a favorite across PLA research projects. In fact, dozens of PLA tenders in 2025 explicitly requested DeepSeek-based solutions, while mentions of Alibaba’s competing model were scarce.

This dominance of DeepSeek underscores Beijing’s drive to build homegrown AI champions under its control. At the same time, Huawei’s AI chips are increasingly appearing in military tech, as Chinese firms respond to government pressure to use indigenous components. (Notably, the PLA still covets advanced NVIDIA chips that are now banned for export; patents filed in 2023 suggest some units experimented with NVIDIA’s A100 and H100 processors, possibly stockpiled before the ban.)

Chinese officials frequently tout breakthroughs as evidence that AI will help close the gap with the U.S. In early 2025, for example, state-owned Norinco unveiled the P60 autonomous combat vehicle, a robotic tank-like platform that can drive at 50 km/h and perform combat support missions without a human crew.

The P60 is powered by the DeepSeek AI engine, and Chinese media lauded it as a sign that China is catching up in the arms race. PLA research institutes have filed patents on AI-powered command systems that can analyze sensor data from satellites and drones to aid battlefield decisions. One paper claimed a DeepSeek-driven system could simulate 10,000 battlefield scenarios in 48 seconds, exploring different terrains and force deployments to recommend optimal strategies.

In another project, scientists at Beihang University leveraged DeepSeek to improve drone swarm decision-making against “low, slow, small” targets (like swarms of hostile drones or light aircraft). These snippets reveal a PLA intent on using AI to outthink, outpace, and outmaneuver – whether through rapid war-gaming simulations or decentralized drone swarms that can adapt in real time.

On the hardware side, China is experimenting with exactly the kind of systems AI enables. The PLA has deployed…