AI-Powered Warfare Systems: Where Is the World Heading? The 2025–2030 Global Competition Map

The Starting Gun Has Already Fired

Somewhere over the Libyan desert in 2020, a quadcopter-sized loitering munition made a decision. According to a United Nations expert panel report filed as document S/2021/229, the drone — a Turkish-made STM KARGU-2 — may have engaged a target without a human explicitly authorizing each strike. The panel’s language was careful, the evidence contested, but the implication was unmistakable: autonomous lethal decision-making had crossed from laboratory conjecture into the fog of a real war.

Five years later, that moment looks less like an aberration and more like a prologue. From the Pentagon’s Replicator program to the People’s Liberation Army’s intelligentization doctrine, from NATO’s revised AI strategy to ASELSAN’s debut of an autonomous low-altitude air defense system at IDEF 2025, the global defense establishment has moved from asking whether artificial intelligence will reshape warfare to arguing furiously about how fast — and on whose terms.

Follow the Money: Defense Budgets in the AI Era

The Stockholm International Peace Research Institute’s 2025 Yearbook places US defense spending at $997 billion for fiscal year 2024. China’s officially declared military budget stands at $314 billion, but SIPRI and independent analysts consistently estimate the real figure at somewhere between $330 billion and $450 billion once defense-industrial subsidies and off-budget procurement are included. Even at the high end of that range, the United States outspends China by roughly 2.2 to 1.

Those headline numbers, however, obscure a more complicated picture. China’s defense-industrial base has become dramatically more cost-efficient over the past decade. A combat drone that costs the US military $15 million to field can, in several documented cases, be replicated by Chinese state manufacturers at a fraction of that price. When the Pentagon’s own strategists talk about “unfavorable cost-exchange ratios,” they are admitting that raw dollar dominance does not automatically translate into systems-level advantage.

China’s Three-Phase Blueprint: Intelligentization by 2049

China’s PLA has articulated a three-phase modernization arc: mechanization (largely complete), informatization (networked information systems across the force, targeted by 2027), and intelligentization (embedding AI into every layer of command, sensing, and weapons employment, with major milestones by 2035 and full-spectrum realization by 2049). This roadmap was detailed in testimony to the US-China Economic and Security Review Commission by CNAS analyst Jacob Stokes in 2024, corroborated by CRS IF11719 and RAND research.

The Center for Security and Emerging Technology (CSET) provided the most granular public analysis of actual PLA procurement in its report Harnessed Lightning, which examined 343 verified PLA contracts to map AI investment. Seven distinct priority areas emerged: autonomous vehicles, ISR enhancement, predictive maintenance and logistics, information and electronic warfare, simulation and synthetic training, command and control decision support, and automatic target recognition (ATR).

That last category — ATR — is the one that keeps Western arms-control lawyers awake at night. ATR combined with autonomous strike authority is the technical prerequisite for a fully autonomous weapons system. China has not officially declared any such capability, but the procurement trail suggests the engineering work is well underway.

Drones by the Hundreds: China’s Export Machine

China is the world’s largest military drone exporter. SIPRI data covering 2013 to 2023 documents the delivery of more than 200 combat-capable drones to 17 countries, spanning the Middle East, Africa, Central Asia, and parts of Southeast Asia. The Wing Loong and CH-series platforms have been used in documented strikes in Libya, Yemen, Ethiopia, and Nigeria.

The next generation is more concerning. China’s AVIC unveiled the FH-97A — a stealthy, jet-powered unmanned combat air vehicle designed as a “loyal wingman,” flying in coordinated swarms alongside crewed fighters. The FH-97A made its public debut at Airshow China 2022, signaling that China has moved beyond simple armed drones into collaborative autonomous combat systems.

America’s Replicator: Quantity as a Quality

The US military’s official response to China’s drone-mass strategy is Replicator. Announced by Deputy Secretary of Defense Kathleen Hicks in August 2023, the program’s central logic is straightforward: counter China’s numerical advantage with thousands of cheap, expendable autonomous platforms that impose impossible intercept demands on an adversary. The full designation — ADA2, for All-Domain Attritable Autonomous — signals the ambition.

By late 2025, Replicator had been restructured and absorbed into the Defense Autonomous Working Group (DAWG), according to CRS Issue Brief IF12611. The program did not die — it evolved, with its core concepts distributed across the Air Force’s Collaborative Combat Aircraft effort and the Navy’s Autonomous Underwater Vehicle expansion. The underlying strategic logic remains unchallenged: in war-game analysis, the side with attritable autonomous systems was able to sustain offensive pressure significantly longer than the side relying entirely on crewed platforms.

NATO’s AI Strategy: Unity on Paper, Fragmentation in Practice

On July 10, 2024, NATO adopted a revised artificial intelligence strategy with four objectives: promote responsible AI use, accelerate AI adoption to enhance interoperability, protect AI systems and manage risks, and counter adversarial AI capabilities. A January 2025 EPRS analysis assessed mixed progress: strong record on responsible use principles, patchier record on interoperability.

The core problem is that NATO’s 32 members are acquiring AI-enabled military capabilities through 32 largely separate procurement systems with varying technical standards, varying data-sharing agreements, and varying appetites for risk. Whether AI systems from different allies can meaningfully fuse their outputs in real-time combat operations remains largely unanswered. On countering adversarial AI, the strategy names the problem clearly; it does not yet solve it.

Turkey’s Autonomous Arsenal: From Libya to NATO’s Frontline

Among NATO’s 32 members, Turkey occupies a peculiar position: it is the only alliance member that has deployed autonomous or near-autonomous lethal systems in actual combat — and then exported them to at least ten other countries across three continents.

The STM KARGU-2 is the system that put Turkey on the map. UN Expert Panel report S/2021/229 documented its deployment in Libya in 2020, with the panel noting it may have engaged targets without persistent human command link — among the first such documented instances in a live conflict. Army Recognition confirmed in April 2026 that KARGU has been exported to more than ten countries across three continents, including at least one European NATO member.

STM’s second autonomous loitering munition, ALPAGU, was formally accepted into the Turkish Armed Forces in October 2025. Smaller and faster than KARGU, it is designed for more contested environments. ASELSAN, meanwhile, unveiled the GÜERZ autonomous low-altitude air defense system at IDEF 2025 — a radar-and-effector package designed to autonomously detect, track, and engage low-slow-small aerial threats.

Taken together, KARGU, ALPAGU, and GÜERZ represent a coherent autonomous-systems architecture — offense and defense, loitering and fixed-site — built entirely within Turkey’s domestic defense industry over less than a decade.

The Governance Gap: Killing Faster Than the Rules Can Keep Up

The Group of Governmental Experts (GGE) on Lethal Autonomous Weapons Systems, operating under the UN Convention on Certain Conventional Weapons (CCW), has been meeting since 2014. After more than ten years of deliberation, it has produced non-binding guiding principles but no treaty, binding protocol, or even an agreed definition of what an autonomous weapons system actually is.

SIPRI’s 2025 “Towards Multilateral Policy” report noted that the GGE’s mandate expires in 2026, and the 7th CCW Review Conference — scheduled for November 2026 — will need to decide whether to extend, strengthen, or effectively abandon the process. A separate SIPRI analysis (DOI 10.55163/YQBY3151) highlights a critical policy gap: existing frameworks were largely designed for AWS but leave AI Decision Support Systems (AI-DSS) largely unregulated, even as they proliferate across nearly every advanced military — systems that shape human targeting decisions at speeds that make oversight nominal rather than real.

Algorithmic Bias and the Laws of War

SIPRI’s August 2025 report examined the intersection of algorithmic bias and International Humanitarian Law (IHL) compliance. Military AI systems trained on historical imagery datasets inherit whatever biases exist in that data. Systems tested in one geographic or climatic context have shown significantly degraded accuracy when deployed in another — potentially misclassifying civilian vehicles as military ones. IHL requires attacks be directed only at lawful targets, that civilian harm be proportionate, and that all feasible precautions be taken. An autonomous system that misclassifies a civilian target because of training data bias may be committing an IHL violation with no human conventionally accountable.

The 2025–2030 Horizon: Key Milestones

• 2025 (Ongoing): US Air Force Collaborative Combat Aircraft program advances; Turkey’s ALPAGU enters service; ASELSAN GÜERZ moves toward procurement evaluation.
• 2026 Q1–Q2: NATO member states submit AI Strategy progress reports; assessment of implementation gaps.
• 2026 November: 7th CCW Review Conference — most consequential international moment for autonomous weapons governance in five years. Outcome determines whether binding legal instruments remain politically viable.
• 2027: China’s formal PLA centenary milestone — the informatization target date. Western intelligence will watch whether PLA exercises reflect genuine AI integration at scale.
• 2028–2029: US Collaborative Combat Aircraft enters limited production; AI-ISR systems redefine reconnaissance aircraft roles.
• 2029–2030: China’s 2035 intelligentization milestone enters its preparatory phase; NATO releases next-generation AI interoperability framework.

The Escalation Problem at Machine Speed

The most profound strategic question raised by military AI is what happens when systems on opposing sides interact without adequate time for human deliberation. The 1983 Soviet nuclear early-warning incident — in which Lieutenant Colonel Stanislav Petrov overrode a false positive alert for incoming US missiles — is the most famous historical near-miss. Petrov had minutes to think. Autonomous systems will have milliseconds. Neither the US-China strategic dialogue nor deconfliction channels have produced any agreed protocols for autonomous system behavior in gray-zone incidents.

Conclusion: The Race Is On, the Rulebook Is Thin

The 2025–2030 period will almost certainly be the decisive window in which military AI transitions from an emerging capability to an established fact of modern warfare. The US will spend more than any other actor. China will deploy at greater scale than Western analysts once predicted. Turkey and a small number of other second-tier military powers will continue to punch above their weight through domestic development and aggressive export strategies.

What will not keep pace, on current trajectory, is the governance architecture designed to ensure these systems are used responsibly, that their failures can be attributed and corrected, and that their deployment does not create escalation pathways that no human intended and no human can reverse. The Libya desert in 2020 was the opening chapter. The chapters ahead will be written at machine speed.

Sources

• SIPRI Yearbook 2025 — World military expenditure data
• Jacob Stokes, CNAS — Testimony to USCC, Congressional hearing on PLA AI, 2024
• CSET, Harnessed Lightning: How the Chinese Military is Adopting Artificial Intelligence
• SIPRI Arms Transfers Database, 2013–2023 — China military drone export data
• Congressional Research Service, IF12611 — Replicator/ADA2 and DAWG restructuring, 2025
• NATO, Revised Strategy on Artificial Intelligence, July 10, 2024
• EPRS, Analysis of NATO AI Strategy implementation, January 2025, PE 769580
• SIPRI, Towards Multilateral Policy on Autonomous Weapons, 2025
• SIPRI, AI-DSS and policy gaps, June 2025, DOI 10.55163/YQBY3151
• SIPRI, Algorithmic bias and IHL compliance, August 2025
• UN Security Council Panel of Experts, S/2021/229 — Libya, KARGU-2
• Army Recognition, KARGU export to European NATO member, April 2026
• STM, ALPAGU acceptance into TSK, October 2025
• ASELSAN, GÜERZ presentation, IDEF 2025