AI in Defense: How Artificial Intelligence is reshaping Modern Warfare and Military Strategy

The Speed Imperative in Modern Warfare

Military history has always rewarded the force that can act faster than its adversary. What has changed in the 21st century is the rate at which that speed advantage has accelerated. Modern battlefields generate sensor data, imagery, signals intelligence, and communications intercepts at a volume that no human analyst team can process in real time. Radar systems track hundreds of objects simultaneously. Satellite imagery refreshes every few minutes over a theatre of operations. Electronic warfare signals fill the spectrum. The commander who can synthesise this data faster than the opponent and convert it into action holds a decisive edge.

This is precisely the gap that artificial intelligence is designed to close. AI systems can ingest data from multiple sensor types, classify threats, assign priority scores to targets, and present decision recommendations to human commanders within seconds of receipt. Tasks that previously required hours of analysis by specialised intelligence teams can be completed in minutes or less. The compression of what military planners call the “kill chain” from detection to engagement is not merely a matter of efficiency. It is a fundamental change in the nature of military advantage.

The Ukraine-Russia War as a Live Proving Ground

No conflict has done more to demonstrate, test, and accelerate AI’s role in military operations than the war in Ukraine, which entered its fourth year in 2025. What began in 2022 as a conventional war with substantial use of off-the-shelf commercial drones evolved rapidly into a full-scale technological arms race in which AI integration became the defining competitive axis.

The evolution moved in distinct phases. Early in the conflict, both sides used drones primarily for reconnaissance and direct strike missions, but these were largely manually controlled or semi-autonomous. As Russian electronic warfare systems improved, causing GPS disruption and signal jamming, Ukrainian operators found that purely GPS-guided drones became unreliable. The response was to embed AI-based visual navigation and target recognition into the drones themselves, making them capable of autonomous flight and targeting even when communication links were severed.

The war also demonstrated that AI warfare is not the exclusive property of the richest military powers. Ukraine, operating under severe resource constraints and supply chain pressures, built a significant AI-enabled drone capability through a combination of commercial off-the-shelf components, open-source software, and rapid field iteration. This finding has profound implications: the barriers to entry for AI-enabled military capability are lower than many had assumed, and the pace of innovation on the battlefield can outrun the pace of formal procurement cycles.

Russia drew its own lessons. In December 2024, the Russian military announced the establishment of a dedicated unmanned systems branch to better integrate autonomous and robotic technologies across service branches. In August 2024, Russian officials announced a new 10-year defence strategy with a dedicated focus on artificial intelligence.

Intelligence, Surveillance, and Reconnaissance

The most mature and widely deployed military AI applications are in the domain of intelligence, surveillance, and reconnaissance, commonly referred to as ISR. AI systems applied to ISR tasks can process satellite imagery to detect changes in terrain, identify military vehicles and formations, track shipping movements, monitor border crossings, and flag anomalies across vast geographic areas continuously. Human analysts reviewing the same imagery would require days or weeks to cover what an AI pipeline can process in hours.

Computer vision models, trained on millions of labelled examples of military hardware, can now identify specific weapon systems, estimate unit strength, and distinguish military from civilian infrastructure with a level of accuracy that approaches or, in controlled conditions, exceeds human performance. These capabilities are directly what Project Maven began building in 2017 when the Pentagon contracted Google to develop AI for drone video analysis, and they are what the Maven Smart System has expanded into a comprehensive multi-domain targeting and intelligence platform by 2026.

Autonomous and Semi-Autonomous Weapons Systems

Autonomous weapons systems span a wide spectrum. At the less autonomous end are systems like precision-guided munitions that follow a pre-programmed flight path or loitering munitions that can be directed by a human operator to a target after launch. At the more autonomous end are systems designed to identify and engage targets within a defined zone without real-time human involvement once deployed. In between are the semi-autonomous systems that are currently most prevalent: drones that can navigate autonomously, avoid obstacles, and conduct surveillance independently but require a human command to engage a target.

Cyber Warfare and Signals Intelligence

In the cyber domain, AI is transforming both offensive and defensive operations. On the defensive side, AI-based systems can monitor network traffic across thousands of endpoints simultaneously, identify anomalous patterns that suggest intrusion or exfiltration, and respond to threats in milliseconds, far faster than any human security team. On the offensive side, AI can assist in identifying vulnerabilities in adversary networks, generate novel attack vectors, and automate aspects of cyber operations that previously required significant human expertise.

Signals intelligence, the interception and analysis of electronic communications and emissions, is similarly transformed by AI. Language models can process intercepted communications at scale, translate and summarise in real time, and flag items of intelligence value for human review. The volume of signals that modern military forces generate far exceeds the capacity of human analysts to process manually, making AI an operational necessity rather than a productivity enhancement in this domain.

Logistics, Predictive Maintenance, and Force Management

Some of the most immediately practical applications of AI in defence are unglamorous but operationally significant. Predictive maintenance systems use machine learning to analyse sensor data from aircraft engines, vehicle drivetrains, and ship propulsion systems to predict component failures before they occur, reducing unplanned downtime and increasing the availability of platforms for operations. Supply chain optimisation AI can model logistics requirements for complex multi-domain operations, identifying bottlenecks and pre-positioning supplies in ways that reduce the logistical tail required to sustain operations in the field.

The US Department of Defense explicitly listed financial systems and human resources management among the priorities for its AI Rapid Capabilities Cell, established in December 2024, reflecting recognition that the institutional overhead of running a large military is itself an area where AI can generate significant efficiency gains that free resources for frontline operations.

Project Maven: From Drone Imagery to the Centre of US Military Power

Project Maven traces its origins to 2017, when the Pentagon contracted Google to develop AI that could process drone surveillance footage, label objects of interest, and help analysts manage the flood of video data being generated over active conflict zones. The initial contract was relatively modest, valued at around $9 million, but the implications were significant. For the first time, the Pentagon was systematically applying commercial machine learning capabilities to an operational military intelligence problem.

Google withdrew from the project in 2018 following internal employee protests about the ethics of contributing to weapons systems. Palantir Technologies subsequently took over the programme and expanded it substantially. By 2024, the Pentagon had awarded Palantir a five-year Maven contract worth up to $480 million, followed by a roughly $100 million expansion later that year. In May 2025, the contract ceiling was raised to $1.3 billion through 2029. In March 2026, Deputy Secretary of Defense Steve Feinberg signed a memorandum designating the Maven Smart System as an official Pentagon programme of record, embedding it as a permanent part of the US military’s institutional infrastructure.

CJADC2 and the Vision of Connected Combat

Beyond targeting, the broader US military AI strategy centres on the concept of Combined Joint All-Domain Command and Control, or CJADC2. The goal is to connect sensors, weapons platforms, and commanders across all military domains, including air, land, sea, space, and cyber, in real time through a common data architecture. AI serves as the connective tissue of this vision, processing and routing information between systems that were originally designed to operate independently.

The Maven Smart System is explicitly positioned as the core software infrastructure for CJADC2. The March 2026 Feinberg memorandum evaluated the option of placing the Maven programme of record under a potential CJADC2 Programme Office as its permanent home, underscoring how central Palantir’s platform has become to the entire joint force connectivity concept.

The AI Rapid Capabilities Cell

In December 2024, the Pentagon announced the creation of the Artificial Intelligence Rapid Capabilities Cell, or AI RCC, overseen by the Chief Digital and Artificial Intelligence Office and partnering with the Defence Innovation Unit. The AI RCC’s priorities include generative AI applications for command and control, autonomous drone development, intelligence processing, weapons testing evaluation, and enterprise management systems. The Pentagon allocated $100 million toward the AI RCC in fiscal years 2024 and 2025. The cell reflects a deliberate effort to move faster than the traditional defence acquisition process allows, recognising that AI capabilities developed in the commercial sector are evolving faster than procurement cycles can respond.

From Mechanisation to Intelligentisation

China’s military modernisation has passed through two conceptual phases in recent decades: mechanisation, which refers to equipping the People’s Liberation Army with modern conventional weapons, and informatisation, which refers to digitalising and networking military systems. The PLA is now pursuing a third phase, which Chinese strategists call “intelligentisation,” meaning the deep integration of artificial intelligence, autonomous systems, and advanced data networks throughout all aspects of military operations. The 2019 Chinese defence white paper identified this transition as a core strategic goal, arguing that “intelligent warfare” was becoming the dominant form of future armed conflict.

AI was listed as the first technology priority in China’s five-year economic plan for 2021 to 2026, the only technology to receive that designation. President Xi Jinping has placed PLA modernisation at the core of China’s 15th Five-Year Plan covering 2026 to 2030, with technological self-reliance in AI, advanced manufacturing, and defence a central pillar. The concept of “algorithmic sovereignty,” meaning China’s goal of reducing dependence on Western AI technology while strengthening domestic control over critical digital infrastructure, guides procurement decisions throughout the defence sector.

PLA AI Procurement and DeepSeek

Analysis of hundreds of procurement tenders from the PLA Procurement Network found that PLA-affiliated entities had been increasing their use of domestically made AI hardware, including Huawei chips, and that DeepSeek AI models were referenced in a dozen PLA procurement tenders filed in 2025, with new military applications appearing regularly. DeepSeek-related procurement notices from PLA entities accelerated throughout 2025. The shift reflects Beijing’s push for domestic technology adoption in strategically sensitive systems, reducing the risk of supply chain disruption from Western export controls.

China’s Drone Manufacturing Advantage

China holds a structural advantage in the drone domain that extends beyond military procurement. China is the world’s leading exporter of commercial drones and has a mature civilian drone manufacturing ecosystem that feeds directly into military capability development. The commercial drone industry provides a base of component suppliers, manufacturing capacity, and engineering talent that military programmes can draw on rapidly. In November 2024, the South China Morning Post reported that Chinese defence scientists at the Chengdu Aircraft Design Institute had developed an experimental generative AI system capable of commanding drones equipped with electronic warfare weapons.

The United Kingdom’s Strategic Defence Review

On June 2, 2025, the UK government published its Strategic Defence Review, explicitly designed to modernise Britain’s armed forces for the AI age, drawing lessons from the Ukraine-Russia conflict. The review called for a shift toward greater use of autonomy and AI within the UK’s conventional forces, acknowledging that the British military faces significant challenges including a reliance on outdated hardware, insufficient ammunition and equipment, and slow procurement processes. The UK Ministry of Defence faces an additional funding requirement of £28 billion over the next four years as the government targets defence spending at 3 percent of GDP by 2029.

The European Union

The European Union’s Security Action for Europe (SAFE) programme, formally adopted on 27 May 2025, provides up to €150 billion (approximately $170 billion) in long-term low-cost loans to EU member states for joint defence procurement. SAFE is the first pillar of the broader ReArm Europe/Readiness 2030 package, which aims to leverage over €800 billion in total EU defence spending. At the regulatory level, the EU AI Act, the world’s first comprehensive AI regulation, establishes requirements for AI systems used in critical infrastructure, though military systems designed solely for military purposes are explicitly excluded from its direct scope. The EU’s broader position is that international humanitarian law applies to all weapons systems, and that EU member states must ensure AI-enabled weapons comply with existing legal frameworks even in the absence of specific military AI legislation.

The Budget Foundation

India’s defence budget for FY2025-26 stands at Rs 6.81 lakh crore, representing a 2.6-fold increase from Rs 2.53 lakh crore in 2013-14, according to Ministry of Defence data published in March 2025. India has also unveiled a record defence budget of approximately $87 billion for fiscal year 2026, a 15 percent increase, signalling accelerating investment in multi-layered air defence systems and indigenous military technology. The Government of India has earmarked Rs 100 crore annually specifically for AI in military projects. The FY2024-25 defence budget increased the allocation to DRDO to Rs 23,855 crore, with an explicit aim to enhance fundamental research and promote private sector involvement in defence production.

The iDEX Framework and ADITI Scheme

The Innovations for Defence Excellence programme, known as iDEX, was launched in 2018 as the central framework for engaging the private sector, including startups and MSMEs, in defence innovation. As of February 2025, iDEX had issued 549 problem statements, engaged 619 startups and MSMEs, and awarded 430 contracts. A total of 26 products developed under iDEX had received procurement orders totalling over Rs 1,000 crore, and Acceptance of Necessity or Requests for Proposals worth Rs 2,380 crore had been issued for 37 products.

The ADITI scheme, standing for Acing Development of Innovative Technologies with iDEX, was launched on 4 March 2024 during DefConnect 2024 by Defence Minister Rajnath Singh. With a total corpus of Rs 750 crore covering FY2023-24 to FY2025-26, the scheme aims to accelerate the development of approximately 30 deep-tech critical and strategic technologies. These technologies span AI, quantum encryption, cyber technology, satellite communications, semiconductors, autonomous weapons systems, and underwater surveillance. Grants of up to Rs 25 crore per project are available to startups and MSMEs, covering 50 percent of product development costs. The iDEX outlay for FY2025-26 is Rs 449.62 crore, covering ADITI and complementary programmes.

India’s Dual Motivation: Capability and Sovereignty

India’s defence AI push is driven by two distinct motivations that reinforce each other. The first is military capability: recent security tensions with Pakistan and the ongoing strategic competition with China on the northern border have created a clear operational demand for AI-enabled ISR, autonomous systems, and cyber capabilities. The second is technological sovereignty: India’s historical dependence on imported defence equipment from the US, Russia, and others has been identified as a strategic vulnerability, and Atmanirbhar Bharat (self-reliant India) policies are driving a determined push to build domestic capabilities in technologies identified as critical for future warfare. India, motivated by concerns about strategic dependence and security tensions, has been leveraging its growing commercial technology ecosystem to strengthen sovereign dual-use capabilities and homegrown defence AI.

The Accountability Gap

The most fundamental legal challenge posed by AI-enabled weapons systems is the accountability gap: when an autonomous or semi-autonomous system takes an action that causes civilian casualties, who is legally responsible? International humanitarian law has historically assigned responsibility to commanders and states, but it presupposes that a human being made the decision to use lethal force. When that decision is made or substantially shaped by an algorithm, the linearity of responsibility that international law requires becomes difficult to establish.

The UN Secretary-General’s report of June 2025, compiled following General Assembly resolution 79/239 of December 2024 (which affirmed that international law applies throughout the life cycle of military AI), noted that AI systems may “obfuscate the linearity of this process,” making it harder to attribute responsibility for specific acts of violence. António Costa, President of the European Council, stated that “the development of lethal autonomous weapons systems threatens to remove human accountability from decisions of life and death” and warned that “the risks are real: miscalculation, escalation, and proliferation.”

Meaningful Human Control

The principle that has emerged as the central normative standard in international discussions of military AI is “meaningful human control,” meaning that a human being must remain in the decision-making loop for lethal uses of force. The Pentagon’s own policy framework focuses on this principle, and NATO reaffirms the full validity of international humanitarian law for all AI weapons while emphasising the maintenance of human responsibility for the decision to use force. The challenge is that “meaningful human control” has not been precisely defined in binding international legal terms, and the speed at which AI-enabled systems operate raises a practical question about whether human review within a compressed kill chain constitutes genuine control or a formality.

The International Governance Vacuum

Despite these concerns, neither the EU nor NATO has yet adopted specific binding legislation that explicitly regulates or restricts military AI, relying instead on the application of existing international law and political declarations. The UN General Assembly’s December 2024 resolution affirming the applicability of international law to military AI life cycles is an important normative signal, but it is not binding on states. The UN Secretary-General has set out four urgent priorities: ensuring human control over the use of force and banning fully autonomous lethal weapons; establishing red lines for AI in nuclear command systems; addressing AI’s role in disinformation and election interference; and ensuring that AI governance decisions are made inclusively rather than by a small number of powerful states.

• 2017
  Project Maven launched

  Pentagon contracts Google to develop AI for processing drone surveillance footage. Google withdraws in 2018 following internal protests. Palantir takes over and expands the programme substantially.

• 2018
  India’s iDEX launched

  India launches the Innovations for Defence Excellence framework to engage startups and MSMEs in defence technology development, with budgetary support and a structured pathway from prototype to procurement.

• 2019
  China’s defence white paper on intelligentisation

  PLA’s modernisation roadmap explicitly identifies AI, quantum computing, and autonomous systems as the core of future military capability under the concept of “intelligent warfare.”

• Feb 2024
  Ukraine establishes Unmanned Systems Forces

  President Zelensky signs a decree creating a dedicated military branch for unmanned systems, recognising drones as a core warfighting domain rather than a supplementary capability.

• Mar 2024
  India’s ADITI scheme launched

  Defence Minister Rajnath Singh launches ADITI at DefConnect 2024, providing grants of up to Rs 25 crore per project for 30 deep-tech defence technologies including AI, quantum, and autonomous systems.

• Dec 2024
  Pentagon’s AI Rapid Capabilities Cell

  The DoD establishes the AI RCC with $100 million in funding for FY2024-25, focused on generative AI for command and control, autonomous drones, and intelligence applications.

• Dec 2024
  UN General Assembly resolution 79/239

  The General Assembly affirms that international humanitarian law applies throughout the life cycle of military AI systems, establishing a normative baseline even in the absence of binding treaty obligations.

• Jun 2025
  UK Strategic Defence Review

  The UK government publishes its SDR, calling for a shift toward greater use of autonomy and AI across conventional forces, drawing explicit lessons from the Ukraine-Russia battlefield and committing to 3 percent of GDP on defence by 2029.

• May 2025
  Maven contract ceiling raised to $1.3 billion

  The Pentagon increases the Maven Smart System contract ceiling through 2029, with over 20,000 active users and integration across 35 military tools anticipated ahead of the system’s designation as an official programme of record.

• Mar 2026
  Maven designated official Pentagon programme of record

  Deputy Secretary of Defense Steve Feinberg’s memorandum institutionalises Maven as the Pentagon’s core AI decision-making and targeting platform, mandating integration across all US military branches by September 2026.

Market Trajectory and Investment Outlook

The global AI in defence and security market was valued at $12.53 billion in 2024 and is projected to reach $14.15 billion in 2025, growing at a compound annual growth rate of 12.9 percent. By 2029, the market is expected to reach $22.75 billion. Separate market analyses project a faster trajectory for the AI in defence segment specifically, estimating it at $8.5 billion in 2026 and reaching $32.8 billion by 2031 at a CAGR of 30.1 percent, with the higher growth reflecting the anticipated acceleration of autonomous weapons procurement and the broader institutionalisation of AI across military supply chains and command structures.

North America holds the largest share of the current market at approximately 36 percent in 2025, driven by US defence spending levels and the maturity of its commercial AI sector. The software component of military AI is expected to hold the largest share through 2035, reflecting the primacy of platforms like the Maven Smart System over hardware in shaping military AI capability. The cybersecurity application segment is forecast to grow at the fastest rate, reflecting the increasing centrality of AI to both offensive and defensive cyber operations as state-sponsored cyber conflict intensifies.

The Three Risks That Governance Must Address

The acceleration of military AI investment across the major powers creates three interrelated risks that existing governance frameworks have not yet adequately addressed. The first is the risk of miscalculation: as AI systems compress decision timelines, the time available for diplomatic intervention or de-escalation in a crisis shortens. An AI-enabled engagement that a human commander might have paused to reconsider can be completed before the political level even becomes aware of it.

The second risk is proliferation. The barriers to entry for AI-enabled military capability are lower than those for conventional advanced weapons. Ukraine demonstrated that a middle-income country under active attack can develop operationally significant AI drone capabilities within months using commercial components and open-source software. This means that the exclusive club of AI-capable military powers will expand rapidly, including to non-state actors and groups that have no interest in observing the norms of international humanitarian law.

The third risk is escalation in nuclear domains. As AI systems are considered for integration into early warning, nuclear command and control, and strategic deterrence systems, the possibility of an AI system generating a false positive alert of incoming nuclear attack, or of adversaries misinterpreting AI-driven military actions as preparation for a nuclear strike, creates pathways to catastrophic miscalculation that purely human-controlled systems are less likely to trigger at the same speed.

Disclaimer: This article is intended solely for educational and informational purposes. It does not constitute investment advice, financial advice, or a recommendation to buy, sell, or hold any security or financial instrument. All market figures, programme data, budget allocations, and policy details are accurate to the best of the author’s knowledge as of the date of publication (June 9, 2026), and are drawn from publicly available official and market research sources. Defence procurement data, market forecasts, and operational details of AI systems may change rapidly. Readers seeking to make decisions based on information in this article should consult primary official sources and independent experts in the relevant domain. The publisher and author accept no liability for decisions made in reliance on information contained in this article.