How AI Is Being Used to Wage War — the technologies behind it, and the companies building the tools

War is changing. Modern militaries no longer only fight with tanks and jets — they increasingly fight with software, sensors, and machine learning models. Below I explain the ways AI is being applied on the battlefield, the underlying technologies that power those capabilities, and the leading companies and contractors shaping “AI-enabled” warfare today. I keep this at a high, policy-and-technical level (no operational instructions), and cite recent industry reporting so you can follow up.

1) How AI is actually being used in modern conflict

1.1 Intelligence, Surveillance & Reconnaissance (ISR)
AI is used to ingest vast sensor streams (satellite imagery, drone video, signals intelligence), flag unusual activity, and automatically classify objects (vehicles, boats, installations) so human analysts can act faster. This reduces the time between detection and decision. Anduril Industries+1

1.2 Autonomous and semi-autonomous systems
From small loitering munitions and unmanned surface/subsurface vessels to autonomous land vehicles and loitering drones, AI gives machines navigation, target-recognition assistance, collision avoidance, and coordinated swarm behaviour. Several defense startups and prime contractors are building “families of autonomous systems” for surveillance, force protection, and strike roles. Reuters+1

1.3 Targeting, decision-support and command-and-control
AI-powered software helps fuse data, propose courses of action, run automated simulations (wargaming) and support commanders with prioritized recommendations. Products are being sold as “decision-support” or C2 (command & control) systems to accelerate planning and execution. Business Wire+1

1.4 Electronic warfare, cyber and information operations
Machine learning improves signal detection, jamming optimization, intrusion detection, and automated exploitation of network data. AI also amplifies misinformation and deepfake production/targeting — tools that can be weaponized to shape perceptions and morale. (High-level discussion; not operational details.)

1.5 Logistics, maintenance and force readiness
Predictive maintenance (ML predicting when vehicles/aircraft will fail), optimized logistics routes, and automated inventory/supply-chain systems keep forces supplied and resilient — an often-overlooked battlefield advantage.

1.6 Training and simulation
Generative models and large-scale distributed simulation let militaries run realistic wargames, model contingencies, and train personnel in virtual environments far faster and cheaper than before. Business Wire

2) The core technologies that make this possible

• Computer vision (CNNs, transformers for images): object detection and tracking in imagery and video.
• Natural language processing & LLMs: extracting meaning from reports, automating translation, generating summaries and even drafting operational briefs.
• Reinforcement learning & planning algorithms: used for autonomous navigation, multi-agent coordination and limited tactical decision-making.
• Sensor fusion & real-time analytics: combining radar, EO/IR, SIGINT, and other feeds into a single operational picture.
• Edge AI and tiny/embedded ML: running models on drones, boats, and field sensors where connection to the cloud is limited.
• High-performance computing (GPUs, TPUs) and simulation: training big models, running high-fidelity simulations, and digital twins. NVIDIA+1

These are powered by an ecosystem of sensors, communications (including resilient mesh networking), secure hardware (trusted execution), and cloud/edge infrastructure.

3) Who builds these systems today? (Companies and categories)

Note: many players are a mix of traditional mega-contractors, cloud/GPU providers, and agile startups. Below are representative companies repeatedly highlighted in recent reporting and government contracts.

Big defense primes (system integrators & platforms)

• Lockheed Martin, Northrop Grumman, Raytheon/RTX, BAE Systems, Thales, Leonardo, Rheinmetall — long-standing defence primes that integrate sensors, missiles, radars and increasingly add AI layers to their products. These firms appear across defense-industry rankings and are investing in robotics, autonomy and AI. Defense News

“AI-native” defense startups and integrators

• Anduril Industries — builds autonomy software (Lattice) and systems for border/force protection, counter-UAS and maritime autonomy; recently involved in undersea and surface autonomous systems partnerships. Anduril Industries+1
• Palantir Technologies — known for data-fusion and decision-support platforms (widely used for military analytics, C2 and simulation); expanding partnerships and enterprise licenses with services like the U.S. Marine Corps. DefenseScoop+1
• Shield AI, Rebellion Defense, Heron Systems and others — smaller startups specializing in autonomy, perception and specialized mission software (these companies often work with DoD programs or allied militaries). Global Venturing

Technology & infrastructure providers

• NVIDIA — GPUs and AI stacks are central to model training and inference; NVIDIA actively engages with defense customers and sponsors research sessions on generative AI for national security. Hardware from companies like NVIDIA is a critical enabler of defense ML. NVIDIA+1
• Cloud providers (Microsoft Azure, AWS, Google Cloud) — provide secure cloud enclaves, specialized “gov/military” cloud services and AI tooling for defense customers (often under compliance frameworks).
• Chip and component firms (Arm, Intel, etc.) — underpin edge compute and embedded systems used in vehicles and weapons.

Specialist simulation and systems companies

• Firms like Hadean (simulation), and other wargaming/simulation vendors, supply digital environments and distributed simulation software used for planning and AI-driven experimentation. Business Wire

4) Why this matters — risks, norms and legal issues

Speed and scale: AI compresses the observe-orient-decide-act loop. Decisions can be faster than human cognition if systems are automated — that improves responsiveness, but raises risks of misidentification, escalation, and loss of human judgement.

Autonomous weapons and human control: The debate about “meaningful human control” continues. Fully autonomous lethal systems (often discussed as “LAWS” — lethal autonomous weapons systems) are controversial; many states, NGOs and researchers call for strict limits or bans. International law (including IHL — laws of armed conflict) still applies, but legal and ethical responsibilities are complex when machines make consequential choices.

Vulnerabilities & adversarial attacks: ML models can be tricked (adversarial examples), attacked (data poisoning), or fooled by spoofed sensors. Dependence on AI creates new attack surfaces that adversaries can exploit.

Proliferation and asymmetric use: Cheaper autonomy and open-source AI lower the bar for non-state actors and smaller states to adopt sophisticated capabilities, changing deterrence and escalation dynamics.

Information warfare: AI makes deepfakes and targeted disinformation more scalable, threatening civilian populations and democratic resilience.

5) Governance, mitigation and paths forward

• Stronger norms & international agreements: multilateral efforts to limit certain autonomous weapons, set verification norms, and agree on human-in-loop requirements.
• Technical safety standards: robust testing, explainability, adversarial-robust models, and fail-safe/human override mechanisms.
• Export controls & supply-chain oversight: controls on advanced sensors, compute, and ML models to manage proliferation risks.
• Transparency and auditability: independent red-team testing, public reporting where possible, and capability audits for high-risk systems.
• Civilian resilience: investments in media literacy and detection tools to blunt AI-enabled information operations.

6) Quick reading / sources (recent reporting & examples)

• Anduril’s autonomy and Lattice platform (company overview). Anduril Industries
• Australia’s recent undersea autonomous drone procurement with Anduril (example of maritime autonomy adoption). Reuters
• Palantir enterprise licensing and Marine Corps acquisition of AI-powered systems. DefenseScoop+1
• Defense industry coverage of drones, AI and robotics reshaping top defense companies. Defense News
• NVIDIA sessions and reporting on generative AI and national security; role of GPUs in defense AI. NVIDIA+1

Closing — big picture in one line

AI magnifies both capabilities and risks on the battlefield: it can make forces faster, more informed and more efficient — but it also accelerates escalation, creates new vulnerabilities, and raises urgent ethical and governance questions that the international community still must resolve.