India’s defence AI push has moved from 75 pilot projects to 129 approved programs, a real combat test in Operation Sindoor, and a nationwide Smart Border rollout in 2026.
In mid-2025, KPMG in India published a piece arguing that leadership in AI-enabled defence “would not be determined by intent, but by speed, scale, and strategic clarity.” At the time, that was a forward-looking claim, built on roughly 75 AI pilot projects across the armed forces, a newly funded Defence AI Project Agency, and early border-surveillance deployments.
A year on, the claim has been tested — literally. Here’s what’s changed in India’s AI-in-defence story since then, and where it stands heading into the second half of 2026.
From Pilots to Proof: Operation Sindoor
The single biggest shift since the original 2025 piece is that India’s AI-enabled defence systems have now been used in an actual conflict, not just exercises. Operation Sindoor, launched on 7–8 May 2025 in response to the Pahalgam terror attack, became India’s first large-scale non-contact war and a real-world stress test for AI-enabled systems.
Akashteer, India’s AI-powered autonomous air defence system, carried out real-time target interception during the operation. The Navy’s MARCOS commandos operated AI-enabled Palm 400 kamikaze drones, while the broader engagement involved hundreds of drones per day on both sides — a scale that validated, in practice, the “speed and decision superiority” argument that was previously theoretical.
In the year since, that experience has directly accelerated indigenous drone and counter-drone development, and drone training has been institutionalized at 19 training establishments, including the Indian Military Academy and the Officers Training Academy. Defence Minister Rajnath Singh has publicly credited Operation Sindoor with strengthening confidence in indigenous, AI-enabled systems.
Border Surveillance Goes National
The 2025 article described the AI-based Intrusion Detection System under the Comprehensive Integrated Border Management System (CIBMS) and roughly 140 smart surveillance systems deployed at India’s borders. That project has now scaled into a national program.
In May 2026, Union Home Minister Amit Shah announced the Smart Border Project, which builds on and significantly expands CIBMS. It fuses advanced sensors, AI, and real-time data analytics into a 24×7 automated surveillance grid — combining UAV-based aerial surveillance, underground fibre-optic sensors that detect ground vibration from human movement, and thermal/infrared cameras on smart poles feeding centralized command stations. The government has targeted a full national rollout within twelve months of the announcement, leaning on India’s expanded domestic drone manufacturing capacity under the PLI scheme.
The Numbers Have Moved
The scale of AI adoption across the armed forces has grown measurably since the 2025 figures:
- AI projects across the services have grown from roughly 75 (the figure cited in mid-2025) to 129 approved AI projects as of 2026.
- Of the 70 AI projects identified under the DPSU-level AI roadmap, 40 have now been completed.
- DAIPA’s dedicated funding — ₹100 crore annually — remains in place, now backed by a record ₹7.85 lakh crore defence budget for 2026–27 (up from ₹6.81 lakh crore in FY2025-26), with a notable rise in capital expenditure aimed at modernisation and domestic capability.
- On the startup side, the iDEX ecosystem has grown to more than 650 startups supported, over 400 contracts signed, and more than ₹2,400 crore in defence procurement channeled through the program — with DISC challenge editions crossing the 13-edition mark.
The New Priority: Sovereign, Air-Gapped AI
Perhaps the most significant strategic shift since 2025 is India’s explicit move toward sovereign AI for defence, rather than adapting foreign models. DRDO is now developing an indigenous AI system purpose-built for cyber defence — covering vulnerability discovery, malware analysis, and threat intelligence — designed to run in a fully air-gapped environment with all model weights, training data, and inference logs kept on defence-controlled infrastructure.
At the India-AI Impact Summit 2026, DRDO Director General Chandrika Kaushik framed this directly: India “cannot afford to depend on AI models coming from abroad” for national security applications. In parallel, DRDO has completed military field trials of an indigenous fibre-based Quantum Key Distribution (QKD) system, aimed at securing strategic communications against future threats, including quantum-computing-enabled attacks.
What Hasn’t Changed
The underlying thesis from the 2025 piece still holds, and arguably holds more strongly now: scaled AI adoption in defence depends on agile collaboration between the armed forces, research institutions, and industry, anchored by trusted data ecosystems and continuous innovation. What’s changed is the evidence. India has moved from “isolated pilots” — the mandate the original article called for — toward integrated, combat-tested, and increasingly sovereign AI systems. The next twelve months, with the Smart Border rollout and continued DPSU project completions, will be the clearest test yet of whether that momentum holds.