Urvashi Sarkar

Thoughts, stories and ideas.

Category: Foreign Policy

  • AI in Warfare: What Military Planners in Delhi and Islamabad Are Thinking

    In a quiet office at a defense research institute in New Delhi, a retired brigadier spread a set of briefing slides across his desk and told me something I had not expected to hear: “We are not worried about autonomous weapons. We are worried about autonomous decision-support.” The distinction matters. Much of the public debate about artificial intelligence in warfare focuses on lethal autonomous weapons systems — drones that select and engage targets without human authorization. But the conversations happening inside South Asian defense establishments are about a different layer of the problem: the AI-assisted analytical tools that compress the time between detecting a threat and recommending a response, tools that do not fire weapons but that shape the informational environment in which a human commander decides whether to fire.

    In Islamabad, I heard a parallel concern from a retired air vice marshal who had participated in Track II dialogues on strategic stability. Pakistan’s defense planners are acutely aware that India’s investment in AI-driven intelligence, surveillance, and reconnaissance capabilities could erode the warning time Pakistan relies on to assess whether a conventional Indian military mobilization constitutes an existential threat — the threshold at which Pakistan’s nuclear doctrine permits first use. “If their AI shortens their decision cycle to hours,” he said, “our decision cycle has to shorten too. And when both sides are making faster decisions with less information, the space for de-escalation disappears.” Neither side has published doctrine on how AI integration affects nuclear thresholds. Neither side has proposed bilateral guardrails.

    The gap between technological capability and doctrinal clarity is not unique to South Asia — the United States and China face similar ambiguities. But the India-Pakistan dyad operates under tighter geographic constraints, shorter missile flight times, and a history of crises that escalated faster than either side intended. Layering AI-accelerated decision-support onto this environment without parallel investment in crisis communication protocols is, as one former Indian national security adviser put it to me, “adding speed to a system that already has too little friction.” The conversations are happening. The doctrines are not keeping pace. And the scholars who study escalation dynamics in this region are increasingly writing papers whose conclusions amount to a single, unsettling admission: we do not yet have the analytical tools to model what happens when both sides’ analytical tools start modeling each other.

  • The Water Wars India and Pakistan Aren’t Fighting Yet

    The Indus Waters Treaty of 1960 is routinely described as one of the most successful water-sharing agreements in the world — a diplomatic instrument that survived three wars, a nuclear standoff, and decades of political hostility between India and Pakistan. That description is increasingly misleading. The treaty was designed for a hydrological reality that no longer exists. Glacier melt rates in the western Himalayas have accelerated beyond the projections that informed the treaty’s allocation framework. Seasonal flow patterns in the Chenab and Jhelum — the “western rivers” assigned to Pakistan — have shifted, with peak discharge arriving earlier and receding faster, compressing the irrigation window that Punjab’s wheat crop depends on. India’s construction of run-of-river hydroelectric projects on these tributaries is legal under the treaty’s terms but functionally alters downstream timing in ways that the 1960 text did not anticipate.

    Water scarcity between nuclear-armed neighbors does not announce itself as a crisis. It arrives as a series of administrative decisions — a dam’s filling schedule, a canal’s diversion rate — each one defensible in isolation, collectively destabilizing.

    Pakistan has twice invoked the treaty’s dispute resolution mechanism in the past decade, and the two processes — one at the International Court of Arbitration, the other through a World Bank-appointed neutral expert — produced contradictory jurisdictional rulings that remain unresolved. The institutional architecture that once made the treaty resilient is now generating its own friction. Meanwhile, both countries face domestic water crises that have nothing to do with the bilateral relationship: India’s groundwater depletion in Haryana and Rajasthan, Pakistan’s urban water stress in Karachi and Lahore. These internal pressures make political leaders on both sides less willing to appear conciliatory on transboundary issues, even when conciliation would serve their own agricultural interests.

    The risk is not a water war in the cinematic sense — tanks rolling toward a dam. The risk is a slow delegitimization of the treaty framework itself, creating a vacuum where every upstream infrastructure decision becomes a bilateral irritant, every downstream shortage becomes a nationalist grievance, and the two governments lose the shared institutional language they once used to manage disagreement. Several water diplomacy scholars I spoke with in both Islamabad and New Delhi used the same phrase independently: “the treaty is being hollowed out from the inside.” Replacing it with something better would require a level of political trust that neither capital can currently afford. Letting it decay carries consequences that neither capital can afford either.

Urvashi Sarkar

Urvashi Sarkar – 2026 – All rights reserved.

web things by Studio Goncharov