In 2021, a man wanted by India for mass murder could choose almost any address in Pakistan and sleep soundly at it. He could rent a house in a Karachi suburb, take a flat in central Lahore, settle near the garrison in Rawalpindi, or run a seminary in a Khyber Pakhtunkhwa border town. Geography itself was a form of protection. The distance from the Indian border, the density of a Pakistani city, the proximity of an army cantonment, the lawlessness of a tribal district: each of these was a defensive asset, and a wanted man could combine them to build a personal fortress out of nothing more than location. The safe haven was not a single place. It was a property of the whole map.

That map no longer offers what it once did. Over roughly five years, a campaign of targeted eliminations has moved across the Pakistani landscape with a logic that becomes visible only when the individual killings are plotted geographically rather than chronologically. The first strikes clustered in places where a foreign hand was easy to imagine: chaotic Karachi, where a body in a working-class lane attracts little official curiosity, and the border districts where smuggling and infiltration already blur the line between civilian and combatant. Then the campaign reached inward. It touched Sialkot, where a figure linked to the Pathankot airbase attack was shot dead inside a mosque. It reached Rawalpindi, the city the Pakistan Army controls more tightly than any other. It crossed the Line of Control into Pakistan-Occupied Kashmir. By 2026 it had penetrated Lahore, the cultural capital and the home of the country’s most carefully protected militant leadership. The geographic safety net was not removed in a single act. It was shredded, one city at a time, until the wanted men of Pakistan could no longer point to a single province and call it secure.

This is the story of that contraction. It is the capstone of a long examination of Pakistan’s terror geography, and it does something the city-by-city guides could not. It holds the entire country in one frame and asks what the shrinking sanctuary means. The argument here is direct. The safe haven has not merely been damaged in places. Its defining feature, the promise that location could substitute for security, has been broken everywhere it was tested. What remains is contested ground, and the men who once treated the Pakistani map as a wall now treat it as a maze with no reliable exit. Tracing how that happened, city by city and year by year, is the work of the pages that follow, and the conclusion they build toward is one this analysis will defend in full: the geography that protected terrorism for three decades has been turned, deliberately and methodically, against the people it used to shelter.

Geography and Strategic Position

To understand how much has changed, it helps to reconstruct the geography as it functioned at the campaign’s start. Pakistan is not a small country, and its terror infrastructure was never concentrated in one spot. It was distributed across four provinces, two contested territories, and a tribal periphery, and the distribution itself was a strategic design. Each region offered a different kind of cover, and the network used them in combination, the way a careful investor spreads risk across uncorrelated assets so that no single shock can wipe out the portfolio.

Punjab was the institutional core. The province holds Lahore, the country’s second-largest city and the seat of Lashkar-e-Taiba’s central apparatus through the Markaz-e-Taiba complex at nearby Muridke. It holds Bahawalpur, the headquarters town of Jaish-e-Mohammed and the home district of Masood Azhar. It holds Rawalpindi, where the Pakistan Army’s General Headquarters sits a short drive from neighborhoods that have sheltered Hizbul Mujahideen figures. It holds Sialkot, a border manufacturing city that doubles as an operations hub for groups staging across the frontier. The deep concentration of anti-India militancy in a single province was not accidental, a pattern explored in the analysis of Pakistan’s terror heartland. The army is a Punjabi institution, drawing the bulk of its officer corps from the province, and the groups it cultivated grew where its officers and recruiters already lived. Punjab’s value to the network was not just that it held the headquarters. It was that the headquarters sat inside the most prosperous, most populous, most politically powerful province in the country, surrounded by the institutions of the state itself. A wanted man in Punjab was hidden not in a wilderness but in a crowd of soldiers, bureaucrats, and ordinary citizens, and that camouflage was supposed to be the strongest of all.

It is worth pausing on how this geography came to exist, because the safe haven was not a natural feature of the landscape. It was built. Over decades, a doctrine of cultivating militant proxies as instruments of policy toward both Kashmir and Afghanistan produced a physical and institutional footprint that spread wherever it was useful. Training camps were placed where terrain favored them. Headquarters were placed where the state could watch over them. Recruitment seminaries were placed where poverty supplied a steady stream of young men. The distribution that looks, on a map, like a scattering of unrelated facilities was in fact the product of a long accumulation of deliberate choices, each one made because a particular location offered a particular advantage. The campaign of eliminations, when it came, was therefore not attacking an accident. It was attacking an architecture, and the fact that the architecture had been assembled piece by piece meant it could, in principle, be dismantled piece by piece. That is the logic the shadow war set out to apply.

Sindh provided a different asset: scale and chaos. Karachi, with a population well past twenty million, is large enough to absorb a wanted man completely. Its sheer size, its history of sectarian violence, its overstretched and politicized police, and its long entanglement with organized crime made it a place where one more body in a lane did not demand explanation. A killing in Karachi could be a land dispute, a gang vendetta, a sectarian hit, or a robbery gone wrong, and the ambiguity worked entirely in the network’s favor. Sindh’s interior, the impoverished districts running north toward Punjab, offered a quieter kind of cover through a dense network of seminaries that fed recruits into the larger organizations. That system of madrassas and charity fronts, registered as educational and welfare institutions, is mapped in the study of the JuD network across Sindh. The interior districts were poor enough that the state’s presence was thin, and a seminary that doubled as a recruitment center could operate for years without serious official scrutiny.

Khyber Pakhtunkhwa and the former tribal agencies offered the cover of ungoverned terrain. The mountainous belt from Bajaur to Landi Kotal to the Waziristans had spent decades outside effective civil administration. Outsiders were noticed instantly, tribal codes protected guests, and the writ of the state thinned to nothing in the high valleys. A Lashkar operative running a madrassa in a border town there enjoyed the protection of distance, terrain, and a social order that treated the central government as a remote and often hostile presence. The texture of that landscape is set out in the survey of the Khyber Pakhtunkhwa terror landscape and the closer look at North and South Waziristan. The tribal belt added a further layer of protection that no city could match: it was already so violent, and so densely populated by armed groups of every kind, that any single death could plausibly be assigned to a feud, a sectarian quarrel, or a clash with the Tehreek-e-Taliban Pakistan. The fog of the frontier was a defensive asset all its own.

Pakistan-Occupied Kashmir functioned as the launching zone. Towns such as Muzaffarabad and Rawalakot sat close to the Line of Control, and their value lay in proximity to the infiltration routes that carry fighters into Indian-administered Kashmir. The PoK towns were never primarily residential safe havens in the way Karachi was. They were staging grounds, and the men who lived in them did so because the work was there, an arrangement examined in the profile of Rawalakot as a launching pad. The infiltration corridors themselves, the river valleys and mountain passes that carry fighters across a fortified frontier, are charted in the analysis of the Line of Control infiltration routes. What made PoK feel secure was not urban camouflage but two physical facts: the heavy military presence around the launching infrastructure, and the fortified, mined, surveilled frontier that separated the staging towns from India itself.

Add to this the diplomatic anomaly of Islamabad, where embassies and intelligence headquarters share a city with the offices of designated organizations, a contradiction examined in the study of the Islamabad diplomatic terror nexus. Add Balochistan’s Quetta, where Afghan-facing networks maintain a supply backbone of weapons markets and smuggling routes. Add the camps themselves, the training facilities scattered across Punjab and PoK that are inventoried in the survey of terror training camps mapped across Pakistan, and the cross-border instability traced in the examination of the Afghan border terror spillover. The result, around 2021, was a country-sized sanctuary with internal specialization. A wanted man did not need a single safe place. He had a portfolio. He could be born into a Bahawalpur seminary, trained at a camp in PoK, hardened in a Waziristan valley, and retired into a Karachi suburb, and at every stage geography worked for him. That portfolio is what the shadow war set out to devalue, and the chapters that follow trace, asset by asset, how the devaluation was carried out.

Terror Organizations Present

The geographic spread of sanctuary mirrored the organizational spread of the network, and the two cannot be separated. Each major group occupied its own territory, and the safe haven’s strength came partly from the fact that no single elimination, and no single strike, could reach all of them at once. To grasp why the contraction took five years rather than five months, it is necessary to see the network not as one organization but as a federation of them, each with its own map.

Lashkar-e-Taiba held the densest and most widely distributed footprint. Its institutional center in Punjab, anchored by the Muridke campus near Lahore, has been documented at length, including in the examination of the Muridke compound and seminary. But the group’s operatives were never confined to one province. Lashkar figures lived in Karachi, ran recruitment through the Sindh seminaries, maintained men in Khyber Pakhtunkhwa border towns, staged operations through PoK, and held senior leadership in the Lahore area. When a Lashkar commander was killed in a Karachi lane and another in a Landi Kotal market, the two deaths were not separate stories. They were strikes against a single organization that had distributed itself across the country precisely so that it could not be decapitated in one blow. The group’s resilience was geographic by design: kill a man in one province, and the organization continued to function in the other three.

Jaish-e-Mohammed concentrated more tightly. Its headquarters at Bahawalpur and its operations hub at Sialkot gave it a Punjab-centered geography, a structure traced in the guide to the Sialkot operations hub and the profile of Bahawalpur as the Jaish headquarters. The Jaish footprint was smaller than Lashkar’s, which made it more exposed to geographic pressure. With fewer cities to retreat into, a campaign that penetrated Punjab penetrated nearly all of Jaish’s home ground. The group compensated with the ferocity of its operational record, including the Pathankot airbase attack of January 2016, in which seven Indian security personnel and one civilian were killed, but it could never match Lashkar’s geographic depth.

Hizbul Mujahideen, the most Kashmir-focused of the major groups, kept its launching apparatus close to the Line of Control and its senior figures in the Punjabi cities, including Rawalpindi. Al-Badr, the smaller and often overlooked Kashmir outfit, shared much of this geography. The Khalistan groups occupied a different map entirely. They maintained a presence in the Punjab of Pakistan, but their center of gravity lay in the diaspora communities of Canada and the United Kingdom, which is why their story runs partly outside the Pakistani frame and follows a different logic of legal cover rather than state sponsorship.

Beneath these named organizations sat the connective tissue: the seminaries, the charity fronts, the financiers, the logistics operators, the recruiters. This support layer was the most widely distributed of all, and it is the part of the network that geographic penetration has touched least, a point this analysis returns to in detail. The full architecture of that connective tissue, organized by facility type and city, is set out in the examination of Pakistan’s terror safe haven network. The organizations, then, were not a single target. They were a federation of targets, spread deliberately so that the map itself protected them, and held together by an infrastructure layer that no targeted-killing campaign could easily see. The shadow war’s response was not to attack any one group. It was to attack the map all of them shared.

One more feature of the organizational geography deserves attention, because it shaped where the campaign could and could not reach. The relationship between the groups and the Pakistani security establishment varied by organization and by location, and that variation produced an uneven defensive shield. In some places the protection was active: a residence watched, a movement escorted, a police inquiry quietly closed. In others it was merely passive: a willingness to look away, to file no charges, to treat a wanted man as a private citizen. The active shield was concentrated around the most senior figures in the most sensitive cities, which is precisely why the penetration of those cities carries so much weight. A killing in a place where the protection was passive proves only that the state was indifferent. A killing in a place where the protection was active proves that the shield itself could be defeated. As the campaign moved from the indifferently guarded margins toward the actively guarded core, each successful strike was therefore worth more than the last, not because the target was more senior but because the defense around him was stronger. The geography of protection and the geography of vulnerability were mirror images, and the campaign read both maps at once.

The Cities That Fell First

A geographic timeline of the campaign reveals a pattern that a simple list of names cannot. The killings did not begin everywhere. They began in the places where a foreign hand was easiest to imagine and hardest to prove, and only later moved toward the places where the network felt safest. Reading the campaign as a sequence of locations, rather than a sequence of dates, exposes its logic and explains why the contraction unfolded in the order it did.

A chronological list of the killings, the kind that news coverage naturally produces, obscures this logic almost entirely. Dates tell the reader when, and a column of dates reads as an accelerating tally, impressive but shapeless. Locations tell the reader where, and a map of locations reads as a strategy. The difference is not cosmetic. A chronological account invites the question of how many, which is the wrong question, because the campaign was never primarily about volume. A geographic account invites the question of which kind of place next, which is the right question, because the campaign was about systematically retiring categories of sanctuary. First the chaotic metropolis. Then the lawless frontier. Then the heartland manufacturing city. Then the garrison city. Then the launching zone across the Line of Control. Then the headquarters complex. Then the cultural capital. Each step retired a type of cover, and once a type was retired, every place of that type was compromised at once. This is why a campaign of a few dozen killings could break a sanctuary the size of a country. It was not killing men in proportion to the country’s size. It was killing the assumptions on which the country-sized sanctuary depended.

The opening phase, from roughly mid-2021 into 2022, clustered in environments of pre-existing chaos. Pakistani officials began, from June 2021, tracking what they would later describe as a series of assassinations of men India regarded as terrorists. Saleem Rehmani, wanted by India, was shot dead in early 2022. Karachi became the recurring setting. The city’s value to the campaign was the same value it had offered the network: scale and noise. In a metropolis of more than twenty million, with sectarian killings and gangland violence already part of the daily register, the death of a middle-aged man during an evening walk did not automatically read as an act of foreign statecraft. It read, at first glance, as Karachi being Karachi. The full picture of why the city became the campaign’s primary theater is set out in the guide to Karachi as a terror capital. The method that took shape there became the campaign’s signature: two men on a motorcycle, a close-range burst from automatic weapons or a pistol, and an escape into traffic before any witness could organize a response.

This was the genius and the limit of the opening phase. By striking first in places where attribution was murky, the campaign established a method, built operational confidence, and accumulated capability without forcing Pakistan into a corner where it had to respond. But it also meant that, for a time, the network could tell itself a comforting story. The killings were happening in Karachi, the reasoning went, because Karachi is dangerous for everyone. A man in a quieter, more controlled city could believe himself outside the pattern. That belief was the green zone, and it was about to start shrinking. The men telling themselves that story were making a category error. They were treating the early Karachi killings as a fact about Karachi, when the killings were in fact a fact about the campaign, and a campaign can move.

Border districts fell next. Khyber Pakhtunkhwa’s frontier towns, where smuggling and infiltration already blurred every category, saw eliminations that extended the campaign into terrain the network had assumed was insulated by remoteness and tribal protection. A Lashkar figure killed in a Landi Kotal market broke the assumption that distance from the heartland equalled safety. Crucially, the killings in the tribal belt also exploited an attribution problem that cuts both ways. The frontier is home to the Tehreek-e-Taliban Pakistan and to sectarian outfits, and any death there can plausibly be assigned to a local feud. The campaign used this ambiguity, just as it had used Karachi’s noise. The deeper texture of that contested ground appears in the study of the Khyber Pakhtunkhwa terror landscape. What the border killings demonstrated was that even the protection of the frontier, the oldest and most trusted form of cover in the entire portfolio, was conditional. The mountains had not moved. The campaign had simply shown it could operate in them.

Sialkot came next, and Sialkot changed the meaning of the campaign. In 2023, a man identified by Indian agencies as a key figure in the 2016 Pathankot airbase attack and associated with Jaish-e-Mohammed was shot dead in the city. In the same broad period, a figure described by Indian outlets as a Lashkar commander was killed nearby. Sialkot is not Karachi. It is a Punjab city, a manufacturing center known for its export industries, a place with a functioning administration and a recognizable civic order. A killing there could not be folded into a story about urban chaos. And the target mattered. The Pathankot figure was not an anonymous foot soldier but a man linked to the planning of a major cross-border attack on an Indian airbase, an attack that had killed Indian security personnel. The message embedded in his death was unmistakable. An attack mastermind, living in a Punjab city, inside a mosque, was reachable. The geographic safety of the Punjab heartland, the network’s institutional core, had been formally questioned for the first time. The implications for that city are drawn out in the profile of the Sialkot operations hub.

This early run of killings also revealed a second signature of the campaign that compounded its psychological effect: the use of prayer time and the mosque as the targeting window. A man shot inside or just outside a mosque, during or near the hour of prayer, is a man whose most predictable routine has been turned into a vulnerability. Everyone knows where a devout man will be at prayer time, and the campaign exploited that certainty. The choice was not careless. It signaled a depth of local surveillance that no wanted man could easily counter, because the only way to defeat it was to abandon the public religious life that anchored a respected figure’s standing in his community. The mosque killings told the network something colder than any single death. They said the campaign knew the targets’ schedules.

The early phase also established the attribution architecture that would protect the campaign for years. A killing carried out by two men on a motorcycle, in a crowded city, with no claim of responsibility and no captured operative, leaves an adversary almost nothing to work with. Pakistani officials, over time, came to attribute a growing number of these deaths to a foreign intelligence hand, and by 2024 international reporting had given those claims wider circulation, with one British newspaper investigation describing documentation alleging direct involvement and connecting close to twenty killings to the pattern. India, for its part, consistently rejected the accusations. The point for this analysis is not to adjudicate the dispute but to observe how the early geographic choices made the dispute possible. By striking first where a death could be many things, the campaign ensured that attribution would always be contestable, and contestable attribution is what allowed the campaign to keep operating without triggering the kind of open confrontation that a clearly claimed act would have provoked. The geography of the opening phase was, in this sense, a legal and diplomatic strategy as much as an operational one.

Reporting on the early killings also captured a texture that the geographic argument should not lose. These were not abstract events. They were intrusions into ordinary domestic life, masked men entering a working-class home in the early afternoon, a few shots fired at close range, and a quiet street left to absorb the aftermath. The very ordinariness of the settings was the point. A wanted man who had retired into the anonymity of a modest urban neighborhood had chosen that neighborhood because it was unremarkable, and the campaign demonstrated that unremarkable was not the same as safe. Every such killing in a quiet residential lane sent a message to every other wanted man living in a quiet residential lane somewhere else. The message did not need to be spoken. It was delivered by the geography itself.

By the end of this opening sequence, the green zone had already lost its outer rings. Karachi was contested. The border belt was contested. A Punjab manufacturing city had been breached, and breached in a way that put a senior planner in the ground. What remained, in the network’s own mental map, were the truly protected places: the garrison cities, the PoK staging towns under heavy military presence, and Lahore, the seat of the most carefully shielded leadership in the country. A wanted man surveying the map in late 2023 could still find places he believed were safe. But the list was shorter than it had been, and it was shrinking, and he had no way of knowing how fast. The campaign moved next against exactly the places that remained.

The Penetration of the Garrison Cities

The second phase of the geographic contraction is the one that matters most for the central argument, because it reached the places the network had treated as structurally unreachable. If a safe haven is a promise that location can substitute for security, then the garrison cities were the strongest form of that promise. Their penetration is the clearest evidence that the promise has failed, and it deserves to be examined slowly, city by city, because each one falsified a different assumption.

There is an order to these falsifications, and the order is itself revealing. The campaign did not move randomly toward the protected core. It moved from the assumption that was easiest to break toward the assumption that was hardest. The protection of chaos was broken first, because chaos was always the weakest shield: it offered concealment but no real defense. The protection of remoteness fell next, because terrain slows an adversary but cannot stop a patient one. The protection of ordinary civic order fell with Sialkot, because civic order was never designed to repel a covert operation. Then came the genuinely hard assumptions: that intense military control could keep a city safe, and that a fortified international frontier could keep a territory beyond reach. Those were the assumptions the network trusted most, and the campaign saved them for the phase when its capability was most mature. The progression from easy to hard is not a coincidence of timing. It is the signature of a campaign that understood exactly which forms of geographic protection were strong and which were weak, and sequenced its work accordingly. By the time it reached the hardest assumptions, it had already proven the method against every easier one.

Rawalpindi is the test case. No city in Pakistan is more closely associated with the military. The Army’s General Headquarters sits there. The intelligence apparatus is headquartered in the same urban area. The cantonment imposes a level of physical control found nowhere else in the country, with checkpoints, restricted zones, and a permanent uniformed presence woven into the fabric of daily life. When Bashir Ahmad Peer, a Hizbul Mujahideen figure India regarded as a launching commander, was shot dead in Rawalpindi in early 2023, the killing carried an argument that no Karachi death could carry. A man could be eliminated in the city the Pakistan Army watches most carefully. The proximity analysis that makes this point unavoidable is laid out in the examination of the Rawalpindi military-terror nexus. The weak-state explanation, which holds that Pakistan simply cannot control its periphery, collapses entirely here. Rawalpindi is not periphery. It is the most controlled ground the state possesses, and the campaign reached into it anyway. Either the campaign penetrated the most secure city in the country, or the men sheltering near military headquarters were never as protected as everyone assumed. Both readings are ruinous for the idea of a guaranteed sanctuary.

Pakistan-Occupied Kashmir fell in the same period. The campaign crossed the Line of Control and reached the staging towns, including the penetration of Rawalakot, where a Lashkar-linked figure was killed inside a mosque. PoK was supposed to be insulated by two things: the heavy military presence that surrounds the launching infrastructure, and the simple fact that it sits across a fortified, mined, and surveilled frontier from India. A killing in Rawalakot defeated both assumptions at once. It meant either that local assets had been recruited inside PoK itself, implying deep network penetration, or that the campaign could project force across the Line of Control, implying a reach the network had no defense against. Either reading was catastrophic for the sense of sanctuary. The strategic meaning of the PoK breach is examined in the analysis of Rawalakot as a launching pad. For decades the launching towns had felt secure precisely because they were close to the army and close to the frontier. The campaign turned both of those comforts into liabilities.

Through 2025 the contraction accelerated, and the early months of that year amounted to a near-complete tour of the Pakistani map. In Khyber Pakhtunkhwa in February, Maulana Kashif Ali, the head of a political front associated with Lashkar-e-Taiba and a brother-in-law of Hafiz Saeed, was shot dead at his residence by attackers who arrived on a motorcycle and used automatic weapons. The detail matters: Ali was killed at home, the place a man controls most completely, and his death struck at the group’s effort to convert itself into a legitimate political movement. In Balochistan in March, Mufti Shah Mir, a cleric described as deeply tied to the intelligence services, was killed by motorcycle-borne gunmen, extending the campaign into a province that had barely featured before. That same month, Abu Qatal, a Lashkar commander Indian agencies linked to the June 2024 Reasi bus attack that killed nine pilgrims and injured dozens, was killed in Jhelum, a Punjab district. The previous year had also seen the killing of a Jaish figure in Khyber Pakhtunkhwa. Read geographically, these deaths form a sequence that touches the tribal belt, Balochistan, the Punjab interior, and beyond, within a single calendar year. No region was exempt. The map had no quiet corner left.

Then, in May 2025, the contraction took a form no targeted killing could match. Operation Sindoor, India’s military response to the Pahalgam attack of April 2025, in which twenty-six civilians were killed, struck nine sites with missiles and precision munitions in a strike that lasted barely more than twenty minutes. Five of the sites lay in Pakistan-Occupied Kashmir, including facilities near Muzaffarabad, Kotli, and Bhimber. Four lay inside Pakistan proper, and two of those four were the holiest ground the network possessed. The Markaz-e-Taiba complex at Muridke, the institutional heart of Lashkar-e-Taiba, was hit. The Markaz Subhan Allah at Bahawalpur, the headquarters of Jaish-e-Mohammed, was hit. These were not anonymous lanes. They were the addresses around which the entire safe-haven geography had been organized, and they had long been considered too sensitive, too deep inside the Punjabi heartland, for India to strike. Analysts noted that an attack so far into Punjab had not occurred since the 1971 war. By hitting Muridke and Bahawalpur, the operation delivered, in a single night, the message that the targeted killings had been delivering city by city for years: old geographic safe havens were no longer safe. A senior Indian statement captured the doctrine behind it, declaring that the country would no longer treat terrorist leaders and the governments sheltering them as separate entities. The covert campaign and the conventional strike had converged on exactly the same conclusion about the map.

The geographic detail of the operation deserves a closer look, because the choice of targets was itself an argument. Five of the nine sites lay across the Line of Control in Pakistan-Occupied Kashmir, the launching belt, and striking them was, in geographic terms, an extension of a pattern India had established before. Hitting the staging zones was aggressive but not unprecedented. The four sites inside Pakistan proper were the genuine departure, and the two that drew the most attention, Muridke and Bahawalpur, were chosen precisely because they had been considered untouchable. Their selection was a deliberate political signal. Both had been linked to anti-India militant groups for decades, and their locations deep in the Punjabi heartland were widely perceived as too sensitive for India to consider striking. By hitting them anyway, the operation announced that the old geographic exemptions had been revoked. The accompanying measures reinforced the message. India suspended a long-standing water-sharing treaty, closed the principal land crossing, and cut bilateral trade, surrounding the military strike with a wider demonstration that the relationship’s familiar boundaries no longer held. The strike on the map was matched by a strike on the rulebook.

It is also worth being precise about what the strikes did and did not prove. They proved that India could reach the symbolic centers of the network with conventional force. They did not prove that the men those centers housed were inside them at the time, and indeed the network’s long-established practice of evacuating known facilities in anticipation of a crisis means the human cost of the strikes is genuinely uncertain. The geographic point, however, does not depend on the casualty count. Whether or not a particular commander was present, the campaign had demonstrated that the headquarters address was no longer a sanctuary, and a headquarters that must be evacuated whenever tension rises is a headquarters that has already lost much of its value. The strike degraded the place as a place, regardless of who was standing in it.

Eliminations did not stop for the war. Within days of the May 10 ceasefire, the targeted-killing campaign resumed. Saifullah Khalid, a Lashkar figure Indian agencies connected to attacks reaching back two decades, including a 2005 assault in Bengaluru and a 2006 attack in Nagpur, was killed in Matli, in the Badin district of Sindh, on May 18. The timing carried its own message. The ceasefire had stopped the missiles, but it had not stopped the motorcycles. The covert track and the conventional track were independently operational, and a pause in one did not pause the other.

The final ring was Lahore. The cultural capital was the network’s deepest sanctuary, the city where the most senior and most carefully protected figures lived, a status examined in the guide to Lahore as a Lashkar-e-Taiba headquarters city. When a co-founder of Lashkar-e-Taiba and a close deputy of Hafiz Saeed was attacked in Lahore, and when a vehicle-borne explosion struck the same city, the last protected address in the country had been contested. The man survived the attack on him, and the survival matters for the debate that follows. But the geographic point was already made. There was no longer a Pakistani city of which the network could say, with confidence, that its senior leadership was beyond reach.

A breach of Lahore carries a weight that a breach of any other city does not, and it is worth being explicit about why. Lahore was not merely one more dot. It was the network’s idea of what a safe place was. If the campaign could reach a Lashkar co-founder in Lahore, then the concept of the safe city had no remaining example to point to. Every wanted man elsewhere in the country had, until then, been able to imagine relocating to the deepest sanctuary if his own situation became untenable. The Lahore attack removed the imagined destination. It did not just redden one more city; it closed the last exit on the map. The contraction was complete in the sense that mattered most: every category of sanctuary had been tested, and none had held.

The Expanding Red Zone and the Contracting Green Zone

The clearest way to see what has happened is to picture the Pakistani map in two colors and watch them move. Color the country green where a wanted man could reasonably feel safe and red where the campaign had demonstrated reach. The shadow war is the story of red expanding and green contracting, and the two movements are not independent. They are the same movement seen from two sides.

In 2021, almost the entire map was green. The red existed only as a possibility, a thing Indian commentators threatened and Pakistani officials dismissed. There was no city where a documented elimination had established that the campaign could operate. The green zone was the whole country, and within it the network enjoyed not just safety but the freedom that safety produces: predictable routines, open movement, public appearances, fixed residences, command exercised face to face. A leader could run an organization the way any executive runs an institution, because nothing about his daily life had to be defensive. He could keep an office, hold a weekly gathering, attend prayers at the same mosque, sleep in the same bed, and trust that the geography around him would absorb any threat before it reached him.

Red began as isolated points. A killing in Karachi placed a single red dot on the map. By itself, a dot is not a zone. The network could quarantine it, treat Karachi as a special case, and keep the rest of the country green. But the campaign did not produce isolated dots. It produced dots that connected. A second Karachi killing, then a border-district killing, then Sialkot, then Rawalpindi: each new point did not merely add to the red, it linked the red into a shape. And a shape implies an area. Once a wanted man could see killings in Karachi and Sialkot and Rawalpindi, he could no longer believe the red was confined to specific unlucky addresses. He had to assume the red was a region, that the region was growing, and that his own city might already be inside it without his knowing.

This is the crucial mechanism, and it explains why the contraction has been faster than the raw number of killings would suggest. The green zone does not shrink only where a body has actually fallen. It shrinks wherever the network can no longer rule out a strike. A city with no recorded elimination still turns from green to grey the moment a comparable city nearby is hit, because the comparison destroys the reason the first city felt safe. Sialkot’s penetration did not just redden Sialkot. It greyed every Punjab manufacturing city of similar size and administration, because the feature that had made them feel safe, ordinary civic order, had been shown to be no defense. Rawalpindi’s penetration did not just redden Rawalpindi. It greyed the entire category of garrison cities, because the thing that made garrison cities feel safe, intense military control, had been shown to be no defense either. Red expands by killing. Green contracts by inference. Inference moves faster than any motorcycle, which is why, by 2026, there was effectively no green left even though most Pakistani addresses had never seen a strike.

The expanding red zone also changes the value of every remaining green patch in a way that compounds the pressure. As the safe ground shrinks, the surviving safe ground becomes crowded and conspicuous. If only a handful of districts still feel secure, the network concentrates its most valuable people there, and concentration is itself a vulnerability. A campaign that has reduced the green zone to a few areas has also told its adversary exactly where to look. The contraction is therefore self-accelerating in its late stages. The smaller the green zone, the more it reveals; and the more it reveals, the smaller it becomes. The geographic logic runs only one direction, and it tightens on itself the longer it operates.

A further effect is worth naming, because it explains why the psychological cost outran the physical one. A wanted man living inside the shrinking green zone could never know, on any given day, whether his city was still green. He could not call the adversary and ask. He could only watch the news, count the dots, and estimate. And estimation under that kind of uncertainty does not produce calm. It produces a permanent low-grade dread, a sense that the red might already have arrived and simply not announced itself. The campaign did not need to strike every city to put every city on edge. It needed only to make the boundary of the red zone invisible, so that no wanted man could ever be sure which side of it he was standing on. Uncertainty, properly weaponized, covers more ground than any number of gunmen.

Worth dwelling on is the grey zone, the band of territory that is neither confirmed red nor reliably green, because that band is where most of the network actually lives. A truly green city is one with no killing nearby and no comparable city struck. A red city is one with a documented elimination. The grey zone is everything in between: cities with no strike of their own but with enough resemblance to a struck city that safety can no longer be assumed. As the campaign progressed, the grey zone swallowed the green. Each new killing converted not just its own city but a whole class of similar cities from green to grey, and grey, for a man whose life depends on certainty, is functionally indistinguishable from red. He cannot relax in a grey city. He can only wait in one. The campaign’s geographic achievement is best measured not by the count of red dots but by the speed at which grey consumed the map, because grey is where the fear lives.

Consider how this looked from inside a single mid-tier commander’s situation. In 2021 he might have lived openly in a provincial city with no particular precautions, confident that his distance from the obvious theaters kept him clear. By 2023, with Karachi, the border belt, and a Punjab manufacturing city all penetrated, his city had not been struck but had lost every reason to feel exempt. By 2025, with the garrison cities and PoK breached and the heartland headquarters hit from the air, there was no argument left that his location protected him. He had not moved. No gunman had visited his street. And yet his geographic situation had been completely transformed, from secure to grey to effectively indefensible, by events that all happened somewhere else. That is the contraction working exactly as designed: it changed his world without ever touching his door.

Set against this is the network’s counter-move, which is real and must be acknowledged. The men inside the shrinking green zone did not simply wait. They adapted. They upgraded residential security, abandoned predictable routines, reduced public appearances, changed addresses, restricted their communications, and pushed operational command downward to subordinates who could be sacrificed without strategic loss. This behavioral shift is documented across the profiles of the surviving leadership, including the examination of how terror leaders still at large have adapted. Whether that adaptation amounts to a genuine restoration of safety, or merely a more anxious way of being unsafe, is the question the next section confronts. But the geographic point stands on its own. The map that was green in 2021 is not green now. The safety net woven from distance, density, terrain, and garrison control has holes in every panel, and the men who depended on it know it.

The Infrastructure of Shelter

A campaign that moves across a map kills people. It does not, by the simple act of killing, dismantle buildings, defund seminaries, or close the offices of charity fronts. This distinction is essential, and it is where the argument must be most careful, because the infrastructure of shelter is the part of the safe haven that geographic penetration has touched least. An honest account of the contraction has to draw a sharp line between what the campaign has done to individuals and what it has, and has not, done to the system that produces them.

Physical and institutional architecture of Pakistan’s terror sanctuary is extensive and, in important ways, intact. The seminaries that recruit and indoctrinate continue to operate. The charity fronts that move money and provide legitimate-seeming employment for wanted men continue to function. The training facilities, even where struck, can be rebuilt, and the campuses that combine a mosque, a school, a clinic, and a dormitory remain difficult targets precisely because they are entangled with genuine civilian use. A missile aimed at such a campus risks killing students, patients, and worshippers alongside militants, and that entanglement is itself a form of protection, a human shield built into the architecture. The full inventory of this architecture, organized by facility type, is set out in the analysis of Pakistan’s terror safe haven network. That inventory describes a system, not a collection of hiding spots, and a system has a resilience that no individual has.

The Operation Sindoor strikes complicate this picture in an instructive way. The missiles that hit Muridke and Bahawalpur did damage physical infrastructure, and they did so at the symbolic center of the network. This was the one moment in the campaign when geographic penetration and structural damage clearly coincided, when the contraction stopped being a matter of individual deaths and became a matter of bricks and concrete. But even here the limits are visible. A campus can be repaired. A headquarters can be relocated. The institutional knowledge, the recruitment relationships, the financial channels, and the social legitimacy that the buildings housed are not destroyed when the buildings are. A defense analysis of the operation noted that the terrorist groups had long learned to vacate their known facilities in anticipation of Indian action, treating evacuation as a routine drill. The strikes degraded the infrastructure and embarrassed it. They did not abolish it.

This is why the safe haven, even now, is best described as damaged rather than destroyed. The campaign has been extraordinarily effective against named, senior, individually identifiable men. It has been far less effective against the diffuse support layer: the unnamed financiers, the seminary administrators, the logistics operators, the recruiters who never appear on a designated-terrorist list. These people enjoy a geographic safety the senior leadership has lost, not because they live in better-defended places, but because they are invisible to the targeting process. A campaign that depends on identifying specific individuals cannot reach a person it cannot name. The support layer is, in effect, hidden by anonymity rather than by geography, and anonymity is a form of cover the campaign has no clear method to strip away.

Infrastructure also retains a regenerative capacity that the geographic argument must not understate. The seminaries are, among other things, replacement factories. They produce new recruits faster than a targeted-killing campaign produces casualties at the leadership level. The deep bench of cadres means that the elimination of a commander creates a vacancy that the system is structured to fill. A field commander killed on a Karachi street is mourned for a week and replaced within a month, because the pipeline that made him is already producing his successor. Geographic penetration removes the man. It does not remove the pipeline that made the man, and it does not remove the next man the pipeline is already producing. The recruitment apparatus mapped across Sindh, the training camps inventoried across Punjab and PoK, and the financial channels that sustain them all continue to function on a map that no campaign of targeted strikes can fully clear.

The dual-use character of the infrastructure deserves its own emphasis, because it is the single greatest obstacle to dismantling the safe haven by force. Many of the buildings that house the network also house genuine civilian functions. A seminary that produces recruits also educates children who will never carry a weapon. A charity front that launders reputation and money also delivers real relief to real disaster victims. A campus that trains operatives also runs a clinic that treats the sick. This entanglement is not accidental. It is a designed feature, because it converts every civilian who depends on the institution into an unwitting shield, and it forces any adversary contemplating a strike to weigh the militant target against the civilian cost. The campaign of targeted killings has largely sidestepped this dilemma by aiming at individuals in the open rather than at institutions. But that very precision is also a limitation: a campaign that cannot strike the dual-use campus cannot easily dismantle the part of the safe haven that the campus represents.

A financial dimension extends the same point. The channels that move money to the network are geographically diffuse and institutionally camouflaged, routed through charities, businesses, and informal transfer systems that cross provincial and national borders. International financial-monitoring bodies have for years pressed Pakistan over exactly this infrastructure, and the pressure has produced some tightening. But money, like ideology, has no fixed address, and a targeted-killing campaign has no obvious way to reach a financier whose role leaves no public signature. The geography of the money is the geography the campaign has touched least, and as long as the money moves, the network can pay for the recruitment, the training, and the replacement of every commander it loses. Geographic penetration has reached the men. It has not reached the means.

The honest conclusion is therefore a divided one. On the question of individuals, the safe haven has contracted dramatically. The named leadership has lost the geographic freedom it once enjoyed, and that loss is the substance of this entire analysis. On the question of infrastructure, the safe haven has been bruised and embarrassed but not dismantled. The system that produces and protects terrorism continues to operate, distributed across a map that no campaign of targeted strikes can fully clear. Holding both findings at once is the only intellectually honest position, and the next two sections push directly on the tension between them, because that tension is where the real strategic question lives.

Adaptation Versus Genuine Contraction

Here the analysis must confront the strongest objection to its own argument. A skeptic could grant every killing described above and still deny that the safe-haven system has truly contracted. The skeptic’s case runs like this: the geography has not changed, only the behavior of individuals within it has. The seminaries still stand. The cities are still there. What the campaign has produced is not a shrinking sanctuary but a population of frightened men who have changed their addresses and their habits. On this reading, the green zone has not disappeared. It has simply become harder to occupy comfortably, and the network has adjusted by occupying it differently.

This objection deserves a serious answer rather than a dismissal, because part of it is plainly correct. The men who have survived the campaign have indeed adapted, and their adaptation has been thorough. They have moved into smaller, more defensible residences, often changing them on no fixed schedule. They have stopped keeping predictable routines, varying the timing of their movements so that no observer can anticipate them. They have reduced or eliminated public appearances, withdrawing from the gatherings and ceremonies that once advertised their authority. They have layered their personal security with guards and screening. They have pushed their day-to-day operational role onto subordinates who can act in their place. The behavioral change is real, well documented, and visible across the profiles of the surviving leadership. A fair analysis cannot pretend otherwise, and this one does not.

But the skeptic’s framing contains a hidden assumption that does not survive scrutiny. The assumption is that a safe haven is a place, and that as long as the place exists, the haven exists. That is not what a safe haven is. A safe haven is a function. Its purpose is to let an organization operate: to let leaders lead, recruiters recruit, planners plan, and the institution renew itself, all with the efficiency that physical security makes possible. The value of geography to the network was never the soil itself. It was the operational freedom the soil permitted. And operational freedom is exactly what the adaptation has cost.

Consider what the adaptive measures actually subtract. A leader who has abandoned fixed routines has also abandoned the predictability that lets an organization run on schedule, that lets subordinates find him, that lets a plan move from decision to execution without friction. A leader who has stopped appearing in public has lost the visibility that builds authority, attracts recruits, projects strength, and reassures financiers that their investment is alive and active. A leader who has pushed command downward has, by definition, reduced his own control over the organization he is supposed to direct, handing real decisions to men less experienced than himself. A leader who has restricted his communications has slowed every decision that depends on him, introducing delay into a system that once moved at the speed of a phone call. These are not cosmetic adjustments. They are reductions in the very capacities that made the leader valuable. The adaptation that keeps a man alive simultaneously degrades the thing he exists to do.

This is why the adaptation-versus-contraction debate is, in the end, not a real opposition. The two are the same phenomenon described from different angles. The safe-haven system has contracted precisely by forcing the adaptation. A wanted man retreating into paranoid, communication-restricted, command-delegating isolation is not evidence that the haven survived. He is the evidence that it did not, because the haven’s whole purpose was to make that retreat unnecessary. The campaign did not need to kill every leader to break the sanctuary. It needed only to make every surviving leader too defensive to function as a leader. Survival inside the shrunken green zone is not the safe haven working. It is the safe haven failing in a way that happens to leave the occupant breathing.

There is a second, harder version of the skeptic’s case worth naming. Perhaps, the skeptic says, the surviving leaders have adapted so well that they retain full operational capability through new methods, and only appear diminished. This is an empirical claim, and it should be tested empirically rather than assumed away. The relevant evidence is operational output: whether the organizations have continued to plan and execute attacks at their former tempo. The picture is genuinely mixed, and certainty is not available. Attacks have continued, which the skeptic can cite; but they have continued under conditions of greater friction, delay, and concentrated personal risk, which the contraction thesis predicts. The structural point holds regardless of how that evidence is finally read. Even an organization that maintains its output has paid a price in friction and exposure, and an adversary that has imposed that price on a sanctuary has, by any reasonable definition, made the sanctuary smaller. A haven that still functions but functions fearfully, slowly, and at constant risk is a haven that has contracted, whatever the body count says.

A closer look at one adaptive measure, the delegation of command, shows how deep the cost runs. When a senior figure pushes his operational role onto subordinates, he is not simply sharing the workload. He is dismantling the chain of authority that made him a leader. The subordinates he hands decisions to are, almost by definition, less experienced, less trusted by the wider organization, and less able to coordinate across the federation of groups. Plans that once moved through a single decisive mind now move through a committee of cautious deputies. Errors that the senior figure would have caught go uncaught. Rivalries among the deputies, previously suppressed by his authority, surface. And because the senior figure must now communicate with his own organization through intermediaries and on restricted channels, even the decisions he does make arrive slowly and partially. The adaptation that protects his life corrodes the institution beneath him. A network run this way does not stop functioning, but it functions like a body whose nervous system has been partially severed: the limbs still move, but the coordination is gone.

There is also a generational cost that the adaptation imposes and that compounds over time. A militant organization renews itself by transferring knowledge from experienced commanders to younger ones, and that transfer historically happened in person, in the training camp, the seminary, the planning session, the unhurried conversation. The defensive crouch makes all of those settings dangerous. A senior figure who can no longer gather his cadres, who can no longer appear at the camp, who can no longer hold the long planning meeting, is a senior figure who can no longer teach. The institutional memory that takes a decade to build begins to leak away, not because the campaign killed the man who held it but because the campaign made it unsafe for him to pass it on. The contraction therefore reaches into the network’s future as well as its present. It does not just threaten the current leadership. It interrupts the making of the next.

A useful way to settle the adaptation debate is to ask what a genuinely intact safe haven would permit, and then to check the surviving leadership against that standard. An intact safe haven would permit a leader to keep one home, attend one mosque, hold one weekly gathering, and command his organization face to face, all without a thought for his security, because the geography would be doing the work. Measured against that standard, the surviving leadership is not living in a safe haven at all. It is living in its ruins. The home is changed on no schedule. The mosque is avoided or varied. The gathering is cancelled. The command runs through intermediaries. Every one of these is a subtraction from the life the safe haven was supposed to guarantee, and the sum of the subtractions is a leadership that survives without functioning as leadership was meant to function. The skeptic who says the haven endures is pointing at the survivors and missing that the survivors are themselves the proof. They are alive because they have stopped doing the things the haven existed to make safe. That is not a haven that held. It is a haven that failed in slow motion, and left its occupants to manage the failure one precaution at a time.

Does Geographic Penetration Equal Strategic Success

Granting that the geography has genuinely contracted, a further and tougher question remains. Does that contraction amount to strategic success? It is entirely possible for a campaign to redraw a map impressively and still fail to achieve what the map-drawing was meant to achieve. Geographic reach and strategic victory are different things, and conflating them would be the central analytical error this section exists to prevent.

The case for strategic success is substantial, and it should be stated at full strength before it is qualified. The campaign has imposed real and rising costs on the network. It has eliminated experienced, hard-to-replace commanders and planners, men whose operational knowledge took years to build. It has forced a defensive crouch on the surviving leadership that measurably degrades their effectiveness. It has stripped Pakistan of the deniability that a guaranteed sanctuary provided, because a state that cannot keep its sheltered men alive in its own garrison cities can no longer credibly claim that those men are simply private citizens beyond its control. It has demonstrated, through Operation Sindoor, that the geographic logic extends to conventional force as well as covert action, closing the gap between what India would threaten and what it would actually do. And it has established a deterrent precedent. A young man weighing a future with a militant organization must now factor in a meaningful probability of a violent and premature end, with no city in the country offering reliable refuge. These are genuine achievements, and they are the substance of the case that the contraction matters. The ranking of which eliminations actually carried this kind of strategic weight, as opposed to merely adding to a tally, is examined in the analysis of the complete shadow war kill list, which makes the point that precedent and deterrent effect, not raw numbers, are where the value lies.

A case against overstating that success is equally important, and it has to be heard with the same seriousness. Geographic penetration removes men. It does not, on its own, remove the conditions that produce men. The seminaries continue to recruit. The financial channels continue to flow. The ideology that motivates the network is untouched by any number of motorcycle-borne strikes, because an idea has no address and cannot be located on a map. An organization with a deep bench of cadres can absorb the loss of named commanders and continue to function. And the period after the 2025 conflict has, by several accounts, seen the operational tempo of eliminations rise rather than fall, which cuts in two directions at once. It shows the campaign’s continuing reach, which supports the contraction thesis; but it also shows that the network has not been reduced to silence, which warns against declaring victory. A campaign can dominate the map and still not have dismantled the machine, and a machine that keeps producing recruits and plans is a strategic problem that no geographic contraction by itself solves.

One strand of the strategic question is worth isolating, because it is where the geographic contraction may matter most over the long run: deterrence and recruitment. A militant organization is, among other things, a career. It offers a young man status, purpose, belonging, and, historically, a reasonable expectation that he could rise through the ranks and live to see middle age. The safe haven was central to that offer, because it was the safe haven that made the career survivable. A campaign that has visibly broken the sanctuary has also, therefore, altered the recruitment proposition. The career it now advertises ends, with meaningful probability, in a motorcycle ambush outside a mosque or a missile through a headquarters wall, and it ends that way regardless of which city the recruit chooses. Whether this shift measurably reduces the flow of recruits is hard to establish from outside, and the seminaries continue to operate. But the logic of deterrence does not require that recruitment stop. It requires only that the cost of joining rise, and the geographic contraction has unambiguously raised it.

Set against this is a sober counter-consideration that the analysis must not skip. Deterrence through targeted killing has a mixed record, and serious scholarship on how terrorist campaigns end has long questioned whether the removal of leaders, by itself, degrades an organization or merely changes its shape. A movement with deep ideological motivation and a steady supply of aggrieved recruits can absorb a great deal of leadership attrition. It can even, in some circumstances, be hardened by it, as martyrdom narratives convert dead commanders into recruiting tools. The geographic contraction does not escape this debate. It is entirely possible that the shrinking safe haven raises the cost of membership without reducing the supply of members, in which case the campaign has imposed pain without achieving decline. The honest position is that the deterrent effect is real but unproven in magnitude, and that anyone who claims certainty about it, in either direction, is claiming more than the evidence allows.

What can be said with confidence is narrower and still significant. The geographic contraction has removed the network’s ability to operate at zero cost, and an enemy that operates at zero cost has no incentive to change. Whatever else the campaign has or has not achieved, it has installed a price. Every plan now carries the friction of a defensive leadership. Every city now carries the possibility of a strike. Every recruit now joins an organization whose senior men live in hiding. The price is not the same as victory, but it is the precondition for any victory, and the precondition has been met across the entire map.

There is a final consideration that connects the geographic contraction to the campaign’s own sustainability. Reaching ever-better-protected men in ever-better-defended cities is not a fixed task that gets easier with practice. It is a task that gets harder, because the surviving targets are precisely those who have adapted most successfully, and because the network, having watched the campaign for years, has learned to vacate, to hide, and to go quiet whenever the threat sharpens. A campaign that wants to keep contracting the safe haven from here will have to identify and reach genuinely difficult personnel, and that demands a level of intelligence tradecraft, a depth of human sourcing and a precision of targeting, that is far more taxing than the work of the opening phase. The geography that remains contestable is the hardest geography of all. This does not undo the contraction already achieved. But it is a reminder that the curve does not extend itself for free, and that the question of whether the campaign can finish what it started is, in the end, a question about intelligence capability rather than about the map. The map has been won as far as the present capability could carry it. Carrying it further is a separate problem, and an unsolved one.

The honest adjudication is that geographic penetration is a necessary component of strategic success but not the whole of it. Breaking the safe haven was always going to be a precondition for any larger result, because as long as the sanctuary held, the network could operate without cost, and an enemy that operates without cost has no reason to change. The campaign has met that precondition. It has made sanctuary expensive, conditional, and unreliable across the entire country. But meeting a precondition is not the same as winning. A defense analysis of the 2025 operation made a related point about military force, arguing that cost-imposition should be understood as one supporting tool within a broader suite of national policy instruments, alongside intelligence work, diplomacy, and pressure on terror financing, rather than as a counter-terrorism strategy complete in itself. The same logic applies to the covert campaign. It is one instrument, and a powerful one, but the conditions that produce terrorism, the recruitment pipeline, the financial flows, the ideology, and the state policy that sustains them all, are not defeated by geography alone.

What emerges as the most defensible conclusion is therefore measured rather than triumphant. The geographic contraction is a major and possibly decisive achievement at the level of denial. It has denied the network the costless sanctuary it depended on for decades, and that denial is real, durable, and visible across the entire map. It is not, by itself, an achievement at the level of destruction. It has not ended the system that produces terrorism, and it was never structured to. A reader looking for a triumphant verdict will not find an honest one here. A reader looking for an accurate one will find this: the campaign has won the geography and has not yet won the war, and the distinction between those two is the single most important thing this analysis has to say.

How the Shadow War Changed Pakistan’s Map

Step back from the individual cities and the running debate, and the transformation can be stated plainly. Pakistan’s terror geography has been inverted. For decades the map worked for the network. Distance from India, the density of great cities, the lawlessness of the tribal periphery, the iron control of the garrison towns: every feature of the landscape was a defensive asset, and the safe haven was, in the most literal sense, made of geography. That is no longer true. The same features now offer, at best, partial and provisional cover, and the men who once read the map as a fortress now read it as terrain with no reliable wall.

It is worth asking what a complete collapse of the safe haven would look like, if only to measure how far the contraction has come and how far it has not. A complete collapse would not be a map fully colored red, because no campaign can station a threat in every district at once. It would be something subtler and more total. It would be a condition in which the network itself no longer believed any location offered protection, and therefore stopped organizing its life around geography at all. In that end state, the seminaries would struggle to recruit, because the career on offer would visibly end in an unmarked grave. The financiers would withdraw, because association would have become lethal rather than merely controversial. The leadership would have ceased to function as leadership, not in a few cities but everywhere, because there would be nowhere left from which to lead. The infrastructure would wither, slowly, for want of the secure environment that let it operate. Collapse, in other words, would be a psychological fact before it was a geographic one: the moment the network stopped believing in sanctuary, the sanctuary would be over.

Pakistan has not reached that end state. The infrastructure stands. The recruitment continues. The deep bench of cadres remains. But the country has moved unmistakably along the path toward it, and the direction of travel is the finding that matters most. In 2021 the network could treat the Pakistani map as a settled asset, a fixed and reliable foundation beneath everything else it did. By 2026 it had to treat the map as a contested and shrinking one, audited city by city, with the audit consistently returning bad news. Karachi, the border belt, Sialkot, Rawalpindi, the towns of Pakistan-Occupied Kashmir, the symbolic heartland of Muridke and Bahawalpur, and finally Lahore itself: each was tested, and the safety each was supposed to provide was found to be conditional. The transformation that began as scattered violence in chaotic places became, over five years, a structural fact about the entire country. And the structural fact is this. The promise that location could substitute for security, the promise on which the whole safe haven was built, no longer holds anywhere it has been tested. That is what it means to say the geography is changing, and the change runs only one way.

The deeper lesson reaches past Pakistan. For three decades, the safe haven rested on an assumption that had quietly hardened into a law of the region: that a state which sheltered militants behind distance, density, and military control had given them something an adversary could not take away. The shadow war has falsified that assumption. It has shown that geography is not a fixed asset but a contestable one, that sanctuary is a claim that can be tested and found wanting, and that a determined adversary willing to operate patiently across a long campaign can convert a country-sized fortress into a country-sized trap. Whether other states sheltering other militants will draw the same conclusion is beyond the scope of this analysis. But the precedent has been set on Pakistani soil, in Pakistani cities, against the most carefully protected network in the region, and precedents of this kind do not stay confined to the place that produced them. The map of Pakistan’s safe haven has been redrawn. The idea that any safe haven is permanent has been redrawn with it. For the wanted men who once treated a Pakistani address as a guarantee, the lesson is now written into the landscape itself: a country can be wide, and crowded, and ringed with garrisons, and still offer no reliable place to hide. That is the inheritance the shadow war leaves behind, and it is an inheritance measured not in a single victory but in the steady, deliberate unmaking of a sanctuary that took thirty years to build.

Frequently Asked Questions

Is the safe-haven system actually shrinking, or is this an exaggeration?

It is shrinking, though the claim has to be stated precisely. The system has shrunk dramatically as a guarantee of safety for named, senior, individually identifiable men. Around 2021 such a person could live openly in almost any Pakistani city. By 2026 every category of sanctuary, the chaotic metropolis, the border district, the Punjab heartland city, the garrison town, and the launching zones of Pakistan-Occupied Kashmir, had been penetrated by a documented strike or, in the case of Muridke and Bahawalpur, a missile. What has not shrunk is the underlying infrastructure of seminaries, charity fronts, and financial channels. The accurate statement is that the safe haven has contracted sharply for individuals and far less for institutions.

Which Pakistani cities are no longer safe for wanted terrorists?

In practical terms, none can be called reliably safe. Karachi became a recurring setting for eliminations early in the campaign. Sialkot saw the killing of a figure linked to the Pathankot airbase attack. Rawalpindi, the army’s garrison city, saw a Hizbul Mujahideen figure shot dead. Towns in Pakistan-Occupied Kashmir, including Rawalakot, were penetrated. Khyber Pakhtunkhwa border areas, Balochistan, and the Punjab interior all recorded killings through 2025. Lahore, long considered the most protected city for senior leadership, saw an attack on a Lashkar-e-Taiba co-founder and a vehicle-borne explosion. The honest answer is that the campaign has demonstrated reach into every type of Pakistani city.

Have terrorists simply adapted by moving rather than the system genuinely changing?

Both are true at once, and they are not really in conflict. The surviving leadership has adapted, by changing addresses, abandoning routines, reducing public appearances, and delegating command. But a safe haven is not a place; it is a function, and its function is to permit an organization to operate efficiently. The adaptive measures that keep a man alive also strip him of the predictability, visibility, and control that made him effective. The adaptation is therefore the form the contraction takes. A leader hiding in paranoid isolation is evidence that the haven failed, not that it survived.

How quickly did the shadow war expand geographically?

The expansion ran over roughly five years, from the killings Pakistani officials began tracking in mid-2021 to the penetration of Lahore by 2026. The pace was not even. The campaign opened with a cluster of killings in environments of pre-existing chaos, then moved into the border districts, then breached a Punjab city, then reached the garrison cities and Pakistan-Occupied Kashmir, and finally touched the cultural capital. The early months of 2025 saw an especially rapid run of eliminations across multiple provinces, and the conventional strikes of Operation Sindoor in May 2025 accelerated the geographic message dramatically.

Which city was penetrated first, and which most recently?

Karachi functioned as the campaign’s early and recurring theater, valued for the scale and noise that made attribution difficult. The penetration that reads as most recent and most significant is Lahore, the seat of the country’s most carefully protected militant leadership. The progression from Karachi to Lahore is the geographic spine of the whole story. The campaign began where a foreign hand was easiest to hide and ended where the network felt safest.

Is there any Pakistani city still considered completely safe?

No city can credibly be described that way. The campaign’s geographic logic means that even a city with no recorded elimination loses its sense of safety once a comparable city nearby is struck, because the comparison destroys the reason the first city felt secure. Once garrison cities, heartland cities, and the cultural capital have all been tested, the network has no category of location left to retreat into with confidence. The remaining cover is provisional, not guaranteed.

Did Operation Sindoor change the safe-haven geography?

Significantly. The May 2025 operation struck nine sites, including the Markaz-e-Taiba complex at Muridke and the Markaz Subhan Allah at Bahawalpur, the institutional headquarters of Lashkar-e-Taiba and Jaish-e-Mohammed respectively. These were the symbolic centers of the entire safe-haven geography, deep in the Punjabi heartland, and had long been considered too sensitive to strike. Hitting them with conventional force delivered, in one night, the same message the targeted killings had delivered city by city: old geographic safe havens were no longer safe.

Does geographic penetration mean the shadow war has been won?

No, and that conflation is the most important error to avoid. Geographic penetration is a denial achievement. It has denied the network the costless sanctuary it depended on, and that is a major and possibly decisive result at the level of denial. But it is not a destruction achievement. The seminaries still recruit, the financial channels still flow, the ideology is untouched, and the network’s deep bench of cadres can absorb the loss of named commanders. The campaign has won the geography. It has not, by that alone, dismantled the machine.

Why did the campaign start in Karachi rather than the heartland?

Because Karachi offered the best conditions for an opening phase. A city of more than twenty million, with existing sectarian and gangland violence and an overstretched police force, allowed eliminations to occur without automatically reading as foreign statecraft. Starting there let the campaign establish a method, build operational confidence, and accumulate capability without forcing Pakistan into a corner. Only after the method was established did the campaign move toward the heartland and the garrison cities, where attribution was harder to obscure.

What does the penetration of Rawalpindi prove?

It eliminates the weak-state explanation for the safe haven, at least for the most protected targets. The argument that Pakistan merely cannot control its periphery collapses in Rawalpindi, because Rawalpindi is not periphery. It is the army’s garrison city, home to the General Headquarters, and the most tightly controlled urban ground the state possesses. A killing there means either that the campaign penetrated the most secure city in the country or that the men living near military headquarters were never as protected as assumed. Either reading is damaging to the idea of a guaranteed sanctuary.

Has the infrastructure of terrorism actually been dismantled?

No. The infrastructure has been damaged and embarrassed but not dismantled. The seminaries that recruit, the charity fronts that move money and provide cover employment, and the campuses that combine religious, educational, and operational functions largely continue to operate. The Operation Sindoor strikes did real damage to the Muridke and Bahawalpur complexes, but campuses can be repaired and headquarters relocated. The campaign has been highly effective against individuals and far less effective against the diffuse, hard-to-name support layer.

Why is the support layer harder to reach than the leadership?

Because the campaign depends on identifying specific individuals, and the support layer is largely invisible to that process. Senior commanders are named, designated, and documented; financiers, seminary administrators, logistics operators, and recruiters often are not. A campaign that works by identifying targets cannot reach a person it cannot name. The support layer therefore retains a geographic safety the senior leadership has lost, not because it occupies better-defended ground but because it does not appear on the lists that drive the targeting.

Could the safe haven recover if the killings stopped?

A partial recovery would be possible, but the deterrent precedent would persist. If the campaign paused, surviving leaders could gradually relax some defensive measures and resume more open operation. But the demonstrated fact that no Pakistani city offered reliable refuge would not be erased by a pause. A recruit or a leader weighing the future would still know that the sanctuary had failed once and could fail again. Geography, once shown to be unreliable, cannot easily be made to feel reliable again.

How does this capstone differ from the individual city guides?

The city guides each examine one location in depth: which organizations operate there, which figures lived there, which eliminations occurred there, and how the campaign changed that specific place. This analysis does something the individual guides cannot. It holds the entire country in one frame and reads the killings as a geographic sequence rather than a chronological list, which exposes the pattern of contraction, the city categories the campaign tested in order, and the inverse movement of the expanding red zone and the contracting green zone.

What would a complete collapse of the safe haven look like?

It would not be a map fully colored red, because no campaign can place a threat in every district simultaneously. It would be a condition in which the network itself stopped believing any location offered protection and therefore stopped organizing its life around geography. Recruitment would falter because the career would visibly end badly. Financiers would withdraw because association would be lethal. The leadership would cease to function everywhere, not just in a few cities. The infrastructure would wither for want of a secure environment. Pakistan has moved toward that state but has not reached it.

Did the period after the 2025 conflict make the geography safer or less safe?

Less safe. By several accounts the operational tempo of eliminations rose rather than fell in the period after the May 2025 ceasefire. This cuts in two directions. It demonstrates that the campaign’s geographic reach was not constrained by the formal end of conventional hostilities, which reinforces the contraction thesis. But it also shows that the network had not been reduced to silence, which is a reminder that geographic penetration and strategic destruction are different things.

How is the geography of the Khalistan groups different?

The Khalistan organizations occupy a different map from the Kashmir-focused groups. They maintain a presence in Pakistan’s Punjab, but their center of gravity sits in diaspora communities abroad, particularly in Canada and the United Kingdom. Their story therefore runs partly outside the Pakistani frame examined here, and the dynamics of those Western safe havens, where legal protections rather than state sponsorship provide cover, follow a different logic from the seminary-and-cantonment geography of the Pakistani sanctuary.

Why does the mosque appear so often as the place of a killing?

Because the mosque is the most predictable location in a devout man’s week. Everyone knows where a religious figure will be at prayer time, and that certainty is a gift to anyone planning a strike. Targeting at or near a mosque also signals a depth of local surveillance that no wanted man can easily counter, because the only way to defeat it is to abandon the public religious life that anchors a respected figure’s standing. The pattern is less about religion than about routine, and routine is what the campaign learned to exploit.

What is the grey zone, and why does it matter more than the red zone?

The grey zone is the band of territory that is neither a confirmed strike location nor a reliably safe one: cities with no killing of their own but with enough resemblance to a struck city that safety can no longer be assumed. It matters more than the red zone because it is where most of the network actually lives. A red city has already been hit. A grey city is one a wanted man cannot relax in, only wait in. The campaign’s geographic achievement is best measured by how fast grey consumed the green map, because grey is where the fear operates.

Did the campaign have to kill anyone to shrink the safe haven in a given city?

No, and that is the most important feature of the contraction’s mechanism. A city turns from green to grey by inference, not by experience. When a comparable city nearby is struck, the reason the first city felt safe is destroyed, even though no gunman ever visited it. This is why the safe haven shrank far faster than the raw number of killings would suggest. Most Pakistani cities that lost their sense of safety never recorded a single elimination. They lost it because of what happened somewhere else.

Does the survival of a targeted leader disprove the contraction argument?

No. A leader who survives an attempt, or who is never attacked at all, has still typically been forced into the defensive crouch that the campaign produces: changed residences, abandoned routines, restricted communications, delegated command. A safe haven exists to make that crouch unnecessary. A surviving leader living in hiding is therefore evidence that the haven has failed in its core function, not evidence that it endures. Survival and contraction are fully compatible, because the contraction is measured in lost operational freedom, not in lives.

What is the single most important takeaway about Pakistan’s changing terror geography?

That the safe haven was built on a promise, and the promise has been broken everywhere it was tested. The promise was that location could substitute for security: that distance, density, terrain, or garrison control could keep a wanted man safe without any effort of his own. Five years of targeted eliminations, capped by the conventional strikes of Operation Sindoor, have demonstrated city by city that the promise no longer holds. The infrastructure survives and the network endures, but the geography that once protected them has been turned, decisively, into contested ground.