In the early hours of 3 September, a group of suspected fighters from the Islamic State-linked insurgency in northern Mozambique crossed into a province they had barely tested in more than a year. ACLED puts the group at around 30. They struck a hunting camp run by Mozambique Wild Adventures in the buffer zone of the Niassa Special Reserve, at Marangira in Marrupa district. That is in the province next door to Cabo Delgado, home to Mozambique’s flagship $20 billion gas project. By the time the smoke cleared, a police post, a health centre and a mobile phone mast nearby had also been set alight.
Mozambican police confirmed at least one person killed and eleven abducted, some of them children. Other reports, including ACLED’s, put the death toll at two. South African officials said one of their nationals was among those taken. They referred to as many as fifteen to twenty abducted workers. Around 2,700 people were displaced. Three American hunters were evacuated by air.
By the standards of this war, it was a modest attack. The conflict has killed more than 6,600 people and displaced over a million since 2017. But this attack landed in a province that had seen almost nothing like it in sixteen months. It has forced an uncomfortable question back into view. After eight years and sustained international security assistance, what has Mozambique’s counterinsurgency actually secured: the country, or an asset inside it?
A war built around one address
The insurgency began in Cabo Delgado in October 2017. It is waged largely by fighters known locally as Ansar al-Sunnah and affiliated with Islamic State Mozambique Province. Its defining moment came in March 2021. More than a hundred fighters overran the town of Palma, just over 10km from the Afungi peninsula construction site for TotalEnergies’ Mozambique LNG project. TotalEnergies declared force majeure the following month and evacuated the site. What followed was not merely a counterinsurgency campaign but a rescue operation for a specific piece of infrastructure.
Rwandan troops deployed to Cabo Delgado in July 2021 at Maputo’s request. They served alongside a shorter-lived SADC mission (SAMIM), which withdrew in 2024. Within weeks, Rwandan forces had helped retake Palma and Mocímboa da Praia, the two towns most directly tied to the gas corridor.
Dr. Egna Sidumo,an analyst writing in Third World Quarterly this year has described the effect in blunt terms. In their account, military stabilisation functioned in practice to protect what they call “extractive enclaves” rather than the province as a whole. A CSIS assessment reached a similar conclusion in softer language. It noted that the intervention reduced insurgent access to the maritime space around Cabo Delgado’s northern LNG corridor.
Rwanda’s role has since deepened, though not without friction. Its deployment is now estimated at 4,000 to 5,000 soldiers and police. ISS estimates that around 3,000 of them are assigned to securing the LNG project and the nearby towns of Palma and Mocímboa da Praia.
On 27 August 2025, during President Daniel Chapo’s visit to Kigali, the two countries’ defence ministers signed a Status of Forces Agreement while Chapo and Paul Kagame looked on. It was the first public legal framework for a deployment that was then already four years old. SOFAs typically govern extended foreign deployments, though Chapo insisted this one did not amount to a new military alliance. The agreement has fed speculation about a longer-term Rwandan footprint, including claims that Kigali could eventually establish a permanent base. Neither government has confirmed this.
That footprint has also proved negotiable. In March 2026, the EU’s €20 million in support for Rwandan forces was due to end in May with no renewal planned. Rwanda then threatened to withdraw. By May, Kigali said Maputo had secured the funding itself, on terms that have not been made public.
That security architecture was the explicit precondition for what came next. On 29 January 2026, TotalEnergies chief executive Patrick Pouyanné and President Chapo stood together at a ceremony in Afungi. There they announced the full restart of Mozambique LNG, nearly five years after it was frozen. The company’s own announcement noted the government’s commitment to “continued cooperation with Rwanda.” Contemporary reporting, including Al Jazeera’s, attributed the restart to improved security, particularly the Rwandan soldiers deployed around the Afungi site. By late April, more than 6,000 workers were on site and the project was 42% complete. By late July it had reached 45%, with first gas still targeted for 2029.
Notably, that improved security has a precise geography. Pouyanné reportedly pushed for the Afungi peninsula to become something closer to a fortress than a construction site, reachable only by air and sea, with no land routes in or out. The company’s own security planning, in effect, concedes that securing Afungi and securing Cabo Delgado are two different undertakings. The insurgency has kept attacking elsewhere in the province throughout 2026, with conflict incidents in the first half of the year up 10% on the same period in 2025, even as the flagship project has moved forward on schedule.
Mobility, not territory
Niassa is where that distinction becomes hardest to ignore. The province recorded its first insurgent incursion in November 2021, but serious attacks have been rare. In April 2025, two attacks inside the Niassa Special Reserve killed ten people in total, including soldiers, rangers and camp workers. The first was on the Kambako hunting camp on 19 April, and the second on the Mariri Environmental Centre on 29 April (Cabo Ligado). ACLED describes the Marangira attack as ISM’s first foray into Niassa since May 2025.
Peter Bofin tracks the conflict closely as an analyst with the Armed Conflict Location & Event Data project (ACLED). He cautions against reading the attack as a deliberate expansion of ISM’s operating range. Speaking to this reporter, he noted that the group has been in Niassa before. It went there briefly in 2021 to evade Rwandan operations, and again in April and May 2025, when it attacked the Kambako hunting camp.
What looks different this time, he said, is how the group arrived. In the past it moved into Niassa undisturbed. This time it was “hotly pursued by the Mozambican military” in Montepuez district, on the Cabo Delgado side of the border. “It is possible,” Bofin said, “that the group that went to Niassa did so to avoid the Mozambican military in Montepuez.” That would make the September attack less a strategic pivot than a byproduct of pressure applied elsewhere. Niassa police offered a similar reading, suggesting the attackers may have been seeking supplies after being cornered by security operations.
In Bofin’s assessment, Niassa’s relative safety has less to do with state security provision than with the terrain itself. “Niassa is not so well protected militarily,” he said. But the province’s dense network of conservation and hunting operations makes it hard for the group to establish itself, with rangers, light aircraft and helicopters on the ground. Its sparse population also limits opportunities to loot supplies or build the informal supply lines the group relies on around small urban centres elsewhere. ACLED’s latest monitoring update makes the same point: security forces themselves depend on the air assets and trained rangers of tourism and conservation operators. In other words, Niassa’s insulation from the insurgency so far has been closer to an accident of geography and private conservation infrastructure than a product of Mozambique’s counterinsurgency strategy.
That distinction matters for how the story should be told, and for how Mozambican and international security planners ought to respond. This is not evidence that the insurgency has expanded into a second province in any territorial sense. It is evidence that state security beyond the Cabo Delgado gas corridor is thin. A group fleeing military pressure in one district can cross into an entirely different province with comparatively little resistance. Niassa’s Secretary of State, Silva Livone, himself said the insurgents’ westward movement had been monitored for three months before they entered Marrupa undetected. Where Niassa has held, it has largely held despite the state’s footprint, not because of it.
A second resource frontier, inside Cabo Delgado itself
Bofin’s account also points to a development that complicates any simple story about geographic expansion: the group’s growing interest in Montepuez district, the corridor through which the Marangira attackers travelled. Montepuez is home to numerous artisanal and small-scale gold mining sites. Bofin said ISM has increasingly tried to extort money from miners there, and he expects that pattern to produce more frequent incursions into the district.
That detail is worth sitting with. I asked Bofin directly whether ISM’s growing interest in Montepuez’s gold miners reflects a group redirecting toward the province’s next most valuable, least protected economic activity. He confirmed the connection in his own terms. ISM fighters “are mostly active away from Palma where the LNG plant is being developed,” he said, because “the main objective of the Rwandan forces is to protect that project, so there are more opportunities for ISM in districts to the south and southwest.” In its current configuration, he added, the group “can’t really threaten the LNG plant” at all.
That is a significant admission from an analyst who tracks the conflict closely: the Rwandan deployment’s success is, by design, geographically narrow. Bofin also explained why that redirection matters less in military terms than it might in economic ones. Since the Rwandan intervention, ISM has shrunk to a maximum of roughly 500 fighters. It has not developed more sophisticated battlefield capabilities, such as the drones other Islamic State affiliates have adopted elsewhere. It is not, in other words, a resurgent force capable of threatening the state directly. But “the presence of significant gold resources across the south of the province presents them with a real opportunity to sustain themselves,” Bofin said. The result is a small, degraded group that no longer threatens Mozambique’s most valuable asset but remains entirely capable of preying on the people and resources the state has chosen not to protect.
Artisanal gold miners in Montepuez have none of the security architecture built around Afungi: no Rwandan battalions, no Status of Forces Agreement, no fortress design. They are the same “protected enclave” problem in miniature. Wherever the state and its partners have not chosen to build a perimeter, the insurgency has room to operate, extort and resupply.
The same logic, a second frontier
A third, less-noticed thread runs through Niassa itself this year. On 8 June 2026, Livone announced the suspension of the management of the Niassa Special Reserve, Mozambique’s largest protected area. The suspension covered its co-manager, the Wildlife Conservation Society. Livone accused them of failing to curb human-wildlife conflict and installed an interim team to be accompanied by police and security contingents.
The suspension did not last. Within about two weeks, a multisectoral team from central government lifted all of Livone’s measures. But it paired that reversal with plans to resize the reserve and restructure its co-management model. The row played out amid a broader debate over Mozambique’s new mining law, which critics say widens the scope for extractive activity inside protected areas. Zitamar argued at the time that the episode reflected a wider shift, in which conservation areas are increasingly framed as obstacles to development, mining and local control.
Read alongside the September attack, the pattern deserves attention, even if it is not evidence of design. In Cabo Delgado, a security crisis became the justification for building a walled-off enclave around a gas project. In Niassa, a dispute over human-wildlife conflict was briefly used to strip management authority from the country’s largest conservation area. Maputo overruled the suspension, but the episode ended with a commitment to shrink the reserve. That reserve sits on land of interest to miners, in a province where the state’s actual security presence, as the Marangira attack showed, is thin. In both cases, the language of protection has been used to reorganise control over valuable land, while the communities living on it are left more exposed, not less.
Whose security
None of this means Mozambique’s counterinsurgency has failed on its own terms. Rwandan and Mozambican forces genuinely degraded ISM’s fighting capacity after 2021, and Afungi’s workers are safer today than they were in 2021. But “safer” has turned out to be a narrow and portable concept. It can be built around a gas terminal. And provincial officials elsewhere have shown they are willing to invoke it selectively, in service of very different interests.
The question the Marangira attack leaves hanging is not whether Mozambique can eventually defeat the insurgency. It is whether the country’s security architecture, heavily shaped by what international partners and investors consider strategically important, is capable of protecting a region rather than an asset within it. Consider the roughly 2,700 people displaced from Marrupa district this month. Consider, too, the communities of the Niassa Special Reserve, whose conservation area is now slated to be resized. For them, the answer so far looks like no. That is a warning worth taking seriously well beyond Mozambique, wherever African governments are asked to organise security around the infrastructure international partners value most, rather than the populations living alongside it.
Kevin D. Mofokeng is a writer and civic leader dedicated to using impactful storytelling to amplify African voices and advocate for social justice and development.
