JNIM’s Mali War Has Moved West Toward Senegal and the Gold Belt, and the Data Predates the Headlines

The Africa Center for Strategic Studies, working from ACLED-derived data over a twelve-month comparison, published on 6 May that nearly 20% of JNIM violent activity in Mali fell in the west and south, against 8% of violent episodes the year before, with fatalities in those regions doubling to more than 450 deaths. The opening move it identifies is specific and dated: seven simultaneous attacks by JNIM’s Maçina Liberation Front near the Senegal and Mauritania borders on 1 July 2025. The same analysis describes the coalition working the Bamako–Ségou corridor along the Niger River, with Diafarabé and Léré in the inner delta and other central and northern towns under blockade, leaving populations dependent on JNIM for goods. It flags the fall of Farabougou as significant.

The economic layer is thinner-sourced and should be read as such. The Africa Report’s Mali index page, crawled around 30 July, states that violence has spread across the mining belts of western and southern Mali, that several Chinese-run operations have been hit since 2025, and raises the possibility that JNIM is extracting resources and forcing deals to tap the gold economy. That is an outlet summary rather than a filed investigation, and no figure attaches to it. Nothing retrievable covers the Barrick Loulo-Gounkoto dispute, B2Gold’s Fekola, or 2026 Malian gold export volumes. The reachable numbers describe violence; the claim about revenue capture remains a claim.

Assessment: Two bars — 8% then 20% — describe a war relocating toward Mali’s solvent half. Kidal is a symbol; Kayes is where the customs receipts and the gold are, and where the roads to Dakar, Nouakchott, Conakry and Abidjan begin. That is the same geography the fuel blockade attacks, which suggests one campaign rather than two. Distrust the gold-revenue framing: it is the most consequential assertion in the file and the least documented, and it circulates because it is intuitive, not because anyone has shown the accounts. Note also the vintage — the strongest dataset here is nearly three months old, and no one has updated it publicly since.