Automation comes with risks and challenges. The operations profiled in this special report include at least one outright failure and one recovery that took far longer and cost far more than anyone initially expected, and both are more instructive than the smoother stories precisely because they show where the real constraints sit.

Geology does not care about your fleet management software

Kamoa-Kakula's dual vendor strategy, splitting equipment between Sandvik and Epiroc specifically to avoid depending on a single supplier, worked exactly as designed. Neither company's fleet was ever the bottleneck during the mine's recovery from the seismic flooding that shut down underground operations in May 2025. What that strategy could not protect against was the flooding itself, a geotechnical and hydrological event that pushed the company's production guidance through three separate downward revisions across ten months, from an initial 600,000 tonne target withdrawn entirely, to 380,000 to 420,000 tonnes by December 2025, to 290,000 to 330,000 tonnes by March 2026. Vendor diversification is a genuine hedge against supply chain risk. It is not a hedge against the physical reality of a flooded underground mine, and no automation strategy currently available solves for that category of risk.

Automation has to fit the mine, not the other way around

Syama's underground automation program with Sandvik was scrapped after the automated system proved incompatible with how the mine's sub level caving design was actually being built. That failure was expensive and took years to fully unwind, and it happened despite both companies publicly describing the original 2018 partnership in the most ambitious terms available at the time. The mine's eventual recovery came from returning to conventional operating discipline: rebuilding a processing plant that was running at only 80 percent availability, reworking the mine plan, and funding the fix through an equity raise the company's chief executive described as genuinely painful for existing shareholders rather than dressing it up as a minor step. Any mine evaluating an automation vendor's pitch should treat Syama as the standard question to ask before signing: has this specific system been proven against a mine design and geology that actually resembles ours, or only against the vendor's flagship reference site elsewhere in the world.

Country risk does not always mean what the headline suggests

Resolute's 2026 disruption at Syama was caused by road insecurity linked to broader security challenges in parts of Mali, which delayed equipment deliveries and forced the sulphide mill onto lower grade stockpiled ore for several weeks. The company's chief executive, Terry Holohan, pushed back directly on the assumption that instability anywhere in a large country automatically applies to a specific mine site, noting that Syama sits in the far southwest corner of Mali, against the Côte d'Ivoire border, geographically distant from the areas driving national headlines. That distinction matters for financiers and mine managers assessing country risk from outside a jurisdiction: a national security headline and a site specific operational risk are not automatically the same thing, though they are frequently reported as if they were.

The labour question is real, but it runs in both directions

Caterpillar's own framing of its CES 2026 autonomy push rests on a labour shortage that is well documented rather than assumed, 349,000 net new construction workers needed in the United States in 2026 alone, with 92 percent of firms reporting difficulty hiring qualified workers. But automation's effect on that shortage is not simply subtraction. Nevada Gold Mines' shift from five operators per truck to one controller per shift changed the shape of its workforce rather than only its size, and Hitachi's remote operation push in Botswana was designed explicitly to help less experienced operators work safely rather than to remove operators from the equation. Any change management plan built around automation needs to budget for retraining and new technical roles as seriously as it budgets for the equipment itself, since the labour question does not resolve on its own once the machines arrive.