Latin America put a price on above-ground risk. $54 billion, and the AI buildout pays it in mines that never start
A risk consultancy told a September 24 webinar that permitting, community conflict and security have cost or frozen more than $54 billion across the region since 2018. Strip out one 2015 tailings settlement and the documented losses fall to $6.3 billion. The number that matters is not the total. It is where capital is going instead.
The AI buildout needs copper that does not exist yet, and the region holding the best undeveloped copper geology has just been handed a bill for the projects it did not build. Americas Market Intelligence put the figure at more than $54 billion since 2018 [MINING.COM, 2026-09-29]. The total is arresting, and it is also the least useful thing in the study. What matters for anyone modelling copper supply into the 2030s is the behavioural change the consultancy describes underneath it: miners are quietly abandoning greenfield development in favour of expanding what they already own.
That is a rational response to risk. It is also the wrong shape of supply response for a demand shock that needs new mines, not bigger versions of old ones.
What’s happening
- Americas Market Intelligence (AMI) estimates above-ground risk has cost or stalled more than $54 billion in Latin American mining since 2018. Roughly $38 billion is write-offs, legal settlements and fines, while $16 billion is capital frozen in delayed or halted projects [MINING.COM, 2026-09-29].
- Sebastián Pérez-Ferreiro, co-director of AMI’s mining and risk practice, gave the breakdown during a webinar on September 24, titled “Avoiding the Costs of Above-Ground Risks in Latin American Mining” [Americas Market Intelligence, event listing]. The MINING.COM write-up appeared five days later. The webinar is the event, not the reporting date.
- Of the frozen capital, about $7 billion sits in Peruvian copper projects and about $4 billion across three Mexican projects: El Arco in Baja California, San Nicolás in Zacatecas and Cordero in Chihuahua [MINING.COM, 2026-09-29].
- The Mining Conflicts Observatory currently lists 284 conflicts involving 301 projects across the region, according to AMI [MINING.COM, 2026-09-29].
- Copper traded at $6.58 per pound on September 29, roughly 36% higher than a year earlier, having set an all-time high of $6.85 earlier in September [Trading Economics, 2026-09-29; these are CFD reference prices, which Trading Economics states are not official benchmark prices]. That is roughly $14,506 per tonne at the standard 2,204.62 pounds per tonne, desk arithmetic.
What the $54 billion actually contains
One line dominates the tally. The largest single loss is the 2015 Mariana tailings dam disaster in Brazil, where BHP, Vale and Samarco agreed with Brazilian authorities to a settlement valued by AMI at $31.7 billion [MINING.COM, 2026-09-29].
Run the arithmetic. Mariana is 58.7% of the entire $54 billion and 83% of the $38 billion documented-loss bucket. Strip it out and every other write-off, fine and legal settlement across eight years and an entire continent comes to about $6.3 billion (desk arithmetic on AMI’s published figures).
This is not a criticism of AMI, whose own framing is careful. It is the reason the headline number should not be read as a permitting scorecard. A catastrophic dam failure and a stalled environmental licence sit in the same taxonomy but carry completely different lessons for anyone working out when new tonnes arrive.
Juan Carlos Guajardo of Plusmining Consulting made the same distinction to MINING.COM, calling the figure “broadly consistent with the scale of disruption we see across Latin America” while noting he had not independently verified the calculation and that losses already incurred should be separated from capital tied up in delays [MINING.COM, 2026-09-29]. Geopolitical mining specialists Eduardo Zamanillo and Marta Rivera went further, cautioning against reading the total as a measure of the region’s investment appeal and describing a region entering a new mining cycle rather than one uniformly losing attractiveness [MINING.COM, 2026-09-29].
The frozen $16 billion is the number with predictive content. That is capital with a project attached, waiting.
The retreat to brownfield
Guajardo’s operational observation is the part that should worry anyone pricing copper into the AI buildout: miners increasingly favour expansions around existing operations, where infrastructure and community relationships are already established, over large greenfield projects with longer and less predictable routes to production [MINING.COM, 2026-09-29].
Brownfield expansion is cheaper, faster and lower risk. It is also capped. You cannot expand your way to the step change in supply that data centre electrification implies, because the increment is bounded by the orebody you already drilled. Every year that above-ground risk pushes the industry further toward incremental expansion is a year subtracted from the greenfield pipeline that would deliver tonnes in the mid-2030s.
Juan Ignacio Guzmán of GEM Mining Consulting framed the mechanism plainly: delays can matter more than direct legal or security costs, because they push cash flows further into the future and erode project value [MINING.COM, 2026-09-29].
Brazil angle
Brazil’s largest single entry in the tally is Mariana, a tailings failure rather than a permitting or community-consent fight. That is a meaningfully different risk profile from the Andean copper belt, and it cuts both ways. Brazil does not appear in AMI’s frozen-capital breakdown alongside Peru and Mexico. It also carries the single largest liability in the entire dataset.
Brazil has pursued a critical-minerals strategy [MINING.COM, 2026-09-29]. This desk covered Lei 15.506 and the CIMCE decree on September 17 and CBMM’s Araxá oxide expansion on September 23. Notably, AMI’s own research shelf carries a September 16 piece titled “Brazil’s New Critical Minerals Law: Sovereignty Ambitions Meet Investor Uncertainty” [Americas Market Intelligence, 2026-09-16]. When the consultancy tallying the region’s losses files Brazil’s remedy under investor uncertainty, that is a signal about how the new law reads from the capital side, whatever its merits as industrial policy.
US angle
Washington’s copper problem is that the tonnes it wants are in jurisdictions its policy tools do not reach. The immediate US lever is trade, not supply: Trading Economics market commentary notes that large volumes of refined copper have been stockpiled in US warehouses on expectations of an import tariff, and that the administration has not yet decided whether to impose one [Trading Economics, 2026-09-29]. Stockpiling refined metal does not add a single tonne of mine supply. It relocates inventory and tightens everyone else’s market.
China angle
The single most blockaded asset in the dataset is Chinese-controlled. Las Bambas in Peru, which AMI cites as having faced more than 700 days of blockades along its transport corridor since commercial production began in 2016, was acquired by MMG (HKEX: 1208). An early plan to move concentrate by pipeline to Cusco was abandoned after that acquisition, with concentrate trucked through communities along the route instead [MINING.COM, 2026-09-29]. This desk reported in August that the mine is wholly Chinese-owned. AMI cites estimates from Peru’s mining union putting stoppages at about $9.5 million a day. John Price, co-director of AMI’s mining and risk practice, drew the causal line: “If you do not take the time to really engage with the community, and if you do not fulfil the promises that you make, it will come back to bite you.”
On the demand side, Chinese industrial profits rose 15.7% in the first eight months of the year, slowing from a 17.6% rise over January to July, which has weighed on the copper demand outlook even as physical markets show tightness [Trading Economics, 2026-09-29].
What it means
The desk’s AI materials index sat at 99.7 as of September 26, with the materials sub-index at 97 and power at 107.9 [Tantalum indexes, as of 2026-09-26]. Materials have been the weaker leg, and this study explains why that weakness is not the same as slack. Physical tightness and soft sentiment coexist when the constraint is not demand but the industry’s willingness to commit capital to new ground.
The AMI figure is best read as a revealed preference rather than a loss statement. Miners have priced eight years of permitting reversals, blockades and security failures, and the price they set is a discount on greenfield Latin America large enough to redirect capital into incremental expansion. That discount is the supply curve the AI buildout will actually meet.
What to watch
- Centinela mediation, first half of October. Two unions at Antofagasta’s Centinela mine rejected the company’s contract offer on Monday, September 28, with 98.73% of participating members voting to strike. A mandatory five-day government-led mediation follows, extendable by five more by mutual agreement, before a strike can legally begin. Centinela produced 240,400 tonnes of copper in 2025 [Reuters via MINING.COM, 2026-09-28]. This is above-ground risk becoming a live tonnage question within days, not a retrospective tally.
- Quebradona’s environmental impact assessment, 2027. AngloGold Ashanti has written off $98 million on the stalled Colombian copper-gold project and is targeting 2027 for a new assessment [MINING.COM, 2026-09-29]. Whether that filing lands on schedule is a clean test of whether Colombian permitting has actually changed.
- Argentina’s RIGI conversion rate. Argentina approved the Vicuña copper-gold-silver project under its large-investment incentive regime this year [MINING.COM, 2026-09-29]. RIGI is the region’s most explicit attempt to buy execution certainty with fiscal terms. The signal is not further approvals but whether approved projects break ground on the timelines the regime promises.
