01 THE SETUP

Last week, Codelco suspended development and construction at Andes Norte, a key sector of its El Teniente expansion, after six months of monitoring flagged a deep-seismic phenomenon its existing models weren’t built to handle — behavior the company called “distinct from the risks historically known and managed in the operation” for decades via panel caving. The measure, while preventive, could set development back two years further compounding ongoing production challenges.

02 WHY IT MATTERS

Copper has one famous supply-side worry: declining ore grades. Miners move more rock for less metal every year, costs creep up, and the marginal tonne gets more expensive. That story is neat, real, well-understood, and thoroughly priced.

However, Andes Norte points at a potential problem that gets far less attention. As the industry’s growth increasingly depends on deeper and more complex mines, a risk shows up that only becomes visible once development or extraction is underway: the rock mass at depth doesn’t always behave how the model predicted.

In other words, while the copper is there and the geology is thought to be understood, there’s still uncertainty as to whether it can be pulled out on the assumed schedule, at the assumed rate and recovery.

03 THE DECODE

Copper forecasting generally runs on a simple de-risking ladder:
resource → reserve → permitted → construction → production. Each rung is supposed to shrink the error bars. And mostly, it does.

However, deep caving mines complicate the logic if a whole category of risk only becomes observable after the fact. You can’t fully validate how a giant rock mass will behave until you initiate a cave, redistribute the stress, and see how it reacts with faults and water at commercial scale. And that produces a split the ladder doesn’t capture — a project can get geologically better understood while staying operationally uncertain.

Andes Norte isn’t a one-off, either. It belongs to a broader family of forecasting blindspots: assumptions that are reasonably well-supported at one stage of an orebody don’t necessarily extrapolate cleanly into the next.

Kamoa-Kakula (DRC)

Kamoa-Kakula in the DRC is one of the industry’s flagship new Tier-1 developments. In May 2025, seismicity and water inflow forced Ivanhoe to suspend underground mining. And it didn’t just repair and resume. The March 2026 technical report rebuilt the geotechnical design: wider pillars, exclusion zones, extraction cut to roughly 60% of resources, output deferred further out.

In short, while the orebody didn’t shrink, the understanding of how safely, and thus, how much of it could be mined did change.

Grasberg (Indonesia)

Grasberg in Indonesia suffered a mud rush on September 8, 2025 at Freeport’s Block Cave which killed seven workers and forced a major suspension. This led Freeport to downgrade 2026 output to roughly 35% below its pre-incident ~770kt estimate, with prior rates not expected back until 2027.

The lesson is asymmetric: a mine can look fully de-risked, sit on a world-class orebody, already be producing at scale — and a single incident can still move hundreds of thousands of tonnes across calendar years. Big caves are efficient precisely because huge volumes move through a concentrated system, which is also why the downside is so large when that system misbehaves.

Oyu Tolgoi (Mongolia)

Oyu Tolgoi in Mongolia might show this isn’t just an isolated 2025-26 run of bad luck. When difficult ground conditions forced a redesign of Hugo North’s Panel 0 — larger pillars, revised layout — the impact was substantial. First, sustainable production slipped 21 to 29 months and development capital rose $1.3-1.9 billion on an original $5.3 billion budget. The mine got there in the end, but the original assumptions proved far less robust than the resource confidence implied.

Spence (Chile)

The complexity isn’t only geotechnical, either. BHP’s Spence in Chile was supposed to be a growth asset after its hypogene expansion. Instead, as it moved into deeper, more complex ore, FY26 copper output fell 21% to 212,600 tonnes, and BHP guides it lower still (210,000-230,000t) before any recovery. To claw that back, the company is now deploying new sulphide-leaching technology (SAL2) and a concentrator upgrade, with first cathode not expected until CY28.

Here, the same basic pattern — what works near surface doesn’t automatically extrapolate downward — plays out, albeit via a different mechanism. Here the problem isn’t seismicity; it’s that deeper ore simply doesn’t process like the ore above it, and closing that gap costs both tonnes and years.

Looking ahead

The forward-looking test case isn’t in Chile or Congo — it’s in Arizona. Rio Tinto and BHP’s Resolution Copper is one of the most important potential future U.S. copper sources, and it sits more than 2,000 metres underground, with rock temperatures near 80°C, intended to be mined by caving.

Of course, there’s no immediately apparent reason why such deep mining can’t work. However, Resolution does represent a meaningful slice of the supply the market already pencils in as a future source. And yet, it faces some of the most demanding engineering in the industry before it becomes dependable annual tonnage.

And that leaves us with a couple of questions. First, how wide should the probability distribution around its ramp-up, rate, and recovery actually be? And second, as the industry leans on deposits this difficult, do aggregate supply forecasts get less certain in any significant way, even as the resource base gets larger?

Either way, copper supply models may be becoming simultaneously more certain about where the metal is and less certain about when and how much of it will arrive.

04 WHAT TO WATCH
 
Now → next quarter: Codelco’s Andes Norte reassessment. The tell is duration and scope — a weeks-long pause with monitoring tweaks is routine; a redesign of pillars, sequencing, or extraction ratios (the Kakula path) signals the problem is structural. Watch whether the language shifts from “preventive suspension” to “revised mine plan.”

H2 2026 → 2027: The Grasberg and Kakula ramp-backs. Do they hit the restated schedules or slip again? Repeated slippage matters more than any single miss — it would mean even the deliberately conservative post-incident assumptions were still too optimistic.

2027 → 2029: Resolution and the next wave of ultra-deep caving projects reaching development decisions. Watch whether the sector starts pricing wider ramp-up distributions for these mines — or keeps applying the same clean de-risking ladder to deeper, more complex deposits.

/ Congo bans copper and cobalt concentrates exports, official order says

The DRC is trying to force more mineral processing inside the country rather than exporting concentrates. The immediate impact appears limited, but the policy direction — producing countries increasingly wanting a larger share of the downstream economics — adds another layer of complexity to future mine development.

/ Copper market crunch brews as US and China compete for metal

Copper flows are being pulled in two directions at once: record shipments into the US ahead of a potential tariff decision, alongside increased shipments to China as buyers there replenish stocks. These competing flows are tightening availability outside the two markets.

/ Chile allows ailing state copper miner to reinvest all profits for first time

Chile will let Codelco retain all $2.42 billion of its 2025 profit, the first time the state miner has been allowed to keep everything it earned. The money is intended to support a recovery plan as Codelco deals with declining output, more than $20 billion of debt and expensive mine-life extension projects.

That's the wire for today. The price is quiet; the ground underneath it isn't.

Know someone who keeps asking you about copper?

They can subscribe free at copperwire.news.