01 THE SETUP

Yesterday, BHP won approval to start a $15bn copper expansion in Chile. Meanwhile, copper majors and juniors alike are pouring capital into brownfield expansions, throughput upgrades, debottlenecking, and restarts of long-dormant mines.

The reflex read is simple — high prices call forth new supply. However, the near-term picture is messier than “structural deficit” headlines suggest: in April the ICSG actually flipped its 2026 balance to a ~96,000t surplus, reversing an earlier 150,000t deficit call after trimming demand growth on Middle East and trade-flow uncertainty.

But the interesting story here isn’t the wave of spend, or a messy short-term balance. It’s the shift in how the industry underwrites new supply — one that may make the response more fragile, more correlated, and more procyclical than the capex figures imply.

02 WHY IT MATTERS

Sort the supply response by how much net new metal it actually adds, and it is conspicuously bottom-heavy — grade defense, throughput protection, restarts, and brownfield work.

There’s the tension the market is actually trading. Near-term, the refined balance may even be in modest surplus. Long-term, the pipeline is structurally short.

03 THE DECODE

The question isn’t why miners are building — it’s how much of the marginal pipeline survives the boardroom math. The answer sits in the price assumptions buried in feasibility studies.

The mechanism:

Mining has always stress-tested new projects against a deliberately conservative price — a buffer below spot, so a mine can survive a downturn and still earn its return. That buffer was the discipline that quietly killed marginal projects. But per new S&P Global research across 573 copper studies, it has thinned dramatically — from a historical discount of 30-50% toward single digits, with developers now comfortably modeling five-figure base-case copper.

Of course, the buffer collapse doesn’t explain every project. BHP doesn’t need a thin price deck to justify sustainment at Escondida — you spend to defend a crown jewel regardless. But it helps explain why the marginal edge of the pipeline — restarts, lower-grade expansions, higher-capex brownfields — can suddenly look financeable at the same time. The price print gets the headlines; the price assumption determines how much of that marginal pipeline clears.

The part investors underweight:

The old buffer made projects individually robust. As buffers thin, a growing slice of the pipeline is underwritten against a similar, elevated band. That converts a set of independent bets into something closer to one shared bet on price.

And the conservatism is procyclical. S&P finds studies anchored 20–31% above spot during the 2015–16 downturn — embedding trough prices makes projects look uneconomic — but tracking at-or-near spot in bull markets. The industry is least cautious exactly when prices are highest. And part of today’s price is arguably froth: U.S. tariff-related stockpiling has pulled metal and tightened spot in ways that may not persist.

So the loop:

Of course, the loop is not guaranteed; boards, lenders, cost inflation, and permitting can all break it. But as long as spot validates the price decks, the industry has less reason to rebuild the old buffer.

With that said, there is a counter to the above. S&P notes that, at prevailing prices, current conservatism may actually understate project economics if the structural-demand thesis is right and the IEA’s deficit is real. Whether thinner buffers are reckless or rational, in S&P’s framing, depends on the sustainability of the current price environment.

04 WHAT TO WATCH
 
Now → H2 2026: Watch sanctioning announcements. Each greenlight modeled above ~$10,000/t widens the pipeline’s shared exposure. The real-time tell is the gap between study assumptions and spot — S&P clocked 2026 assumptions at $10,647/t against a ~$12,970/t LME average, an 18% discount that flatters a still-thin buffer.

The trip wire: If copper corrects toward $8,500–9,000/t, projects approved above $10,000 base cases face synchronized margin squeeze and write-downs. The likeliest triggers are the two soft spots: an unwind of U.S. tariff stockpiling, and weakness in Chinese demand. The ICSG’s downgrade already pinned the near-term surplus on softer-than-expected usage.

2027–2029: The reflexive payoff. If price holds, aggressive base cases get validated and more marginal projects sanctioned. If it breaks, a wave of write-downs and mothballing hits a pipeline already delivering thin net-new tonnes. That would then re-tighten the market, setting up the next leg in a classic capital-cycle whipsaw, with the amplitude turned up because the damping has been trimmed out.

The real wager:

If electrification and grid buildout are strong enough, today’s thinner buffers will eventually look rational.

If, however, the price breaks first, the industry will discover that it sanctioned a lot of marginal supply on the same assumption at the same time. Copper may still be structurally short in the long run, but the path could get a lot more violent.

/ LME metals whipsawed by war and peace in first half of 2026
War premium and peace relief made for a messy first half across LME metals. Copper’s version had an extra twist: macro demand worries on one side, sulfuric-acid and supply-chain frictions on the other.

/ Once-Great Copper Miner Faces Reckoning Just as AI Demand Soars
Chile’s state-owned Codelco is struggling under $25 billion of debt, its lowest output in 28 years, and a series of controversies. This has prompted concerns the model will fail to capture the AI and the energy transition boost.

/ The copper the world needs above ground: tailings reprocessing lens for mine planners
Reprocessing a claimed 282 billion tonnes of already-mined copper tailings is emerging as a faster, potentially bankable supply source that also cleans up legacy pollution — a pathway majors like BHP are already testing.

That’s the wire for today. Until next time, don’t let spot do all the thinking.

Know someone who keeps asking you about copper?

They can subscribe free at copperwire.news.