|
Macro · Hard Assets · Commodities
The last basket is not a rock. It is red wire, and nobody is pulling enough of it out of the ground.
Every gigawatt the substation piece described has to travel through several hundred pounds of it to reach a single rack of GPUs. The mines that would supply the next decade's worth take longer to permit than the decade itself.
Wall Street Logic
· 8 min read
Start with the number that makes the rest of this piece unnecessary for anyone in a hurry. A single gigawatt-scale data center campus, the kind now being proposed routinely across the corridors Issue 011 described, requires on the order of ten thousand tonnes of copper to build, threaded through the switchgear, the busbars, the transformer windings, the miles of cabling that carry power from the substation fence to the individual rack. A single offshore wind farm uses roughly eight thousand tonnes of copper per gigawatt of capacity, nearly five times what an equivalent gas plant needs, because wind and solar generate electricity in diffuse, distributed form and copper is the only economic way to gather it back into something a grid can use. An electric vehicle carries roughly two and a half times the copper of a comparable gasoline car, in the motor windings, the battery pack, the charging electronics. None of this is a future requirement. It is being drawn down against existing above-ground supply this year, this month, on top of the roughly twenty-six million tonnes a year the world already consumes to keep the existing economy running. The property piece told you where the next decade's electricity has to be delivered. This piece is about the metal that has to carry it there, and the metal is running out faster than the mines can be built to replace it.
Copper is not a rare metal in the way gold or the rare earths are rare. That is precisely what makes the shortage worth understanding, because it is not a shortage of atoms in the crust. It is a shortage of permitted, financed, producing mines, and those are two entirely different kinds of scarcity that get talked about as if they were the same thing. The world is not running out of copper ore. It is running out of the ten to twenty years it takes, on the most careful published estimates, to take a fresh discovery from a drill core to a permitted, financed, operating mine, a timeline that has stretched rather than compressed over the past two decades as environmental review, water rights litigation, and community consultation requirements have all lengthened in nearly every major copper jurisdiction on earth. A silver deposit can sometimes be brought into production inside five years. A major copper porphyry, the kind of deposit large enough to matter at the scale the electrification build now requires, routinely takes twice that, and the last genuinely world-class discovery of that scale was made the better part of two decades ago. The pen that can print a trillion dollars of stablecoin-backed Treasury demand overnight, the mechanism Issue 006 walked through in detail, cannot compress a fifteen-year permitting timeline into three by decree. It can subsidize a smelter. It cannot conjure an ore body that has not been found.
A mine that takes fifteen years to permit cannot be summoned by the same pen that can summon a trillion dollars of new debt demand by Tuesday.
The demand side of this equation is not a modest, linear trend that the mining industry can quietly grow into over a normal investment cycle. It is a demand shock arriving from four directions that used to move independently and are now moving together for the first time. The electrification of transport is one. Global electric vehicle sales have moved from a rounding error a decade ago to close to a fifth of new car sales worldwide, and every one of those cars needs roughly the copper load of two and a half combustion vehicles. The renewable build-out is the second, and the arithmetic there is the starkest of the four, because a unit of electricity generated by wind or solar and delivered to a home or a data center requires several times the copper per unit of a unit generated by a gas turbine sitting next to the load it serves. The grid rebuild is the third, the one Issue 011 already introduced, an American transmission and distribution network built largely in the middle of the last century now being asked to absorb a demand curve nobody on any utility's planning staff was trained to expect, and every mile of new transmission line, every new substation transformer, every upgraded feeder into a growing suburb, is copper that has to be pulled from somewhere before it can be strung. And the fourth is the one this letter has spent three months documenting directly, the data center build itself, which the most careful industry estimates now suggest could account for a meaningfully larger share of incremental global copper demand by the early 2030s than it does today, a share that barely existed as a line item in any forecast five years ago.
Set against that four-sided demand shock is a supply picture that the industry's own analysts describe, in language usually reserved for far more dramatic commodities, as a structural deficit rather than a cyclical one. The ore grades at the world's largest existing copper mines, the deposits that supply the bulk of the world's refined metal, have been declining for years, meaning more rock has to be mined and processed to produce the same tonne of copper, which raises costs and lengthens mine lives in the wrong direction. Chile and Peru together supply somewhere in the region of two-fifths of the world's mined copper, a concentration risk that ought to sound familiar after Issue 010's treatment of rare earths, except that in copper's case the risk is not a single government's export license but water scarcity in the Atacama, community opposition, and ore grades at flagship mines like Escondida that have fallen for years running. The Democratic Republic of Congo has emerged as the fastest-growing major source of new supply, principally from a small number of exceptionally rich deposits, and every serious forecast of the 2030s copper balance now leans on continued expansion from a jurisdiction whose political and infrastructure risk profile is, to put it gently, not the kind a utility planner enjoys underwriting a multi-decade grid buildout against. And the smelting and refining step, the part of the chain that turns ore concentrate into the metal an electrician can actually use, is now dominated by China to a degree that would be the headline of this piece if rare earths had not already claimed the word chokepoint for themselves this year. Chinese smelters process something in the order of half the world's mined copper concentrate into refined metal, a dependency that sits quietly underneath every American grid upgrade the way the rare earth magnet sits quietly underneath every F-35 control surface.
Every projection of the resulting gap tells roughly the same story with different decimal points attached. Industry consultancies that model global copper supply and demand out to the middle of the next decade have, in their more sober published work, put the cumulative shortfall between what committed mine projects will produce and what the electrification pathway requires somewhere in the range of several million tonnes a year by the early 2030s, widening further if the data center build proceeds anywhere near the pace this letter has been documenting since Issue 011. Recycling, the answer most people reach for instinctively, genuinely helps and genuinely is not enough on its own, because scrap supply is mechanically bound by how much copper was installed decades ago and is only now reaching end of life, a figure that grows slowly and predictably while the new demand curve grows anything but. The market has, in its own quiet way, already begun pricing part of this in. Exchange-tracked copper inventories across the major global warehouses have spent stretches of the past two years sitting near multi-year lows, even as the price has periodically traded at record levels, the same divergence between thin visible stock and firm physical price that Issue 006 flagged in gold's paper market, except here the tightness is industrial rather than monetary, and there is no central bank vault of unused copper sitting off the books waiting to be revalued the way Issue 009 described for gold.
It is worth pausing on why this basket earned the label the consensus has stopped watching, because copper is not an obscure metal nobody has heard of. It trades on every major exchange, it has a nickname on every trading desk, and every macro strategist alive has an opinion about what it says about the global business cycle. That familiarity is exactly the problem. Copper has been priced for two generations as a cyclical bellwether, a metal whose price rises and falls with the health of Chinese construction and global manufacturing, the so-called Doctor Copper read on the economy. Investors who have watched that pattern for a career look at today's price and read it through the old lens, a China property slowdown here, a manufacturing PMI wobble there, and conclude the metal is behaving normally for the stage of the cycle. What that lens misses entirely is that the demand base underneath the price has quietly changed character. A meaningful and growing share of copper demand no longer comes from the cyclical, PMI-sensitive part of the economy at all. It comes from a structural, multi-decade electrification build that does not pause when a factory survey disappoints, because a data center campus under construction, a transmission line already permitted, or an EV plant already tooled up does not stop pulling copper because Chinese cement output softened for a quarter. The old framework says watch the business cycle. The new demand base says the business cycle is no longer the whole story, and the investors still trading the old framework are the ones calling this metal boring at the exact moment its underlying buyer base has been rewired.
This is where the basket rejoins the rest of the series, and it does so more directly than any of the other three. The metal cannot be printed, which is the same line that opened the silver piece in Issue 005, except silver's scarcity is largely monetary, a matter of an agreement about what counts as reserve-worthy, while copper's scarcity is purely physical and industrial, a matter of rock in the ground and years on a permitting calendar, neither of which bends to a central bank's decision about interest rates. The mine cannot be summoned any faster than the property in Issue 011 can conjure a faster transformer delivery schedule, because both are gated by the same real-world constraint, the years it takes to turn a regulatory approval into poured concrete and, in copper's case, into a functioning concentrator and smelter circuit besides. And the demand pulling on the metal is not a speculative story invented by a newsletter. It is the same four-sided buildout, transport, generation, grid, and compute, that every previous issue in this run has already documented from a different angle. Property needs the substation. The substation needs the transformer. The transformer, and every mile of cable feeding it, needs the copper. Four issues, one thread, arriving at the same physical bottleneck from four different directions.
None of this is a guarantee the price behaves in a straight line, and the discipline this letter has argued for in every prior issue applies here without exception. Copper is a genuinely cyclical commodity in the short run even as its structural floor rises, which means a real growth scare, a sharp Chinese slowdown, or the same kind of hawkish rate shock that hit gold and silver in Issue 008 can and likely will knock the price down hard at some point in this build, for reasons that have nothing to do with whether the electrification thesis is correct. That is not a reason to avoid the basket. Issue 008 already wrote the manual for reading that kind of selloff correctly, and the manual does not need rewriting here: the price answers to the cycle in the short run, the value answers to the multi-decade demand curve, and the two are not the same thing on any given Tuesday. The entry that has mattered most across every basket in this series has never been the parade. It has been the washout the market mistakes for a verdict on the thesis rather than a verdict on the quarter.
Which closes the run this letter opened eleven Mondays ago. Issue 004 named four scarce assets the pen cannot rewrite: the metal, the property, the toll, and the industrial commodity the consensus had stopped looking at. Silver ran in Issue 005. The toll booth closed the first sprint in Issue 007. The substation and the fenced field ran last week. And the fourth basket, the one hiding in plain sight inside every other issue's argument without ever getting its own headline, is the wire that connects all three of the others to the wall socket. It is not exotic. It is not a new discovery. It is the oldest industrial metal humans ever learned to mine, sitting under a shortage that has nothing to do with scarcity in the crust and everything to do with how long it takes to turn permission into production, at the exact moment four independent demand curves decided to arrive at the smelter gate together for the first time in the metal's long history.
◆ ◆ ◆ |
