The copper that matters is the copper still for sale. Two governments are taking tons off the table. Everyone else is left to bid for what remains. That is the case for exposure. It is not a case for a thrill.
On September 28, 2026, Daniel Ghali, Deutsche Bank’s head of metals research, put a hard edge on an old worry. Available copper inventories, he said, have fallen to unprecedented lows. U.S. and Chinese stockpiling is squeezing everyone else. In his framing, the price rallies to about $22,050 a ton by the second quarter of 2027. He called it the most acute copper scarcity on record, and a de-globalization endgame.
Investors need exposure to copper because a ton locked in a strategic pile cannot wire a house, fill a shell, or feed a data center. Price is how the rest of the world rations what is left. If you do not have a claim on that metal, you are the person who finds out last that the warehouse door is shut.
This is an argument, not an order. A bank target is not a promise. Copper can fall. Exposure can lose money. The theme is still the theme. The free pile is the asset. The headline is not.
One theme
Say it in one line. Investors need exposure to copper because the tons still free to trade are being absorbed by two national stockpiles, and the only peaceful way to free them is a price high enough to kill some demand.
Everything else in this piece has to serve that line. The 50 percent rally. The war chart. The tungsten warning. The data-center story. If a fact does not change the case for owning a claim on copper, it does not belong in the center. Decoration is how commodity stories go bad.
What the bank actually said
Keep the numbers in one place, and keep them labeled as the bank’s.
Ghali estimates that China’s strategic reserves hold about 2.05 million tons. He puts that at about 43 percent of global above-ground inventories. He estimates that U.S. stockpiling, driven by tariff fear, could tie up about 1.3 million tons in warehouses by year-end. Together, he says, U.S. and Chinese stockpiling will encumber about 71 percent of global inventories.
The arithmetic is worth doing in the open. If 2.05 million tons is 43 percent, the whole above-ground pile he is using is about 4.8 million tons. Add 1.3 million tons of U.S. metal and you are near 3.35 million. That is about 70 percent of 4.8 million. The 71 percent line is not a magic number. It is those two piles against that one pool. If the pool is defined differently, the percent moves. Read the note before you tattoo the percent on a thesis.
Do not let anyone turn inventories into mine supply. The world mines on the order of more than 20 million tons of copper a year. Seventy-one percent of a year of mine output would be a different, much larger claim, and it is not the claim in the figures above. A stockpile share is a share of metal already sitting above ground. A supply share is a share of what comes out of the ground. Mixing them is how a tight market becomes a fairy tale.
At the current pace of stockpiling, Ghali said freely available inventories would approach zero by the end of 2028. He added the sentence that keeps the call honest. The market must prevent that zero through demand destruction, or through higher prices. Zero free metal is not a forecast of empty factories. It is a forecast of a price that stops the emptying. The price is the brake. Exposure is how an investor owns the brake instead of standing in front of it.
The 50 percent, and the number behind it
A rally of about 50 percent to $22,050 a ton only makes sense from a price in the mid-$14,000s. Recent published prints put London copper in that neighborhood, within a short step of a record set earlier in September. Feeds differ by the day. They do not differ about the neighborhood. Copper is already near a record. The bank is not talking about a recovery from a crash. It is talking about another large step from a high place.
Check the multiplication. A 50 percent gain from $14,700 is $22,050. That is the headline. It is also the easy part. The harder part is the rest of the bank’s own path. A fuller reading of Ghali’s work has Deutsche Bank at an average near $20,900 for 2027, and near $18,500 for 2028. The second-quarter point of $22,050 is a peak inside that path, not a new floor that lasts forever. The bank itself fades the price after the squeeze. Anyone who hears only $22,050 has heard the exciting half.
That fade is not a reason to ignore the call. It is the reason to know what kind of exposure you want. A spike into 2027 and a lower average in 2028 is a trader’s shape. A world that still cannot build enough mines is an owner’s shape. The stockpile math can support both for a while. It cannot support a person who needs the peak to be permanent and has bought the story at the peak.
Jeff Currie, the former Goldman Sachs commodities chief, put the wider mood in a line in August. Get long and buckle up. The line is a mood, not a model. The model is Ghali’s piles. Use the mood if you want a feeling. Use the piles if you want a reason. Investors need exposure to copper for the reason, and they should buckle up because the path the bank drew is not a straight line.
A stockpile is not a supply
A ton in a strategic reserve is still a ton of copper. It has not vanished. It has changed jobs. Its job is no longer to be the next cathode a wire mill can buy. Its job is to sit, in case a border closes, a tariff hits, or a war lasts longer than a headline.
China has been building that sitting pile for years. The bank’s 2.05 million tons is the part Ghali is willing to estimate in public. The exact number inside a state reserve is never as clean as a chart. Direction matters more than the last decimal. The direction is a large buyer that does not need to show a profit next quarter.
The United States is the newer buyer in this story. Tariff fear pulls metal into American warehouses. U.S. copper futures have traded at a premium to London. The London exchange also has warehouses on U.S. soil. Ghali’s warning, in the fuller note, is that metal which enters that system may keep moving from one American shed to another even if the premium shrinks, and even if the tariff never arrives. The ton came in scared. It does not have to leave when the scare fades.
That is why stockpiling is worse than a normal deficit. A deficit says the world used more than it mined this year. Stockpiling says two rich buyers removed metal from the market on purpose, and they may not sell it back when you need it. The free float is what sets the price you can actually trade. When the free float shrinks, the same purchase moves the price more. Investors need exposure to copper before that float is a puddle, not after the puddle has a crowd around it.
2028 is the date the bank used as a wall
Approach zero by the end of 2028. Those are Ghali’s words for the free pile, if the pace holds. A wall on a calendar is useful because it can be missed. If the pace slows, the wall moves back. If China sells, the wall may never arrive. If a recession cuts demand, the wall was a scare. Write the date down. Then write the conditions next to it. A date without conditions is a slogan.
The bank does not pretend the wall is fate. High prices are supposed to stop the approach to zero. Users who can switch to aluminum will switch, once copper is painful enough. Ghali has also said today’s price is still not high enough to force that switch at scale. That is the hinge. The metal can stay tight until the price changes behavior. The price that changes behavior is the price in the call. It is also the price that hurts anyone who must buy copper and cannot pass it on.
There are two sides of that hinge, and only one of them is an investment in copper. The miner, the royalty, the holder of metal, and the fund that actually owns copper all want the hinge to swing up. The cable maker, the builder, and the car plant want it not to. Investors need exposure to copper if they want to be on the side that owns the scarce ton. They need to know they are short copper, in economic life, if their business must buy it.
Demand destruction is not a gentle phrase. It means a project that does not get built, a substitution that makes a product worse, or a factory that runs less. The bank is saying the world will choose that pain, or it will choose a higher price, or it will get both. Exposure is a bet that the price does a large share of the work before the projects die. If the projects die first, the scarcity story eats itself. That is the risk inside the theme. It is not a footnote. It is the theme’s exit.
The old story was data centers. The new story is who may buy.
For two years the popular copper story was electric cars, grids, and the buildings full of computers that train models. That demand is real. It is not the new part of Ghali’s note. The new part is a liquidity crisis. Free-floating inventories are the market’s cash. When cash in a market thins, price gaps. You do not need a new data center to get the gap. You need two governments buying and everyone else arriving at a smaller window.
De-globalization is the name for the reason those governments buy. Decades of underinvestment are the name for why a mine cannot answer them quickly. A copper mine is not a factory you switch on. It is a permit, a pit, a plant, and a decade. The tons that can answer a 2027 squeeze are the tons already mined or already financed. The tons in a slide deck are a 2032 conversation. Investors need exposure to copper that exists, or to the companies that already produce it, more than they need exposure to a story about a mine that might.
Resource nationalism is the political twin of the stockpile. China has tightened the tap on rare earths and other critical metals. The West has started to talk about defense the way it talked about it before the long peace. Copper sits in that talk because copper sits in the weapons, the grid that powers the weapons, and the economy that pays for them. A metal can be an industrial input on Tuesday and a security stock on Wednesday. Wednesday changes who is allowed to sell, and to whom. That is a supply cut even when geology has not changed.
The war chart is a map of dependence, not a buy signal
A graphic built from NATO risk assessments, circulated with the bank’s work, ties common exchange metals to machines of war. Fighter aircraft, tanks, missiles, submarines, ships, artillery, ammunition, torpedoes. The squares are aluminum, cobalt, copper, nickel, and the rest of the familiar list. Copper is marked at the highest risk on ammunition and on the torpedo. Other metals carry other colors. The picture is simple even where the legend is busy. Modern force is a metals list. Copper is on the list in red where the round and the torpedo need it.
A risk color is not a price target. It does not say copper goes to $22,050. It says a defense planner who cannot get copper has a problem that a market price is supposed to solve, or that a stockpile is supposed to solve. Governments that reach this conclusion buy piles. Buying piles is what Ghali is measuring. The chart explains the buyer. The inventory math is the trade.
There is a cousin metal that shows what “we will get to it later” looks like after 25 years of later. A Stifel graphic last week tracked the U.S. National Defense Stockpile of tungsten. It fell from about 137,000 tons to about 3,000. The graphic’s point was blunt. A 25-year drawdown will not be reversed quickly. The United States once held years of use in government sheds. It spent the peace dividend. It has now decided, in that telling, to be a buyer again, into a domestic industry that largely does not exist yet. China, in the same graphic, sits on the dominant share of both the mines and the refining.
Tungsten is not copper. The tonnage is smaller and the uses are narrower. The habit is the same. Sell the pile in the quiet years. Discover, in the loud years, that the pile was the policy. Copper is the large version of that habit, playing out while the pile is still being built rather than after it is gone. Investors who wait for the tungsten chart to be redrawn in copper units will be buying the story from the people who stocked up earlier. Investors need exposure to copper on the way into that chart, not as a souvenir of it.
What exposure is, and what it is not
Exposure means a holding that rises and falls with the copper price, or with the cash a copper business earns from that price. It does not mean a headline. It does not mean a single tweet about a record. It does not mean a junior stock with the word copper in the name and no copper in the mill.
The metal itself is the cleanest claim. A futures contract, or a product that holds copper, gives you the price and little else. You do not own a mine’s politics. You do own the roll of contracts, the storage, and the chance that a spike reverses. If Ghali is right about a 2027 peak and a lower 2028 average, the metal can pay you and then take it back. Size the claim for a round trip, not for a shrine.
A producing miner is a louder claim. Costs are the difference. A company that pulls copper from the ground at a cost far under the price earns more on each dollar the price rises. It also earns less, fast, if the price falls. Strikes, water, permits, governments, and debt sit between you and the ton. The ton in the annual report is not the ton in Ghali’s free pile. Both can be scarce. They are not the same scarce.
A royalty or a stream is a thinner claim. You take a slice of revenue or metal and you do not run the pit. You still live and die by the price, and by whether the operator keeps the mine open. A diversified commodity basket can be a diluted claim. If copper is 10 percent of the basket, you do not have the exposure this theme is about. You have a hint of it. Hints do not answer a 71 percent lockup.
There is also false exposure. A technology company that uses copper is not a copper investment. It is a customer. A customer is hurt by the squeeze the bank describes, unless it can raise its own prices. Buying the customer because you liked the metal is how people get the sign wrong. Investors need exposure to copper. They do not need accidental shorts dressed up as growth stocks.
Canada and other mining markets are full of companies that say they are the way in. Some produce. Some explore. Some will dilute you twice before the first ton. This article will not name a winner. A winner is a cost, a life of mine, a balance sheet, and a country, checked one at a time. The theme does not pick the ticker. The theme says the ticker should be a real claim, and that a real claim is worth having while the free pile shrinks.
Who gets hurt if you do not have it
The bidding war Ghali describes ends when someone stops bidding. The someone is usually the user with the worst ability to pass the cost on. A defense ministry can pay. A grid builder backed by a state can pay. A small manufacturer with a fixed contract cannot. The price does not feel unfair inside the warehouse. It feels unfair on the loading dock of the firm that was last in line.
Investors who own only the firms in that line are exposed to copper in the painful direction. Their margins are the shock absorber for Ghali’s rally. A portfolio of builders, cable makers, and equipment firms can look diversified and still share one input. When that input is the metal two governments are locking up, diversification among customers is not diversification. It is the same bet, repeated.
The reverse portfolio is not complicated to describe, and it is complicated to live with. Some metal. Some production. Maybe a royalty. Not so much that a 2028 fade, the one the bank already penciled in, wrecks the plan. Not so little that a real squeeze is a story you read rather than a price you own. Investors need exposure to copper in a size they can still explain on a quiet day. Quiet days are when oversized commodity bets turn into forced sales. The oil market offered that lesson in the same week. It travels.
What would make the theme wrong
The theme dies if the free pile is refilled. China can sell from the 2.05 million tons. A sale from a strategic reserve is a policy choice, and policy choices happen when domestic industry screams. Anyone who treats the Chinese pile as locked forever has not watched China use a stockpile as a tool in both directions. The tool cuts both ways. Exposure that assumes it cuts only up is a half-reading of a state.
The theme dies if the U.S. metal comes back out. Ghali’s own caveat is that it might not, because of where the warehouses are and how the premium works. Might not is not cannot. A tariff that never arrives, plus a premium that vanishes, can still pull some tons toward the highest bidder outside America. If those tons move, the 1.3 million is not a permanent hole. It is a delay. Delays do not pay for a permanent scarcity multiple.
The theme dies if aluminum and other substitutes arrive earlier than the bank thinks. The bank says today’s price is not enough to force the switch. A price near $22,000 might be. That is the bank arguing with itself, and the argument is healthy. The rally is supposed to create the substitution that ends the rally. Investors need exposure to copper for the trip. They do not need to marry the destination. The destination, in the bank’s 2028 average, is already lower.
The theme dies if demand dies first. A recession, a halt in grid spending, or a pause in the buildout of data centers would cool the bid. Stockpiles would look like a mistake. The price would fall from a record, which is a long way to fall. Near a record is the worst place to confuse a structural story with a guarantee. Structures can be true and still have a down 30 percent year. Copper has done that inside bull markets before.
The theme dies, more quietly, if you buy the wrong claim. A mine that does not produce, in a country that changes the rules, financed by a share issue every year, is not exposure to Ghali’s free pile. It is exposure to hope. Hope is allowed. It should not be the whole position. The pile is a physical fact. Match it with something physical, or with a producer whose costs you can read.
How to hold the idea without holding a fantasy
This is a way to think, not a portfolio you should copy.
Start from the free float, not from the target. Each time a new inventory number is published, ask a rude question. Did the metal available to everyone except the two stockpilers rise, or fall? If it fell, the theme is intact. If it rose, the theme is aging. Price can rise for a month on a headline while the float is healing. Price can fall for a month while the float is still shrinking. The float is the boss. The price is the employee.
Then read the bank against the bank. $22,050 in the second quarter of 2027. About $20,900 as a 2027 average. About $18,500 as a 2028 average. A person who needs all three numbers to be beaten has built a fantasy. A person who owns copper because the free pile is shrinking can survive the third number. The third number is the bank admitting that extreme tightness is a period, not a permanent address.
Separate the war demand from the war headline. Ammunition and torpedoes do not set the copper price by themselves. The grid and the ordinary economy are bigger. The NATO chart matters because it explains why a government will pay up and why it will not sell the pile back on your schedule. It does not matter as a reason to buy on the day a headline mentions a ship. Headline days are when the wrong size of exposure gets shaken out. The slow buyer of the pile does not care about your headline. Match that patience or own less.
Keep a sentence you can say without the word rally. Mine: two governments are locking up most of the visible pile, mines cannot answer quickly, and I want a claim on the copper that is still for sale. If you cannot say your version of that sentence, you do not have a thesis. You have a target someone else published. Targets get revised. The sentence should survive a revision, or you should leave.
The close
Ghali’s warning is easy to sell as a 50 percent rally and a date in 2027. The part that matters is duller. About 2.05 million tons in China’s strategic pile. About 1.3 million tons that tariff fear could pin inside the United States. About 71 percent of the above-ground inventory those figures are measured against, if you accept the bank’s pool. A free float that, at this pace, he thinks approaches zero by the end of 2028, unless price or pain stops it.
Copper near a record in the mid-$14,000s already tells you the easy money, if there was any, was earlier. The bank’s own path goes up toward $22,050 and then down toward an $18,500 average. That is a squeeze, not a promise of a new forever price. Aluminum can steal demand. China can sell. A recession can empty the bid. A fake copper stock can empty your account even if the metal does what the note says.
None of those exits removes the center. Investors need exposure to copper because the tons still for sale are the tons that will clear the market, and two of the largest buyers on earth are taking tons out of that sale. A portfolio that owns only the customers of copper owns the pain. A portfolio that owns no claim at all owns the surprise. A claim, sized for a spike that the bank itself thinks fades, is the exposure the scarcity argues for.
Get the claim while the warehouse still posts a price. After the free pile is gone, the price will be posted by whoever is left, and they will not post it for your convenience. Buckle up is optional. Exposure is the point.
Important information
This article is for information and education only. It is not investment advice and not a recommendation to buy or sell copper, a miner, a fund, or any other security. The case for exposure is a view about scarcity. It is not a personal plan. Commodity prices are volatile. A price near a record can fall a long way. Miners face cost, political, operational, and dilution risk. Futures and options can cost more than the cash you posted.
Inventory figures of about 2.05 million tons, 43 percent, 1.3 million tons, and 71 percent, the end-2028 free-inventory warning, and the $22,050 second-quarter 2027 price are from Daniel Ghali’s Deutsche Bank work as reported on September 28, 2026. A fuller public account of that work also cites a 2027 average near $20,900 and a 2028 average near $18,500, and treats aluminum substitution and a later return toward balance as real limits. Recent published LME prints put copper in the mid-$14,000s per ton, near a September record. Feeds differ. The NATO graphic and the Stifel tungsten stockpile chart, including a fall from about 137,000 tons to about 3,000, are used as context from those published graphics, not as findings of this desk. Jeff Currie’s August line is a comment, not a forecast by this desk. Readers should read the bank note and check live prices. This article does not consider any person’s goals or finances.