—— ◆ ——
| Core Summary |
|---|
| This may be the most awkward collective misjudgment in recent years. Goldman Sachs set its year-end 2026 target price at $13,735/ton, UBS at $13,228/ton, and Morgan Stanley gave an even lower figure of $10,650/ton. Meanwhile, LME spot copper has left $14,500 behind, with August cash contracts briefly touching a historical high of $14,830/ton. More intriguing is the process: LME inventory fell for 42 consecutive trading days, marking the longest streak since 2014, dropping to a five-year low of 204,975 tons, nearly half of which was already marked for withdrawal. The premium of spot over three-month futures reached $434/ton at one point, the widest spread since the short squeeze in 2021. This was not simply a case of "being wrong on direction," but rather a collective failure of the entire predictive framework based on "annual supply-demand balance sheets" when faced with three variables: tariff arbitrage, geographical mismatch of inventories, and sulfur shortages. This article attempts to dissect where investment banks went wrong. |
| Comparison |
|---|
◎ I. First, Look at This Awkward Comparison Table
Placing the target prices of major institutions side-by-side with spot prices makes the gap immediately apparent. It should be noted that Goldman Sachs’ $13,735 is not a pessimistic forecast from the beginning of the year; quite the contrary, it is the number just raised on June 1—a rise of more than 10% from its previous target of $12,465. That is to say, even after being revised upward and then considered an "aggressive bullish" target, it still lagged behind spot prices.
| Institution | 2026 Target Price | Adjustment Context | Gap vs. Spot |
|---|---|---|---|
| Goldman Sachs | $13,735/ton (year-end) | Raised on June 1, previously $12,465 | ~$800 below spot |
| UBS | $13,228/ton | — | ~$1,300 below spot |
| Morgan Stanley | $10,650/ton | — | Nearly $4,000 below spot |
| Bernstein | $12,419/ton (average) | Expected $11,750 for H2 | Significantly low |
| LME Spot | ~$14,500/ton | August cash price touched $14,830 | — |
Morgan Stanley’s $10,650 is particularly noteworthy because it represents another line of thinking—Macquarie holds a similar stance, arguing that the market does not lack copper and forecasting a surplus of 262,000 tons in 2026, with annual surpluses exceeding 700,000 tons in both 2027 and 2028. Such judgments are not without basis: data from the International Copper Study Group shows that the refined copper market experienced a cumulative surplus of 221,000 tons in the first five months of 2026, up from 117,000 tons during the same period last year. In other words, looking at the global annual balance sheet, copper is indeed in surplus—but prices are hitting record highs. This paradox is precisely what this article aims to explain.
| Reason for Error One |
|---|
◎ II. The First Mistake: Confusing "Global Balance" with "Deliverable Balance"
The core of traditional copper price prediction frameworks is the global annual supply-demand balance sheet—calculating how many tens of thousands of tons of surplus or shortage exist for the full year, and then deriving prices from that. This method has been broadly effective over the past two decades, but in 2026, it failed due to a historically rare geographical mismatch in the physical distribution of copper: while the global total is in surplus, the portion available for delivery is in extreme shortage.
Data depicts this mismatch with remarkable clarity. On one hand, there is frantic hoarding in the United States—the US imported over 200,000 tons of refined copper in July, its largest monthly import volume in at least 12 years. Combined with metal already held in Comex warehouses, LME registered inventories, and private port storage, total US copper inventories have surpassed 1 million tons. On the other hand, London is experiencing continuous bleeding—LME warehouse inventories have halved since May to less than 210,000 tons; by August 12, they had further dropped to 214,550 tons, a 14% decrease (over 35,000 tons) from the end of July; the latest data has fallen to a five-year low of 204,975 tons.
| The key here lies in the special role of "LME Inventory": it serves as the ultimate source of supply for the physical copper market and is also the metal used to settle expiring futures contracts. As the US absorbs large quantities of copper due to tariff expectations, the freely circulating volume available for delivery within the LME system is drained—even if the global total is in surplus, short sellers may still be forced to close positions at high prices because they cannot find physical metal. This is why "an annual surplus of 221,000 tons" and "spot prices hitting record highs" can coexist: they describe fundamentally different markets. |
|---|
There is a term that accurately captures this state—copper piled up in US warehouses due to tariff politics is becoming "economically trapped" tonnage, while other parts of the world are quietly suffering shortages. This copper exists physically and is counted in global supply statistics, but it cannot economically flow back to where it is needed—because once shipped out of the US, it means abandoning the bet on the tariff price differential.
| Reason for Error Two |
|---|
◎ III. The Second Mistake: Treating Tariffs as a Single Event Rather Than a Continuous Variable
The second flaw in the investment bank framework is their handling of tariff policy. Most models treat it as a "discrete event" that will be announced at a certain point, after which uncertainty is eliminated. Goldman Sachs’ scenario is typical: its base case assumes the US government announces a 15% tariff on refined copper in mid-2026, implemented in 2027; once tariff uncertainty is removed, investors would refocus on large-scale global surpluses, thereby bringing new downward pressure on prices.
But the reality is that this "boot" has yet to drop. The US Department of Commerce was originally scheduled to provide recommendations before June 30, but that deadline has passed without a public ruling, and traders are still hedging bets on both outcomes. The current policy framework proposes implementing a 15% tariff on refined copper starting January 2027, rising to 30% in 2028, but the market assigns a very low probability to its realization—as of August 14, the probability implied by Comex premiums that the 15% tariff takes effect as scheduled in January 2027 is only 14.6%; Societe Generale gives the same probability of 14.6%.
This creates a state difficult for investment bank models to handle: tariffs are neither implemented nor canceled, but persist in a state of "pending"—and it is precisely this pending status that continuously creates arbitrage motives and continues to siphon copper toward the US. If tariffs were clearly implemented, the incentive to hoard would disappear; if tariffs were clearly canceled, the hoarded copper would flow back. Only the state of "never deciding" allows the siphoning effect to continue indefinitely—this is exactly the scenario that any model based on "event-driven" logic finds hardest to depict.
| Reason for Error Three | |
◎ 4. The Third Error: Omitting Sulfur as a Variable
The third blind spot is the most concealed and best illustrates the limitations of forecasting frameworks. Goldman Sachs itself acknowledged this when it revised its forecasts downward in July—analysts explicitly stated that potential supply disruptions from the Middle East were not included in their predictive models, specifically highlighting Congo (DRC)’s reliance on seaborne sulfur. This self-admonition precisely identifies the most undervalued transmission chain in this market rally.
The logic chain is as follows: The Middle East accounts for approximately 24% of global sulfur production, and half of global seaborne sulfur trade must be exported via the Strait of Hormuz; sulfur is the raw material for sulfuric acid, which is an indispensable basic reagent in hydrometallurgical copper processing (acid leaching processes). When passage through the Strait of Hormuz is obstructed, the shock transmits directly to copper’s hydrometallurgical capacity—Robert Friedland, founder of Ivanhoe Mines, put it bluntly: “I hear traders are having difficulty sourcing any supplies; sulfur prices in Africa will rise significantly... If the disruption lasts more than about three weeks, oxide copper operations will have to shut down because they will run out of acid.”
Hydrometallurgical copper processing constitutes a significant share of global copper supply, and it is concentrated in regions like Africa and South America that have high dependence on imported sulfur. More problematically, sulfur is a byproduct of oil refining, and its output is entirely determined by upstream refining activity levels; it cannot respond to price signals through “increased production”—meaning its supply elasticity is even lower than that of crude oil itself. An investment bank model can predict ore grades, capital expenditure, and construction cycles, but few models treat “seaborne routes for a refining byproduct” as a constraint on copper output—this is one of the deep-seated reasons for the collective failure of these forecasts.
| Superposition |
|---|
◎ 5. Misfortune Never Comes Alone: A Chain Reaction of Supply-Side Shocks
Beyond inventory mismatches and sulfur shortages, the supply side itself experienced a series of incidents within the same time window. Listing recent events together reveals a startling density: the export ban on copper concentrates from Congo (DRC) has led Chinese smelters to reduce operating rates; storms caused a shutdown at Antofagasta’s Los Pelambres mine; Codelco has just delayed the Andes Norte project to 2029; Chile’s national output remains stagnant around 5.5 million tonnes; and Freeport’s Gresik smelter in Indonesia has been offline since August 8.
The foundation was already shaky earlier on—Goldman Sachs reduced its 2026 global mine supply forecast by approximately 350,000 tonnes in June due to reassessing the production recovery prospects at two key operations: Indonesia’s Grasberg and Congo (DRC)’s Kamoa-Kakula. The commonality among these events is that none of them will replenish LME warehouses in the short term; in other words, bad news on the supply side overlaps completely with bleeding on the inventory side in terms of timing.
| Disturbance Event | Impact | Status |
|---|---|---|
| DRC Concentrate Export Ban | Chinese smelters reduce operating rates | Ongoing since June 2026 |
| Los Pelambres Shutdown | Production loss due to storms | Phased/Temporary |
| Codelco Andes Norte | Expansion project delayed to 2029 | Long-term impact |
| Freeport Gresik Smelter | Offline since August 8 | Ongoing |
| Grasberg/Kamoa-Kakula | Goldman Sachs cuts global supply by ~350k tonnes | Slower recovery than expected |
| Mechanism |
|---|
◎ 6. How the Short Squeeze Formed: A Self-Reinforcing Cycle
Connecting the above factors reveals the complete mechanism of this short squeeze. Step one: Tariff expectations drew copper toward the United States, causing LME deliverable inventories to decline continuously; step two: Falling inventories pushed up the cash premium. On August 11, the LME cash premium over the three-month futures widened sharply to $207.5 per tonne, compared to approximately $34 per tonne at the end of July, subsequently expanding further to $434 per tonne—the widest level since 2021; step three: Wide backwardation itself became a strong incentive to hold physical metal, further suppressing willingness to deliver.
Step four is the most critical: Declining inventories and widening premiums forced traders holding short positions to bid for scarce physical metal. Short covering demand may intensify further, especially given that near-month contract open interest remains substantial relative to available warrants. With canceled warrants (metal marked for withdrawal) accounting for nearly half of LME inventories, the actual free-floating volume available for delivery is much smaller than the book figures suggest. This creates a self-reinforcing cycle: less availability → higher premiums → fewer willing to deliver → less availability.
Notably, the market has a clear understanding of the nature of this state—analysis explicitly points out that wide backwardation indicates a “shortage of immediately deliverable copper,” not necessarily an “annual supply shortage.” This distinction is the key to understanding this rally: investment banks forecast annual supply and demand, while the market trades immediate availability. These two factors are highly correlated in normal years but were completely torn apart by tariff politics in 2026.
| Assessment |
|---|
◎ 7. Judgment and Advice: When Will This Short Squeeze End?
First judgment: The conditions for ending this short squeeze lie not in mines, but in Washington. Because the core driver is geographical inventory mismatch caused by tariff arbitrage, any clear tariff ruling—whether implemented or cancelled—will weaken hoarding motives and allow copper under “economic freeze” to re-enter global circulation. This also means traders should focus not on ICSG monthly balance sheets, but on the Department of Commerce’s announcement calendar.
Second judgment: Investment banks’ target prices will likely continue to be revised upward, but this does not constitute a reason to chase longs—in fact, it is a lagging indicator. Goldman Sachs raised its target from $12,465 to $13,735 in June; even so, it still trails the spot price. This shows that in such structurally mismatch-driven markets, sell-side research serves to explain rather than predict. Following target price adjustments to enter positions usually means buying into sentiment peaks. The market has indeed developed spontaneous correction mechanisms—some traders note that the premium and backwardation of near-month contracts relative to far-month contracts will begin to attract metal flows back into the LME system.
Third judgment, and perhaps the most easily overlooked: The end of the short squeeze does not mean prices will fall back to investment bank target prices—because beneath the short-term factor of inventory mismatch lies another layer of genuine medium-to-long-term supply constraints. Four structural factors remain: declining ore grades, insufficient capital expenditure, ten-year construction cycles, and community and approval risks. Data from the International Copper Study Group (ICSG) shows that global mine production fell by 1.1% year-on-year in the first half of 2026, potentially marking the first annual decline since 2017. In other words, the short-term squeeze premium will dissipate as tariffs become clear, but the long-term logic supporting the price center will not disappear simultaneously.
Three Recommendations for Supply Chain Participants: First, incorporate the “LME Cancelled Warrants Ratio” into daily tracking metrics—it reflects true availability more accurately than total inventory levels. With nearly half of current stocks marked for withdrawal, this signal is far more significant than the headline figure of “200,000 tonnes of inventory.” Second, for enterprises with physical procurement needs, it is inadvisable to schedule purchasing rhythms based on investment bank target prices at present. Instead, greater reference should be given to the spot premium structure and delivery lead times; in a backwardation environment, the opportunity cost of “waiting for a pullback” is extremely high. Third, include sulfur/sulfuric acid supply within the risk register for copper supply chains—the “three-week” inventory threshold highlighted by Friedland serves as the critical window for assessing whether African hydrometallurgical copper production capacity can maintain operations.
Finally, there are three signals worth monitoring closely: the timing and conclusion of the U.S. Department of Commerce’s final ruling on refined copper tariffs; whether the LME cancelled warrants ratio will peak and decline; and whether the spot premium will converge from its extreme level of $434. Among these three signals, if any two turn, it can be considered that the short squeeze is entering its final phase—but until then, using annual balance sheets to predict spot prices remains a methodology destined to err.
—— ◆ ——
“They calculated correctly how much copper there was for the full year, but they miscalculated where that copper was—and the market has always traded the latter.”
— END —