On Thursday, October 15, 2026, Duncan Wanblad told a Financial Times summit in London that technology companies have talked to Anglo American about buying equity. The point of the equity, he said, would be to lock in metals that go into data centers. He said the talks were "absolutely being had." He said there had been "deep expressions of interest." He also said no stake had been taken. The choice on the tech side, as he framed it, was still a choice: a long-term offtake, or a share of the company, as a way to hold a claim on future supply.
Here is the only idea. Interest is not a cathode. A conversation can be real and still not be a pound of copper. No tech firm has bought a piece of Anglo. The merger that would make Anglo a larger copper miner has not closed. The last government that has to say yes is China, and the remedy reported on the table is not a data-center contract. It is a promise to keep copper concentrate flowing to Chinese smelters.
Nothing here is a recommendation to buy or sell Anglo American, Teck Resources, copper, or any technology stock. A CEO's sentence is not a supply contract. You can lose money being right about a shortage, and you can lose money being early to a story that has not happened.
What he said, and what he did not
Wanblad was speaking at the FT Metals and Mining Summit. Mining.com reported the remarks the same morning. A separate account from the same day, on Miningmx, has him saying the conversations are high on the mind of people downstream, "certainly in the new forms of demand — data centres, AI," and that some of those talks include direct investment. Both accounts agree on the limit. Deep interest. Nothing signed.
That limit is the whole story. An offtake is a contract for metal. An equity stake is a share of the company that might produce the metal. Neither one has been announced. A reader who turns "talks" into "tech is buying Anglo" has added a fact the CEO did not give. A reader who turns "scarcity" into "the price must rise from here" has added a forecast. He said a lot of people are looking for ways to secure supply, because of the scarcity he and others have been talking about. Looking is not securing.
There is a signed version of this idea, and it is small enough to keep the Anglo talks in scale. In January 2026, Rio Tinto and Amazon Web Services announced a two-year collaboration. AWS would take Nuton copper from the Johnson Camp mine in Arizona for use in U.S. data centers, and would lend cloud tools to the process. The deployment is aimed at about 30,000 tonnes of refined copper over four years. That is real metal, with a named customer. It is also about 7,500 tonnes a year. Anglo and Teck, if they combine, are described as a group with about 1.2 million tonnes of copper output a year. One signed data-center deal does not feed a year of that merger. It shows the form. It does not show the size.
The price fell on the morning of the speech
Copper did not rise to greet the scarcity talk. It fell for a third day.
Comex copper for December delivery traded as low as about $6.52 a pound on Thursday, near $14,370 a tonne, and by late morning in New York was about $6.58, or roughly $14,510 a tonne. The September 9 settlement of $6.89 is still the record on that contract. Late morning left the price about 4.5 percent under that print. Mining.com's Anglo piece, filed earlier, said the active contract was within 6 percent of the record. Both descriptions fit a morning that started near $6.52 and firmed a little. Neither description is a forecast.
In London, cash copper had settled at a record $14,822 a tonne on Monday. By Thursday the three-month price was lower, around $14,360 in official trading after a dip to $14,289. LME stocks were reported at 228,000 tonnes, down about 8 percent since the start of October. Visible stocks can fall while a headline price falls. Those are different facts. One is a warehouse. One is a screen.
The drop had more than one reported cause. Hotter U.S. inflation put talk of a Federal Reserve rate hike this month back on the table. A rout in Asian chip stocks spilled into metals. And a deal at Escondida, the large Chilean mine, took the nearest strike risk off the table. A CEO can talk about scarcity at a podium in London while the screen in New York prices a rate scare, a chip selloff, and one less disruption. If you only hear the podium, you will think the metal confirmed him. It did not. Not that morning.
A tight concentrate market and a softer refined price can share a week. Smelters buy concentrate. The screen mostly shows refined metal, or a futures claim on it. China's smelters have built capacity that runs ahead of the concentrate they can get. That is a feedstock problem. It is not the same sentence as "the world has no copper." The International Copper Study Group's preliminary read for the first seven months of 2026 was a small apparent surplus of refined copper, about 32,000 tonnes, against a surplus of about 157,000 tonnes in the same months of 2025. Apparent demand is not a warehouse count. China's unreported stocks are outside that sum. Use it as a warning against the word "scarcity" when you mean "the price I like," not as proof that metal is long.
Two market caps are not a mine
Mining.com's Thursday morning prices put Anglo's New York listing at $26.86, up 0.6 percent, for a market value of $58.0 billion. Teck Resources was $64.96, up 0.3 percent, for $31.9 billion. Add those two indicative values and you get just under $90 billion. The same piece said Teck had gained about 50 percent over twelve months and Anglo about 36 percent, and set the pair beside Glencore at $89 billion and Freeport-McMoRan at $99 billion.
Those are share prices at one hour, summed. They are not cash. They are not the value of the copper in the ground. The chart Mining.com ran, rebased to the last close before the September 9, 2025 announcement, marked the pre-deal pair at about $33 billion for Anglo and $17 billion for Teck, and the merger announcement at $53 billion. Later wire copy has called the same transaction about $54 billion. The number moved because the language moved, and because the shares moved. Do not treat $53 billion, $54 billion, and $90 billion as three measurements of one rock. The first two are what people called the deal. The third is what two separate stocks were worth on a Thursday morning in 2026, added together, before the companies are one company.
A merger of equals does not become worth the sum of the two market caps in any simple way. There is an exchange ratio. There is a share count after the deal. There is whatever China still requires. Mining.com labeled the chart's market caps as indicative. Leave them indicative. A 0.6 percent uptick on the morning of a conference quote is not a verdict on the talks. No stake was bought. The stock did not have to reprice a deal that does not exist.
The stamp that is still missing
Canada, the European Union, the United States, Japan, Chile, and South Korea have signed off on Anglo-Teck, on the timeline Mining.com's chart records: Canada in mid-December 2025, the EU at the end of January 2026, the United States, Japan, and Chile in mid-February, South Korea by the end of March. China has not. The parties still say they expect to close by March 2027. That is inside the eighteen months Wanblad has said these approvals now take. He said in August that deals need twelve to eighteen months, a bit longer than five years ago, and that regulation was not making them impossible. Thirteen months in, one regulator is left. "Not impossible" and "done" are different words.
On October 2, Reuters reported that China's State Administration for Market Regulation had asked Anglo to commit to a steady flow of copper concentrate to Chinese smelters, including tonnes sold through traders, as a condition of approval. Three people familiar with the talks described the ask. Anglo's public line was that it was working through SAMR's process and would not add detail. Teck declined to comment on the regulatory process. SAMR did not immediately respond to Reuters. The remedies then under discussion did not include asset sales.
At the London summit, Wanblad would not be drawn on what China wants in return. That silence is not a denial of the Reuters report. It is a refusal to negotiate in a ballroom. Read it as silence.
The competition arithmetic, as Reuters and Mining.com both give it, is that the combined group would control around 5 percent of global copper supply. Thresholds that usually wake a competition authority are often put at 10 to 15 percent. Five percent is not a corner on the world's mines. China is not reviewing this deal because Anglo-Teck would own the market. China is reviewing it because China smelts a very large share of the world's copper and its smelters are short of feed. A 5 percent miner that sells concentrate can still matter to a smelter that cannot fill the furnace.
Do not launder the share. Mining.com wrote that China refines up to 60 percent of the world's copper cathode. The International Copper Study Group's September 2026 note is more careful. It put China and the Democratic Republic of Congo together at about 59 percent of global refined output in the first seven months of the year. It put China's share of world refined use at about 59 percent. Those are not the same as "China refines 60 percent." Congo's metal is not China's furnace. Use the ICSG split. The point survives the correction. The regulator who has not signed is the regulator in the country that uses the most refined copper and runs a huge smelting industry. The ask that leaked is for concentrate, the dirt before the cathode, including dirt that currently moves through traders. Traders are how tonnes stay flexible. A commitment that reaches through the trader is a commitment to reduce that flexibility.
A tech firm that bought equity in Anglo would not sit in front of that smelter. It would sit behind it. The cathode a data center wants is refined metal. The document China is reported to want is a promise of concentrate. Those are two ends of the same chain. Equity in the miner does not cut the chain. If the remedy is a steady flow east, then a slice of the mine is a slice of a flow that has a prior claimant. The prior claimant is not Amazon. It is the approval.
What 1.2 million tonnes does not settle
About 1.2 million tonnes a year would put the combination among the five largest copper producers, which is the rank Mining.com used. Fifth is the rank Reuters used. Rank is not a surplus. Rank is a place in a list. The world's refined use is many times 1.2 million tonnes. ICSG had Chinese apparent demand up about 3.5 percent in those seven months, and world apparent use up about 2.4 percent. A large copper company can still be a small part of the world's need. That is what 5 percent means. It is large enough to be worth a regulator's time. It is not large enough to end a shortage by existing.
Data centers are a real new bid for copper. Cable, transformers, busbar, and cooling all take metal. They are not a new geology. They do not move the ore body at Los Bronces or at Quebrada Blanca. They move the people who would like a contract in front of the people who do not have one. Wanblad described that queue. He did not say Anglo had stepped out of it and handed someone an allocation. The Johnson Camp figure is the right yardstick for the queue. Tens of thousands of tonnes over several years is a contract. A million tonnes a year is a mining company. Wanting the second thing, and talking about buying a share of it, is not the same as receiving either.
The price of the metal is a third yardstick, and it was lower on the day. Record prints are recent. Monday's LME cash record and the September 9 Comex record are not ancient history. A market can be strong over a year and down on a Thursday. Anyone who uses Wanblad's sentence as a reason the next tick is up has skipped the three reasons the last ticks were down. Inflation, a chip-stock break, and a quieter Escondida are not copper's long story. They were Thursday's story. The long story is still a question about concentrate, permits, and grade, not a question about whether a software company has a meeting on the calendar.
The other gate is a closure
Anglo is trying to stop being a collection of businesses and become a copper and iron-ore company with Teck's assets inside it. Two exits are still open. One is diamonds. In early October Wanblad said he hoped to finish a sale of De Beers by the end of the year. Mining.com's Thursday piece said Anglo is seeking about $1 billion for that business. Treat the billion as the figure in that report, not as a cheque. A hope for a year-end close is not a close.
The other exit is nickel in Brazil, Codemin and Barro Alto. The buyer is MMG. The price on the deal is $500 million. The European Commission opened an in-depth probe late last year, on the fear that MMG could steer ferronickel away from European stainless-steel makers. Mining.com said a formal warning went out last month. MMG has offered long-term supply commitments to Europe and says the diversion case is built on selective quotes. The Commission has not ruled.
Reuters obtained the remarks Anglo's Brazil chief operating officer, Ruben Fernandes, took into a closed hearing in Brussels. He said MMG was the only credible buyer the company had found. He said a prohibition would leave care and maintenance as the path to closure. The hearing was held. MMG says it told the Commission it would keep supplying Europe and would run the business on its own, with the local team. A threat of closure is a negotiating fact. It is also an operating fact if the sale dies. Care and maintenance is not a mine. It is a mine that has stopped, with the option, later, of starting again or not. Stainless steel in Europe and a data center in Virginia are not the same customer. They are fighting over different chapters of the same company. One chapter is for sale and may shut. The other chapter is the one tech firms are asking to buy into, and it is not theirs until a merger closes and someone signs.
Put the two gates side by side. China can withhold a copper merger unless concentrate keeps moving to Chinese smelters. Europe can withhold a nickel sale unless ferronickel keeps moving to European mills. In both cases the state is not buying the metal. It is pricing its permission in metal. That is the pattern. A tech company's wish to "secure supply" by owning equity runs into a prior version of the same wish, held by governments, and written as a condition. The company does not get to satisfy all of them with one pile of rock.
What to write on the card
Write the quote with its limit. October 15, 2026, London: Wanblad said tech talks about equity or offtake are happening, that the interest is deep, and that no stake has been taken. Mining.com and Miningmx both carried that limit. Do not drop it when you retell the story.
Write the copper price with the hour. Thursday's December Comex contract traded near $6.52 and was about $6.58 by late morning, against a September 9 settlement record of $6.89. LME cash set a record at $14,822 on Monday. The third daily drop came with hotter U.S. inflation, a chip-stock rout, and less immediate strike risk at Escondida. LME stocks were about 228,000 tonnes, down about 8 percent on the month.
Write the merger as unfinished. Announced September 9, 2025, at about $53 billion on Mining.com's chart, and called about $54 billion in later Reuters copy. Canada, the EU, the United States, Japan, Chile, and South Korea have approved. China has not. Reported ask, October 2: a steady flow of copper concentrate to Chinese smelters, including trader volumes, not an asset sale. Expected close: by March 2027. Combined copper output: about 1.2 million tonnes a year, around 5 percent of global mine supply. Thursday morning indicative market values: Anglo about $58 billion, Teck about $31.9 billion. The sum is not the deal.
Write the nickel line as a threat, not a shutdown. Sale to MMG at $500 million. EU in-depth review still open. Fernandes told the Commission, in remarks Reuters reported, that a block means care and maintenance on the way to closure, because MMG was the only credible buyer they had. MMG says it will supply Europe. De Beers is a separate, unfinished sale.
Write the one signed tech contract you can actually count. Rio Tinto and AWS, January 2026, Nuton copper at Johnson Camp, about 30,000 tonnes over four years. That is a cathode with a customer's name on it. Anglo's version is still a meeting.
What an investor can do with a meeting
You can refuse the headline that skips the verb. "Come knocking" is not "bought." "Deep expressions of interest" is a mood. A mood does not change the share count. If a stake is announced later, read the contract. How many shares. At what price. What metal, if any, is promised. For how many years. From which mine. At a fixed discount, or at the market, or at a formula. A stake with no offtake is a financial bet that wears a supply costume. An offtake with no stake is a supply contract that does not need the costume. Wanblad said the tech side had not chosen. Until it chooses, you do not have either.
You can separate concentrate from cathode. The merger's political price, if Reuters' sources are right, is concentrate for China. The product a data center installs is cathode, wire, and component. A miner sells, in the main, concentrate or cathode depending on the asset. Owning a few percent of the miner does not let a software company skip the smelter, unless the mine is one of the few that makes cathode on site, as Nuton is trying to do at Johnson Camp. Anglo's Chilean concentrates do not work that way. They need a furnace. China has the furnaces. That is why the furnace's government is still in the deal.
You can separate Thursday's price from the decade's need. A third down day, a record still close overhead, a small apparent refined surplus, and falling LME stocks can all be true. None of them is a target price. None of them is a reason to treat a conference quote as a bid for the stock. The stock was up less than 1 percent when the quote was news. The market did not hear a transaction. You should not invent one to make the column balance.
You should not do the flattering thing. Do not call five percent of mine supply a cure for data centers. Do not call a Chinese supply commitment a defeat of the merger if the merger still closes. A condition can be the price of a yes. Do not call care and maintenance in Brazil a copper story. It is a nickel story, and it is still a threat. Do not add Anglo's market cap to Teck's and spend the $90 billion in your head. The companies are not one company. China has not said they may become one.
What would make this reading wrong
The reading is wrong if a tech firm has in fact bought Anglo shares, or signed an offtake, and Wanblad's "no stakes" line was already out of date by the time it was reported. Check the filings before you repeat the sentence next week. A stake can appear on a Friday. This piece is about a Thursday on which he said none had.
The reading is wrong if the October 2 Reuters account of SAMR's concentrate demand was wrong, or if the remedy has since changed to an asset sale, a fine, or a ban. Anglo did not confirm the ask in public. Three unnamed people did. Unnamed people can be early, partial, or wrong. Wanblad's refusal to discuss the ask is not proof the ask is real. It is also not proof the ask is false. If SAMR approves with no supply condition, the line about a prior claimant has to come out. The line about an unfinished merger stays until the close.
The reading is wrong if China's share of refining is the figure you need and you use the ICSG's use-share instead. This piece does not claim China refines 60 percent. It claims China and Congo together were about 59 percent of refined output in the first seven months of 2026, and that China's share of apparent world use was about 59 percent. If ICSG revises those shares, use the revision. The regulator is still in Beijing either way.
The reading is wrong if the Thursday copper prices were not in that range. December Comex near $6.52 at the low and about $6.58 later in the morning, LME cash at a record $14,822 on Monday, and the September 9 Comex settlement at $6.89, are the prints used here. A revised settlement changes the "how far under the record" sentence. It does not change the fact that the metal fell while the CEO talked about scarcity.
The reading is wrong if the $58.0 billion and $31.9 billion were misread from early trade, or if they were already stale by the close. They are morning indications from one newsroom. They are not an official merger value. The $53 billion and $54 billion labels for the September 2025 deal are two newsrooms' names for one announcement. Do not average them into a phantom price.
The reading is wrong if Fernandes did not deliver the care-and-maintenance warning, and the Reuters text was only a draft that was never spoken. MMG's account of the hearing says the hearing happened and that MMG made its supply case. The closure path is Anglo's argument. It becomes a fact only if the Commission blocks the sale and the company then stops the mines. A threat in a hearing room is not a shutdown notice.
The reading is wrong if Johnson Camp's 30,000 tonnes over four years is not the target in Rio's January release. It is the scale used here, from that release, to show that a named data-center contract can be real and still be small next to 1.2 million tonnes a year. If the four-year target changes, change the comparison. Do not promote it into a second Anglo.
The idea, once
Technology companies are asking miners for a way to hold copper before a data center needs it. At Anglo, on October 15, 2026, that ask was still a conversation. Duncan Wanblad said the interest was deep and that no stake had been taken. The same morning, copper fell for a third day, to about $6.58 a pound after trading near $6.52, below a September record of $6.89. The merger that would lift Anglo's copper toward 1.2 million tonnes a year, about 5 percent of world mine supply, still needs China. The condition reported on October 2 is a steady flow of concentrate to Chinese smelters, including tonnes that now pass through traders. A share of the miner, if anyone ever buys one, would be a share of a company that has not finished that merger, has not finished selling its nickel mines, and has not finished selling De Beers. Interest is not inventory. A meeting is not a cathode. The cathode still has to be mined, smelted, and permitted, and the last permit on the table is written in concentrate, not in a data-center contract.