Oppenheimer cut Almonty to a $19 target and still calls the shares attractive. The mill, in the company’s own late-September letter, was running 13 hours a day. The ban and the bags are not the same fact.
This is not a recommendation to buy or sell Almonty Industries, tungsten, a defense stock, or any fund. It is not a price target. It is not a view on whether the shares are cheap at 5.7 times one analyst’s guess of next year’s profit. A mill can miss a ramp. A booked ship can sit. A rule can slip, or it can arrive and still not buy your concentrate. You can lose money on a metal a government says it needs. Need is not a purchase order. A purchase order is not a check.
Here is the only idea. The ban is a date. The mill is a ramp. On January 1, 2027, a U.S. defense rule reaches certain tungsten back to the mine, not just the furnace. Almonty Industries is being treated as the Western answer to that date. What the company has, in writing, is narrower. Sangdong, in South Korea, produced its first tungsten concentrate in 33 years in late September. The chief executive said the mill was on 13-hour days and would go to 24 hours “over the coming days.” A first cargo has been talked about for October 24. Monday’s investor day can confirm hours, a ship, and a plan. It cannot make January buy the ore. If you cannot say which of those you own, you own a calendar with a mine’s name written on it.
A rule is not a customer
The rule people are trading is real. It is also smaller than the slogan. Through December 31, 2026, the defense acquisition clause on covered materials mostly stops the Pentagon from buying certain metals that were melted or produced in China, Russia, North Korea, or Iran. On January 1, 2027, the same clause, in 48 CFR 225.7018-2, steps upstream. For tungsten metal powder and tungsten heavy alloy, the restriction then includes mining or production of the ore or feedstock, including recycled material, through the powder and the alloy. Melted in Ohio does not wash rock that was mined in a covered country. A third country that only tolls Chinese concentrate does not, by itself, make the metal clean for this rule.
Read the nouns. The clause names tungsten metal powder and tungsten heavy alloy, and finished or semi-finished parts that contain the heavy alloy. It does not say “all tungsten, from every seller, for every use.” A drill bit for a civil mine is not this clause. A missile part that is tungsten heavy alloy is the kind of thing the clause is built for. Armor-piercing rounds, some missile and aircraft parts, and other defense metal sit in that neighborhood. The list of uses is long. The list of what the rule actually bans is a column in a regulation. Trade the column. Do not trade the poem.
There is a second paper, and it is not the same sentence. A July executive order set a January 2027 push for companies to stop using certain critical materials from China, Russia, Iran, and North Korea, with waivers if the government accepts a plan to switch suppliers. An Army Materiel Command official said in mid-October that mapping a supply chain down to the raw material is not done, and that the service expects some impact in January. A September report from the Aerospace Industries Association said significant domestic processing and refining for aerospace-grade material is not expected until 2028 or later, and that qualification can push the date further. The buyer at the center of this story has been told, by its own world, that the deadline will bite before the replacement capacity exists.
That bite is not a contract with Almonty. A rule can raise the value of rock that was never in China. It can also be waived, delayed in practice, or satisfied by a competitor who is already pouring powder. China mines the large majority of the world’s tungsten. The United States does not have a producing primary tungsten mine at the scale Sangdong is aimed at. Those two facts are why the date matters. They are not a tonnage award. The Pentagon does not mill ore because a calendar flips. A mill does, if it is running, and if someone qualified to buy defense metal will take the product.
What the mill has actually done
Sangdong last produced tungsten in 1993. Almonty said that on September 23, 2026, when it marked the first concentrate since then. Lewis Black, the chairman and chief executive, called it a 33-year gap and credited the crew that stayed through construction. That is a real milestone. It is a bag, not a year.
Six days later, on September 29, the company put a shareholder letter out of Dillon, Montana. Black wrote the line that should be taped above the model. “Sangdong is no longer a future promise. The mill is running for 13 hours a day, seven days a week, which will be increased to 24 hours a day, seven days a week, over the coming days.” He also said the first bags were ready for export. Ready is not shipped. Shipped is not paid. Paid is not a quarter.
Hold the arithmetic of 13 hours. A mill on 13 hours is a little more than half of a 24-hour day. “Over the coming days” is a hope with a short fuse. If those days have passed and the mill is on 24 hours, Monday’s room should say so in a sentence with a date. If the mill is still on a short day, the ramp is the story, and the ban is a spectator. Do not let a September verb do an October job.
Oppenheimer’s Ian Zaffino, writing Thursday ahead of the investor day, said the processing plant received final certification on September 17 and can produce concentrate. He said grinding is at full throughput and the company is still tuning the flotation circuit. Flotation is where a lot of tungsten mines make or miss their recovery. Full grinding with a circuit still being tuned is not nameplate. It is a plant that is allowed to run and is still learning the ore. He also said the first shipment is expected in late October, and that this is slightly behind the house view he started with. He cut his near-term revenue and profit guesses because of that lag. A late ship is a small sentence. It is the sentence that separates a restart from a business.
Other notes have said Black has booked that first cargo for October 24. Booked is a plan on a calendar. It is not a bill of lading, and it is not revenue. One secondary write-up has already talked as if the commercial shipment happened. The company’s September words were “ready for export.” The analyst’s Thursday words were “expected in late October.” Use those. If the ship left between Thursday and the hour you read this, the proof is a company line with a date, not a recap that got ahead of the dock. A cargo that leaves can still miss the quarter if the export paper or the customer’s acceptance sits in between. DA Davidson has pointed at fourth-quarter revenue for that reason: the gap between a physical ship and a booked sale. Production and revenue are cousins. They are not twins.
Nameplate is not one number
Ask what “full” means before you multiply a price by a dream. The company’s own second-quarter release said Phase I, once fully operational, is aimed at about 640,000 tonnes of ore a year. A fully permitted Phase II is contemplated to take throughput as high as 1.2 million tonnes a year. Those are company targets. They are not last quarter’s mill report. As of the June quarter, the release still described Sangdong as in commissioning and ramp-up. September’s bags came after that sentence. They do not retire it.
A Bank of America note, as carried ahead of Monday, put the nameplate investors should listen for near 1,100 tonnes a day, and the Phase II step-up near 1.2 million tonnes a year, with concentrate roughly doubled. Do the year. Eleven hundred tonnes a day, if the mill ran every day, is about 400,000 tonnes a year. The company’s 640,000 tonne target is a different year. It implies something closer to 1,750 tonnes a day if you smear it across 365 days, and more than that if the mill takes maintenance days. Both figures can be honest if they describe different stages, different recovery, or a loose use of the word nameplate. They cannot both be the same steady state. Monday is useful if someone says which number is the 2026 mill and which number is the study. It is useless if both stay on the slide.
Zaffino’s model, as reprinted this week, expects Phase I to deliver about 640,000 tons of ore a year, or about 240,000 metric ton units of concentrate, and about $500 million of revenue at the prices he was using. A metric ton unit is ten kilograms of contained tungsten trioxide. It is the unit the concentrate trade uses. It is not a ton of rock, and it is not a ton of missile parts. Two hundred forty thousand of those units times a headline ammonium paratungstate price will not equal cash. Payability, grade, freight, and the offtake formula sit in between. Treat $500 million as an analyst’s sketch at a price that has already moved. Do not put it in a checking account.
The offtake is the closest thing to a customer the mine has published. On July 14, 2026, Almonty amended its Phase I agreement with Global Tungsten and Powders, a Pennsylvania firm in Austria’s Plansee group. The term went from 15 years to 21, counted from first delivery. Total contracted volume rose 40 percent, from 3.15 million metric ton units to 4.41 million. The minimum after ramp-up is 210,000 units a year. The contract covers about 90 percent of Phase I concentrate. Pricing on the contracted volume improved by about 6.3 percent. The company’s headline cast the amendment as roughly $490 million of annual contract revenue at the APT price then in force. The same release also uses wording that can be read as a 21-year sum. Those two readings are not the same business. At a multi-thousand-dollar unit price, 210,000 units a year is a large gross line before the formula. It is still a formula on rock that has to be mined. Ninety percent of a plan is not 90 percent of a year already dug. Phase II is outside that contract. So are Spain and Rwanda and Montana. The offtake is a seatbelt for Phase I. It is not a receipt.
The price that fell is not the price you sell
Zaffino said tungsten APT prices had moderated from $3,140 per metric ton unit to $2,700 in the latest week. He put the July-through-September stretch near $3,000. He also said something easy to skip, and it is the market structure. Western APT and oxide plants are running at capacity under long-term contracts. They have limited room to take extra concentrate. So the finished price can stay tight while the concentrate price softens. Almonty sells concentrate. It does not, today, sell the powder the Pentagon clause names. A high APT print on a screen can be a full plant that will not bid for one more bag. A softer concentrate price is the price that hits the mine.
Management, in his telling, still thinks upstream concentrate stays tight, and should stay tight, as the defense rule arrives. Both lines can live in one note. Finished plants are full. Mine supply outside China is not loose. And the bid for the bag can still fade because the plant cannot take it this month. Tight in 2027 is not a bid in October. If you buy the stock because APT was $3,140, you are buying a week that already left. If you buy it because concentrate is scarce into the rule, you need the scarcity to show up in the price of the thing Sangdong ships, not in the price of the thing a full plant refuses to make more of.
This is why the pullback in mining shares since the spring is not, by itself, a sale or a verdict. The early spike came when Beijing tightened exports of tungsten and a set of other defense metals. The later drop came as the tape cooled and as investors remembered that a restriction is not a mine. Zaffino cut his 12-month target from $25 to $19 and kept an Outperform rating. He said the cut reflects lower multiples on peers, not a decision that the rock got worse. He still called the shares attractive at 5.7 times his fiscal 2027 EBITDA guess and 4.2 times fiscal 2028. A rating is a word. A target is a multiple times a profit that assumes the ramp. The multiple can fall because every miner de-rated. The profit can fall because the ship is late and the concentrate price is $2,700 instead of $3,140. He already cut the near quarter for the late ship. The target cut and the “attractive” line are the same model looked at from two ends. Neither end is a bag on a truck.
The quarter he just marked down
The note, as carried this week, is blunt about the next two quarters. For the third quarter he now estimates revenue of C$49 million and EBITDA of C$25 million. His prior pair was C$75 million and C$47 million. He put the Street at C$94 million and C$66 million. For the fourth quarter he now estimates C$75 million and C$57 million, down from C$110 million and C$99 million, with the ramp toward full Phase I doing the repair. These are his numbers, in Canadian dollars. They are not company guidance. Almonty has not promised that third-quarter profit. If the Street is still up at C$94 million while he is at C$49 million, Monday’s risk is not only the mine. It is a room full of models that have not all moved to the same clock.
A cut from C$75 million to C$49 million is not a footnote. It is the admission that “late October” moves money out of a quarter. Investors who own the stock for the January rule are owning a 2027 story. Investors who own it for a beat this quarter are owning a different clock, and the analyst who likes the stock just moved that clock back. Know which meeting you are in. The same man can be right that the rule helps the mine and right that this quarter will disappoint people who used his old sheet. Those are not contradictory. They are sequential. Sequence is the whole risk.
Cash is the other number that must not be blended. For the quarter ended June 30, 2026, the company reported cash of C$1.2 billion, up from C$268 million at the end of 2025. On June 9 it closed an oversubscribed offering of 2.25 percent convertible notes due 2031, with gross proceeds of US$800 million. That pile is why a development story can talk about Phase II without an immediate share sale. It is also debt. Zaffino’s later note put cash near $865 million and net cash near $290 million. He said that net figure can fund Sangdong Phase II, an oxide facility, and buybacks. Do not average C$1.2 billion and $865 million into a third cash balance nobody published. One is June 30. One is an analyst’s October snapshot, and the currencies are not even labeled the same way in every recap. The spread between gross cash and net cash is the lesson. The converts are in the building. Spendable surplus is what is left after them. A buyback funded by a convert is not the same thing as a mine funding itself. Phase II that depends on the net cash is a choice. It is not free metal.
The other rocks are not Sangdong
The notes also list work that is not the Korean mill. Treat each as its own stage, or the investor day will feel like five mines when it is one mine and a set of options.
Los Santos, in Spain, is an existing site. The plan described by the analysts is a multi-year deal to recycle a tailings stockpile. Tailings are rock that has already been mined. They can be a faster pound than a new decline. They are also a finite pile, with a grade the old mill left behind. A tailings contract is a bridge. It is not a second Sangdong.
Rwanda is earlier still. The language in the notes is acquire, upgrade, and export existing ore, pre-concentrate, and tailings, with exports hoped for in November. November is a hope until a permit and a port say otherwise. Ore in a stockpile is not concentrate in Pennsylvania. If the company shows a contract and a date, file it under near-term pounds. If it shows a slide, file it under options.
Gentung Browns Lake, in Montana, is the U.S. option. Both notes call the pace slow and steady. Bank of America wants an update on permits and on whether the Department of War will fund or offtake. A U.S. government customer would matter more to the January story than another bag in Korea, because the rule is American. A prospect in Montana does not satisfy the rule. A funded, permitted, built mine might, years from now. Do not let the word Montana do the work of a feasibility study.
There is also molybdenum at Sangdong, and a drilling program described at about 12,000 meters. Molybdenum is a different metal. A drill program is not a reserve, and a reserve is not a second plant. It can be a real option beside the tungsten mill. It cannot be counted as tungsten EBITDA. If Monday spends its time on moly while the 13-hour question is still open, the room is hiding in a side tunnel.
Phase II is the tunnel that matters after Phase I is boring. The company has said it could lift throughput toward 1.2 million tonnes a year and, in the analysts’ telling, about double concentrate. Engineering, capital, underground development, and the gates that raise plant capacity are the checklist Bank of America wrote down. None of those items is a pour. The July offtake does not cover Phase II. Extra tonnes from Phase II need a buyer, a price, and a plant that has first learned to run Phase I for a full day. Doubling a mill that is still tuning flotation is a drawing. Drawings are allowed. They are not shipments.
What you are actually long
Strip the slogan and the position is a list. You are long a Korean underground mine that has made its first concentrate since 1993. You are long a mill that, on the last company letter, was not yet on a full day. You are long an offtake with one Western powder maker for most of Phase I, and long the hope that a second phase finds a buyer of its own. You are long a cash balance that looks huge until the convertible notes are subtracted. You are long the chance that a January rule makes non-Chinese rock more valuable. You are short time, recovery, and the right of a full plant to say it does not need another bag this month.
That list is a tungsten position only in the last line. The rest is a single-asset ramp with a defense label. A single asset can be the right asset. It is still one geology, one workforce, one circuit, and one port. A flood, a permit, a recovery miss, or a customer who takes the contract minimum and nothing more will not be saved by a paragraph in the Federal Register. The Register does not run the night shift.
Compare that with what the early 2025 rally was buying. Beijing tightened exports of tungsten and of other defense metals. The shares of miners who might someday stand outside that fence went up together. They have since given a large part of that move back. The giveback is not proof the fence came down. It is proof the market can reprice hope faster than a mill can learn an ore. Almonty is further along than a hope. It has bags. Bags are the start of the proof, not the end. A stock that still trades as the whole Western answer is being asked to carry a fence that one mine cannot hold. China still supplies the majority of mined tungsten. One Phase I, even at 210,000 units a year under contract, is a crack in that majority. It is not a replacement. Size the crack. Do not size the country.
There is a cleaner way to say the same thing in a portfolio. If you want the rule, you want every qualified non-Chinese pound, and you will not know the qualified list until primes do the paperwork. If you want Sangdong, you want the hours, the recovery, the ship, and the payable. Those two wants can share a ticker. They should not share a reason. Write the reason in a sentence you can still read if the APT price is $2,200 and the mill is on 16 hours. If the sentence dies, it was the price and the calendar. It was not the mine.
What Monday can settle, and what it cannot
Almonty will host the investor day by webcast at 9 a.m. Eastern on Monday, October 19, 2026. Management has said the operating team will walk through the tungsten market and the mines. Bank of America framed the third quarter as a re-risking quarter, not a victory lap. The useful questions fit on one card.
Is the mill on 24 hours, and since what date? Thirteen hours was the September 29 fact. Anything else is a claim until it has a start time. What recovery is the flotation circuit actually getting, not the recovery in the study? Grinding at full throughput with a young circuit can still ship weak concentrate. A weak bag does not fill a 210,000 unit year.
Did the October 24 cargo leave, or is it still a booking? If it left, when does it become revenue? If it did not leave, what paper is in the way? An export permit is not a detail. It is the difference between a dock and a sale.
Which nameplate is Phase I, 1,100 tonnes a day or 640,000 tonnes a year? When does the company expect to be there, in months, not in “ramp”? What concentrate price, not what APT price, sits under the revenue sketch? APT at $2,700 is a headline. The payable on a bag is the business.
How much of the cash is left after the converts, the ramp, and any buyback? Will Phase II be funded from that net cash, and what does it cost? Is the oxide plant a plan or a bid? Oxide is further down the chain, closer to the powder the rule actually names. A concentrate seller who cannot make oxide is still one step away from the clause. That step can be GTP’s plant in Pennsylvania. It can also be a promise. Ask which.
Monday cannot settle January. It cannot make a prime contractor qualify a new source. Qualification in defense is slow on purpose. The aerospace trade group already said the processing the country needs is a 2028 problem, and maybe later. A Korean concentrate ship in October 2026 is allowed to be early for that qualification and late for a quarterly model. Both clocks are honest. Only one of them pays the quarter. Only the other one is the reason the stock became a defense story. If you need both clocks to strike noon on Monday, you will be disappointed by a meeting that can only wind one of them.
The order boom is a third clock
Bernstein’s European aerospace analyst, Adrien Rabier, has said the rearmament order boom is expected to begin in 2027 and run through 2030. That is a third clock, and it is not Almonty’s mill. An order for a missile in 2028 needs tungsten that was qualified years earlier. It does not need the APT price from a week in October 2026. It also does not automatically need Sangdong. A prime can qualify a recycler, a Spanish tailings line, a Vietnamese or Australian source, or a waiver. The boom, if it comes, is a demand ceiling. It is not a named offtake.
Put the three clocks on the table and the slogan breaks. Clock one is January 1, 2027, when the clause steps back to the mine. Clock two is the mill: 13 hours, then maybe 24, then maybe a ship on the 24th, then maybe revenue in the fourth quarter. Clock three is a defense order cycle that a European desk thinks starts in 2027 and runs to 2030. Wall Street has been writing notes as if the three clocks were one watch. They are three watches. You can be early on the rule and late on the ship. You can be right about rearmament and wrong about which mine gets the contract. The investor day will be filmed as if it synchronizes them. It can only report the middle one.
That is the use of the meeting. Not a new price target. A separation. If Black says the mill is on 24 hours, the ship is booked and dated, and the payable price is still high enough to fund the offtake math, the ramp is doing what the September letter asked. The ban remains a 2027 rule with waivers and a buyer who is not ready. If he says the circuit is still learning, the ship has slipped again, and concentrate bids are soft because the plants are full, then the stock is a 2027 hope sitting on a 2026 ramp that has not cleared the easy part. The hope can still be right. It is not the same position.
What would make this reading wrong
The reading is wrong if the mill has already been on 24 hours for weeks, the October cargo has sailed, and the customer has accepted it at a payable price near the old APT print. Then the ramp is no longer a hope. It is a shipment. You still do not own the January rule. You own a mine that has started to behave like a mine. That is a better fact than a slide. It is still one cargo.
The reading is wrong if the defense clause is waived so widely that origin stops mattering. The Army has already said it expects some impact, which cuts against a total waiver. It does not cut against a pile of mitigation plans. If the practical rule in 2027 is “file a plan and keep buying,” the scarcity premium in the stock is a story about a law that will not be allowed to bite. Watch the waivers, not the speeches.
The reading is wrong if concentrate stays scarce and the price of the bag rises even while APT cools. Then the structure Zaffino described flips, and the mine’s product is the tight market, not the leftover. That would show up in the realized price, not in a chart of Chinese APT. Ask for the realized price. A company that will not say it is asking you to trade the screen.
The reading is wrong if you needed a safe way to own rearmament and you bought one developer to get it. There is no safe tungsten mine in this piece. There is a Korean restart with a large offtake, a cash pile that includes an $800 million convert, a Spanish tailings idea, a Rwandan hope, and a Montana prospect. China still mines most of the metal. Plants in the West are full and slow to take more feed. Each of those can hurt you in a different way. The clause will not refund you.
The reading is wrong if your tungsten view is a price and a date. This piece does not have a target. A forecast that the shares “must” return to $25, or “must” fall because a target was cut to $19, is a mood with a verb. The facts in hand are the clause, the 13-hour letter, a late-October ship that was still expected rather than received in Thursday’s note, an offtake on 90 percent of a plan, two cash snapshots that should not be averaged, and a pair of nameplate figures that do not describe the same year. They do not multiply into Monday’s closing price.
The idea, once
Almonty is being asked to stand in for a date. The date is January 1, 2027, when the U.S. defense rule on tungsten metal powder and tungsten heavy alloy steps back from the furnace to the mine. Chinese or Russian rock, processed somewhere else, does not by itself satisfy that clause. The Army has said January will have an impact. The aerospace industry has said the processing to replace China is more of a 2028 problem. The rule is a real constraint. It is not a purchase order with Sangdong’s name on it.
The mill is the other fact. Sangdong produced its first concentrate since 1993 in late September. On September 29 Lewis Black said the plant was running 13 hours a day, seven days a week, and would go to 24 hours over the coming days. The first bags were ready for export. Oppenheimer still had the first shipment in late October, a bit behind its old sheet, and cut its third-quarter sketch to C$49 million of revenue and C$25 million of EBITDA. A cargo booked for October 24 is a plan. It becomes a business when it is paid. Phase I is aimed at about 640,000 tonnes of ore a year in the company’s words, while a Bank of America note has used about 1,100 tonnes a day. Those are not the same year until someone shows the calendar. The GTP offtake covers about 90 percent of Phase I, at least 210,000 metric ton units a year after ramp-up, for 21 years from first delivery. It does not cover Phase II. It does not cover a quarter that has not shipped.
APT cooled from about $3,140 a unit to about $2,700 in the week Zaffino described, after a summer near $3,000. He also said Western plants are full under old contracts, so concentrate can soften while the finished price looks tight. Almonty sells the concentrate. The target went from $25 to $19 because peer multiples shrank, and the rating stayed Outperform. June 30 cash was C$1.2 billion after an US$800 million convert. A later note put net cash near $290 million. Gross cash is not the fund that builds the next plant.
Monday can tell you the hours, the ship, the recovery, and which nameplate is real. It cannot make the Pentagon a customer, and it cannot make 2027 arrive early. The ban is a date. The mill is a ramp. Buy only the one you can check. The other one does not mill the ore.