← Nomi Prins hub  Β·  Research hub  Β·  Research library

🌎 The Company Keeping American Aluminum Home

The U.S. imports most of the primary aluminum its defense industry and power grid depend on. This month's recommendation is one of the few companies still operating domestically.
2026-AUG-27 · Prinsights Pulse Premium (Substack) · Nomi Prins (ex-Goldman Sachs MD; Prinsights Global) · paid post β€” August monthly issue (body captured via a logged-in session) · β†— Read on Substack · transcript · actionable insights
One-line take: the formal August Pulse Premium recommendation teased the day before β€” Century Aluminum (NASDAQ: CENX), "Action to Take: Consider buying shares… up to $52." The framing is a deliberate move up the value chain: July's Semi-Annual Update closed Constellium (CSTM), the aluminum fabricator, "at a more than double return… in just eight months," and this month rotates to a smelter, "because where a company sits on the aluminum value chain determines how it makes money." The mechanism: aluminum trades on one global benchmark (LME ~$3,225/t), but delivery inside the U.S. costs LME plus the US Midwest Transaction Premium, now close to $2,000/t β€” so an American buyer pays about $5,200/t, roughly 60% above the world price. Imported metal owes the 50% Section 232 tariff; a domestic smelter does not, "so the premium is theirs to keep on all they sell," and because energy β€” a smelter's largest cost β€” does not rise with the aluminum price, "most of that premium becomes margin, far more than a fabricator or a diversified miner keeps." The asset base: Century runs two of only four smelters still producing primary aluminum on American soil β€” Sebree (KY, ~220kt) and Mt. Holly (SC, ~220kt, at full nameplate since end-June 2026) β€” plus Grundartangi (Iceland, ~320kt) on 100% hydro/geothermal power, branded Natur-Al at <4t CO2 per tonne (~a quarter of the industry average, ASI-certified/ISO 14064-verified) β€” ~760kt of capacity with all three running at or near capacity for the first time in over a decade; plus 55% of Jamalco (Jamaica, ~1.4Mt/yr of alumina) for captive feedstock. In February 2026 it sold its oldest plant, Hawesville, to a TeraWulf affiliate for $200M cash for redevelopment as an AI/HPC data-center campus. The numbers: Q2-2026 net sales $752.1M, adjusted EBITDA $326.9M (from $231M in Q1), adjusted EPS $2.46 (vs $2.35 consensus), shipments 130,632t (+6% q/q); cash $343.4M, liquidity $784.9M, $66M of debt repaid in Q2 and cash exceeding total debt by late July; next-quarter guidance $325–345M EBITDA. The growth option: the Inola, Oklahoma JV with Emirates Global Aluminium (EGA 60% / Century 40%) β€” 750kt/yr, ~$4B, up to $500M DOE funding plus >$275M Oklahoma incentives and discounted power; a July 2026 White House proclamation lets Century import up to 300kt/yr at a reduced 25% tariff from 2027, with the savings earmarked to fund its share. The risks: the tariff itself (an Aug 19 Bloomberg report of a 50%β†’25% Canada deal knocked the stock ~6% in a session β€” "even a partial tariff reduction doesn't destroy the thesis"), Glencore owning 30% of the shares and taking ~44% of Q2 sales, execution (Mt. Holly cast-house/carbon issues to be resolved by Q4; a Grundartangi Line 2 failure in Oct 2025), and the Oklahoma AG's federal lawsuit to block Inola. The entry: the stock ran ~$21 β†’ ~$70 then pulled back to ~$44, βˆ’36% from the highs, "largely tariff-sentiment-driven" while "the operating business has only gotten stronger" β€” "the stock is trading as if that premium disappears entirely. We don't think it will."

1. Stocks & names mentioned

TickerNameResearchViewWhat she saidAt
CENXCentury AluminumQT Β· SA Β· STK · FAPositiveThe August Pulse Premium recommendation β€” "Action to Take: Consider buying shares of Century Aluminum (NASDAQ: CENX) up to $52." It "operates two of only four smelters still producing primary aluminum on American soil, and it collects the record domestic premium on everything it sells" β€” imported metal owes the 50% Section 232 tariff, Century's domestic smelters don't, so "they sell into the same U.S. market at the same inflated price, but they keep the premium as margin," and because energy doesn't rise with the aluminum price "most of that premium becomes margin, far more than a fabricator or a diversified miner keeps." ~760kt of capacity across Sebree (KY ~220kt), Mt. Holly (SC ~220kt, full nameplate since end-June 2026) and Grundartangi (Iceland ~320kt, 100% hydro/geothermal, Natur-Al at <4t CO2/t vs ~4Γ— that industry-wide, ASI-certified and ISO 14064-verified β€” an edge as EU carbon border adjustments tighten), with Q2-2026 "the first time in over a decade that all three were running at (or near full) capacity simultaneously"; plus 55% of Jamalco (~1.4Mt/yr alumina) for captive feedstock and "some insulation from the spot market." Q2 was "its strongest quarter in years": net sales $752.1M, adjusted EBITDA $326.9M (up from $231M in Q1), adjusted EPS $2.46 (beat $2.35), shipments 130,632t (+6% q/q), on top of a strong Q1 ($649.2M revenue, $170.7M adjusted net income). Balance sheet "improved dramatically" β€” $343.4M cash, $784.9M liquidity, $66M of debt repaid in Q2 and cash exceeding total debt by late July: "A year ago, Century was a leveraged, sub-scale producer running below capacity. Today, it's net cash positive with all plants at full utilization." Guidance $325–345M EBITDA. Growth option: the Inola, Oklahoma JV with Emirates Global Aluminium (EGA 60% / CENX 40%), 750kt/yr for ~$4B β€” the first new U.S. smelter in almost 50 years, which "would more than double current U.S. primary aluminum output" β€” backed by up to $500M DOE funding, >$275M of Oklahoma incentives and discounted power, plus a July 2026 proclamation letting Century import 300kt/yr at a 25% tariff from 2027 with the savings earmarked for its share; construction targeted to begin by end-2026, production by end of decade. Risks, in her order: the tariff (an Aug 19 report of a 50%β†’25% Canada deal cut the stock ~6% in a session β€” "even a partial tariff reduction doesn't destroy the thesis… at 25%, the tariff still generates a substantial domestic premium"), Glencore at 30% of shares and ~44% of Q2 sales, execution (Mt. Holly cast-house/carbon issues due to clear by Q4; the Oct-2025 Grundartangi Line 2 failure), and the Oklahoma AG's federal suit to block Inola. Entry: from ~$21 to a ~$70 peak and back to ~$44, "about 36% from its highs… largely tariff-sentiment-driven" while EBITDA is at records and the balance sheet is net cash β€” "The stock is trading as if that premium disappears entirely. We don't think it will."read β†—
AluminumAluminum (commodity)β€”PositiveThe structural case behind the pick. LME ~$3,225/t plus a US Midwest Transaction Premium near $2,000/t means an American buyer pays about $5,200/t, "roughly 60% above the world price"; the premium hit a record near $2,180/t in February, "more than triple its level just 2 years earlier." Supply: "the structural deficit is also widening" β€” Gulf smelter outages this spring cut an annualized 2 million tonnes and Mercuria called it "a black swan," putting this year's global deficit at "a minimum of 2 million tonnes"; LME warehouse stock "has roughly halved this year to about 250,000 tonnes, the lowest since 1990"; and China, "the source of roughly 60% of the world's aluminum, cannot lift its output past the 45-million-tonne cap the government imposed… a ceiling its smelters have now reached." Demand: aluminum "is the conductor for the high-voltage lines that carry power across the grid, and nearly every new AI data center requires vast new transmission systems"; EV production alone is projected to lift demand to 31.7Mt from 19.9Mt in 2020. U.S. supply has fallen from ~30 smelters and 4.65Mt in 1980 β€” then the world's largest producer β€” to about 680,000 tonnes, less than 2% of world supply, because smelting takes ~15 MWh per tonne ("roughly what a home uses in a year and a half") and American energy "became too costly to compete." Washington classifies aluminum a critical mineral and tariffs it under Section 232 of the Trade Expansion Act, a 1962 law letting the president restrict imports that threaten national security; the rate reached 50% in June 2025 and in April was widened "to include the full value of each shipment, metal plus fabrication and freight." "If the supply picture stays this tight, the LME price has room to rise."read β†—
CSTMConstellium SEQT Β· SA Β· STK · FANeutralPosition closed β€” the prior aluminum winner, exited rather than re-recommended. In July's Semi-Annual Portfolio Update "we closed Constellium (CSTM), the aluminum fabricator, at a more than double return in our model portfolio in just eight months, as the physical supply squeeze we forecast set in." It is now used as the value-chain contrast: "Constellium was selected because it buys aluminum and fabricates it into aircraft parts, car panels, and cans," whereas the new recommendation "is one of the few companies that still smelt primary aluminum on American soil" β€” and a fabricator keeps less of the domestic premium than a smelter does, since it pays the inflated U.S. price on its own input.read β†—
GLENGlencoreSA Β· STKNeutralNamed only as the concentration risk inside the CENX thesis: "Glencore owns 30% of Century's shares, and accounted for roughly 44% of its Q2 2026 consolidated net sales. Glencore has been a supportive shareholder and partner, but that high concentration is worth noting." No view expressed on Glencore itself.read β†—
WULFTeraWulfQT Β· SA Β· STK · FANeutralPassing mention as the buyer on the other side of Century's February 2026 divestment: an affiliate of TeraWulf paid $200 million in cash for the Hawesville, Kentucky smelter, "being redeveloped as an AI and high-performance computing data center campus." Prins frames it from Century's side β€” "an unusual exit, but one that makes sense. Century got $200 million for its oldest and least efficient plant" β€” with no view on TeraWulf. (Note the theme rhyme: a defunct smelter's power interconnection becoming AI compute.)read β†—
Emirates GlobalEmirates Global Aluminium (private, UAE)β€”NeutralCentury's joint-venture partner on the Inola, Oklahoma smelter, announced January 2026: EGA holds 60%, Century 40% of a ~$4 billion, 750,000 t/yr project that "would be the first new aluminum smelter built in the United States in almost 50 years." Mentioned as the partner and majority owner of the growth option, not as a rated name.read β†—
MercuriaMercuria Energy Group (private)β€”NeutralCited as a research/market source: "one of the world's largest commodity traders" called the spring Gulf smelter outages β€” an annualized 2 million tonnes of lost output β€” "a black swan" and estimated this year's global aluminum deficit at "a minimum of 2 million tonnes." A supply data point, not a rated company.read β†—
NovelisNovelis (private β€” Hindalco subsidiary)β€”NeutralCited as the demand-forecast source: "According to research by Novelis, a leading provider of sustainable aluminum solutions, EV production alone is projected to increase aluminum demand to 31.7 million tonnes, up from 19.9 million tonnes in 2020." A forecast reference, not a rated company.read β†—

Written post β€” no video timestamps; the read β†— link in the "At" column opens the source article. Model-portfolio chart and closed-track-record images from the issue are not captured.

2. Talking points

The January call, cashed once already

Same commodity, different rung on the ladder

The tariff turns into a transfer to the surviving smelters

From 30 smelters to a handful β€” an electricity story

The world is short too β€” and the buffer is gone

Demand keeps climbing β€” grid, AI and EVs

Section 232 β€” the legal machinery behind the premium

The premium, not the price, is the variable

The recommendation β€” Century Aluminum (NASDAQ: CENX)

What a smelter actually is β€” the value chain, spelled out

The three plants β€” and the Icelandic carbon edge

Captive alumina, and the plant it sold to an AI data-center developer

Why the premium matters more than the price β€” and why it's stickier than assumed

The numbers β€” a balance sheet that flipped in a year

Inola, Oklahoma β€” the growth option, and who is paying for it

Risks β€” in her stated order

What moves the stock β€” the risks read forward

Price and the action to take

3. In plain English

CENX β€” Century Aluminum Positive

Century melts a white powder called alumina into raw aluminum metal, using colossal amounts of electricity. That's the whole business β€” it doesn't dig the ore and it doesn't bend the metal into car panels; it does the one step in the middle that turns rock into metal. Two of its three plants are in the United States, and there are only four such plants left in the entire country.

Why that matters right now comes down to one quirk of pricing. Aluminum has a world price set in London β€” about $3,225 a tonne. But if you want the metal physically delivered inside the U.S., you pay that world price plus a surcharge, currently around $2,000 a tonne, because a 50% import tax makes every foreign tonne that expensive. So an American can-maker or carmaker pays roughly $5,200 for a tonne that costs $3,225 anywhere else. Here's the trick: Century's American plants sell at that inflated $5,200 U.S. price but never pay the import tax, because they never import anything. That extra ~$2,000 is simply extra revenue. And since a smelter's biggest cost is electricity β€” which doesn't go up just because aluminum does β€” almost all of that extra revenue drops through to profit. A fabricator like Constellium, by contrast, has to buy aluminum at that inflated price, so it keeps far less of it.

The business is also in the best shape it has been in years. All three plants (two in the U.S., one in Iceland running entirely on hydro and geothermal power) are running flat out for the first time in over a decade. Last quarter's operating cash profit was a record $326.9 million, the company paid down debt, and it now has more cash than total debt β€” a year ago it was over-leveraged and running half-empty. It also sold its oldest, worst plant to an AI data-center developer for $200 million cash, which cleaned up the fleet. On top of that sits a lottery ticket: a 40% stake in a $4 billion new smelter in Oklahoma β€” the first built in America in nearly 50 years β€” with $500 million of federal money and $275 million of state money behind it.

The catch is that the whole premium exists because of a tax the President can change with a signature. When a Bloomberg story on August 19 suggested the tariff might be halved to 25%, the stock fell 6% in a day. Prins' argument is that even a 25% tariff leaves a domestic premium far above anything seen before 2025, that Washington has too much invested in domestic aluminum to abandon it, and that after falling from $70 to $44 the stock is priced as though the premium vanishes completely. Two other things to keep in mind: Glencore owns 30% of the company and buys 44% of what it sells, so one relationship carries a lot of weight; and Oklahoma's Attorney General is suing to stop the new smelter. Her instruction: buy up to $52.

Aluminum β€” the commodity Positive

Aluminum is having a genuine shortage, and for once the reason is simple arithmetic rather than speculation. The metal the whole world can freely buy sits in London Metal Exchange warehouses; that pool has roughly halved this year to about 250,000 tonnes, the smallest it has been since 1990. Outages at Gulf smelters this spring knocked out the equivalent of two million tonnes a year, which one of the world's biggest commodity traders called a "black swan." And the country that makes 60% of the world's aluminum β€” China β€” has capped itself at 45 million tonnes and has already hit that ceiling, so it cannot simply make more.

Demand, meanwhile, keeps growing for unglamorous reasons. Aluminum is what long-distance power lines are made of, and every new AI data center needs a lot of new power lines. Electric cars use far more of it than ordinary cars; one industry forecast has EV-driven demand rising from about 20 million tonnes in 2020 to nearly 32 million.

The U.S. sits on the wrong side of all this. In 1980 it had about 30 smelters and made 4.65 million tonnes β€” more than anyone. Today it makes about 680,000 tonnes, under 2% of world supply, because turning alumina into aluminum takes roughly 15 megawatt-hours of electricity per tonne (about what a house uses in eighteen months) and American power got too expensive versus Canada, China and the Gulf. That is why Washington now treats aluminum as a "critical mineral" and taxes imports at 50% under a 1962 national-security law, and why the shortage looks structural rather than cyclical.

CSTM β€” Constellium SE Neutral

Constellium is the position Prinsights just sold, not one it is recommending. It sits downstream of Century in the aluminum chain: it buys finished aluminum and shapes it into aircraft parts, car body panels and drink cans. Prinsights bought it in late 2025 on the view that aluminum would get physically scarce, and closed it in July 2026 for more than double the money in about eight months.

It appears in this issue as the contrast that explains the new pick. A fabricator has to pay the inflated American aluminum price for its raw material and then try to pass it on to customers, so it keeps only a slice of the tariff-driven premium. A smelter is on the other side of that same transaction β€” it collects the inflated price and pays no import tax. When the premium itself is the opportunity, you want to own the seller, not the buyer. That is the entire logic of moving from Constellium to Century within the same commodity.


Summary derived from the Prinsights Pulse Premium paid post for personal study. Not investment advice. © Nomi Prins / Prinsights for source material.