DK Street Journal

A price increase landed; the shares came off

Containerboard's third price increase of 2026 showed up in the published benchmark last month. Fastmarkets RISI recognized linerboard up $70 a ton and corrugating medium up $100 month over month — the first movement of a round in which Packaging Corporation of America had announced $140, Smurfit Westrock $100 and International Paper $80. Then the cost leg turned the producers' way: old corrugated containers' eleven-month price run ended in October, flat in most regions and down $10 a ton on the Los Angeles and San Francisco docks.

Both items widen the spread between what a mill sells a ton of board for and what it pays to make one. Yet over the five sessions to 2 October, International Paper fell 8.7% to $31.97, Smurfit Westrock 8.3% to $42.54 and Packaging Corp 3.6% to $228.54, while the S&P 500 tracker slipped 0.2% and the average change across 3,297 priced symbols was −0.9%. All three had rolled over from summer uptrends in mid-September. What settles the argument is dated: Packaging Corp reports after the close on 21 October, guided to $2.91 a share excluding special items, and International Paper on 28 October.

Revenue up, profit flat

Packaging Corp makes containerboard and converts it into shipping boxes, retail displays and protective packaging, with a smaller segment selling cut-size office paper. June-quarter revenue rose 14.7% to $2.49bn, helped by the Greif containerboard business bought for $1.8bn in September 2025; operating income was flat at $333.2m. That completes four consecutive quarters of accelerating revenue growth — 6.0%, 10.1%, 10.6%, 14.7% — in which operating income fell year over year every time. The operating margin compressed to 13.4% from 15.4%. Volumes are the best in the group: total corrugated shipments rose 24.3% per day and legacy shipments 4.1% per day to a quarterly record. At 21.8x forward earnings and 13.1x trailing EV/EBITDA it is the most expensive of the three, priced for consensus earnings of $13.61 a share in 2027 against $10.48 this year.

The two cheap ones, and why

International Paper's June quarter was the worst in the group: revenue down 11.3% to $6.00bn, operating income down 78% to $45m — an operating margin of 0.75% — and a $12m net loss. After a $3.52bn loss in 2025 its trailing earnings and enterprise multiples are unusable; the anchors are 0.70x sales, 1.17x book and roughly 10.5x 2027 consensus earnings of $3.04. The business is taking share in a shrinking market, with North American box volumes up 1.7% a day against industry shipments down 1.9%, and it is withdrawing supply it controls: five sites closing for an expected $230m earnings improvement, the Georgetown pulp mill shut, $500m of DS Smith synergies targeted by 2027. "We are taking cost and complexity out of the business," chief executive Andy Silvernail told investors on the 30 July second-quarter call, on which management also cut its second-half demand assumption to generally stable.

Smurfit Westrock, the Dublin-headquartered group formed from Smurfit Kappa and WestRock, grew June revenue 1.1% to $8.03bn while operating income fell 44% to $309m, its margin down to 3.85% from 6.95% a year earlier. Paper markets "are as strong as I have seen in my lifetime within this industry," chief executive Tony Smurfit said on the 29 July call, with the company sold out of nearly every grade. It is the cheapest of the three on capital-structure-neutral measures — 8.6x trailing EV/EBITDA, 1.34x book, a 4.4% free-cash-flow yield, about 12.2x 2027 consensus earnings of $3.50. The two dated items in its sell-off window cut the other way: a $420m agreement on 24 September for CMPC's Chilean containerboard and corrugated assets, tonnage added in South America rather than into the North American balance, and Citi trimming its target to $54 from $57 on 2 October while keeping a Buy.

What the business earns and what it does not

Twelve months do not describe one story: International Paper is down 31.5%, Packaging Corp up 6.0%, Smurfit Westrock up 1.6%. Nor does the economy explain the week — ISM manufacturing printed 54.5 in September with new orders at 55.3, a ninth straight expansion month, while box shipments fell, because boxes are paid for in units of goods shipped, not dollars of factory output. Absent discoverable company news, the likelier reading is a sector-wide de-rating ahead of the October prints.

Only one of the three declines is earned by the numbers. Packaging Corp's step down is the one four quarters of falling operating income support, against a multiple priced for a near-30% earnings jump in 2027. International Paper's and Smurfit Westrock's share declines are explained by their margins, not by anything that happened to tons, price or fiber cost in late September — both of which moved favorably. And the increases stick only because supply was withdrawn: the industry operating rate reached nearly 95% in the second quarter after roughly 3.5m to 4m tons of permanent closures since 2023, and buyers are resisting — the Association of Independent Corrugated Converters publicly opposed this round, noting producers announce large numbers and secure smaller ones.

Packaging Corp has already told investors what the September quarter is worth: $2.91 a share excluding special items, a sharp step up from what June delivered. If a confirmed price increase and a cooling fiber bill cannot produce that, the question stops being why these shares broke in five sessions and becomes what the published index is worth to the companies that set it.

Paid in the year it spends

On 1 January the network service rate charged by American Electric Power's eastern transmission companies rose to $60,050.97 per megawatt-year from $56,289.43. No commission heard argument over it. The figure is the output of a forward-looking formula on file at the Federal Energy Regulatory Commission, computed off the coming year's projected costs, lifting the annual revenue requirement those companies collect to $1.44bn from $1.257bn, with the prior rate year's difference trued up and billed with interest at an approved 9.85% base return on equity plus a 50 basis-point adder — half a percentage point — for membership in a regional transmission organization.

That machinery is why AEP, the Columbus, Ohio holding company whose four segments include a standalone transmission arm, is a different business from most of the regulated owners it trades beside. Forty-two percent of its raised $78bn five-year capital plan is transmission, with transmission rate base scheduled to grow from $32bn in 2025 to $55bn in 2030. For a utility in a buildout, the decisive variable is not how much it spends but how long it waits to earn on it.

Entergy waits for a test year

Entergy, the New Orleans utility serving 3 million customers across Arkansas, Louisiana, Mississippi and Texas, spends through the other kind of mechanism. Its $57bn 2026-29 plan runs mostly through state jurisdictions, and Entergy Louisiana filed its formula rate plan evaluation report with regulators on 1 June 2026 for test year 2025 — this year's rates set off last year's costs. The filing sought $149m of rate change, of which only $15m came through the base plan; riders for transmission and distribution carried $36m each and market-based recovery $40m.

The plan itself has to be renewed. "Yeah, I think that we have a long history of extending Formula Rate Plans within Louisiana," chairman and chief executive Drew Marsh told analysts on the July 29 call. "Our expectation is that we would probably be able to extend."

In the meantime the gap is funded with shares. Entergy's June quarter grew revenue 5.9% to $3.524bn and net income 3.4%, yet diluted earnings came to $1.03 a share against $1.05, because the diluted count rose 4.6% to 466.3m. Management widened its 2026 dilution drag to $0.20-$0.25 a share from $0.15-$0.20. AEP issues equity too — a $3bn forward transaction in the second quarter, which it says covers all anticipated marketed equity needs for the $78bn plan — but it collects a current return on the largest slice of what the money builds.

Who carries the stranded-cost risk

The protection for existing customers lives in contract terms rather than announced gigawatts. AEP's Ohio data-center tariff, effective 23 July 2025, obliges new loads of 25 MW or more to pay for at least 85% of subscribed capacity whether they use it or not, over a ramp of up to four years plus eight, with an exit fee of three years of minimum charges. Entergy's agreement to serve Meta's Hyperion campus runs 15 years, with Meta covering a substantial portion of roughly $3.2bn of gas-plant cost only if a further 15-year contract is signed; intervenor witnesses argued for 25 years to match the plant's life.

What the quarter's selling earned

Neither company gave the market a reason. AEP raised 2026 operating guidance to $6.25-$6.55 a share and lifted its capital plan; Entergy affirmed its range. Yet the eleven largest regulated electric owners fell an average of about 12.8% in the three months to 2 October while the S&P 500 rose 3.3% — AEP down 13.7%, Entergy down 12.3% — and the Utilities Select Sector SPDR fell 6% in September, its steepest month since 2023. The likelier reading is arithmetic rather than news: the 10-year Treasury yield went through 5% in mid-September and reached 5.28% on 2 October, while allowed returns are fixed by commissions for years. NextEra showed what that costs at the margin, selling $3.75bn of junior subordinated debentures on 22 June at initial fixed rates of 6.000% to 6.625% against the 10.95% return Florida allows its utility through 2029.

So the discount rate explains the direction for both, and the multiples decide the rest. Entergy carries the group's richest forward earnings multiple at 22.9x, against a peer median near 17.3x, and 2.51x book, while delivered per-share earnings go backwards and the rate base it is buying waits on a test year — a de-rating the business has earned. AEP sits at 18.7x forward against 20.5x trailing and 2.02x book, with guidance raised twice and the shortest recovery lag of the two; that part of its decline is harder to attribute to anything the company is doing. One further constraint bounds both plans from upstream: heavy-duty gas turbines from the three dominant makers are sold out into the end of the decade, with reservation slots stretching to 2031 delivery.

Entergy's next Louisiana rates will be set off a 2026 test year, filed next June. AEP's transmission rate resets again on 1 January, trued up and billed with interest, whatever the 10-year yield happens to be doing by then.

WEG, the Brazilian maker of electric motors, generators and transformers, looks on American screens like the cheapest grid-equipment supplier in the world: a little over six times next year's earnings, at a moment when every peer selling switchgear into data centers trades near thirty. The two figures in that ratio are not in the same money. The price is in dollars; the earnings estimate behind it is in reais.

Convert the estimate at the October 2 rate of 5.2225 reais to the dollar — the real is down 6.78% across 2026 — and WEG is roughly 33 times forward earnings, the dearest of the eight foreign grid suppliers an American can buy over the counter rather than the cheapest. The same unit mismatch runs through almost every one of those lines, and fixing it reorders the group that has become the main way to own electrification demand from a US account.

The wrapper is honest; the screen is not

A depositary receipt pays its holder the local share's return, multiplied by the currency, divided by the number of ordinary shares each receipt represents, net of fees. That machinery works. Siemens' receipt and Prysmian's — two independent euro lines, each one receipt to half an ordinary share — implied exchange rates within 0.4% of each other on every sampled date over the past year, and within 0.4% of the 1.1252 euro spot print for October 2. Schneider Electric's unsponsored line, five receipts to one share, prices within 0.3% of the €303 Paris close. Hitachi's one-for-one and Mitsubishi Electric's one-for-two lines imply the yen within 0.5% of each other. WEG's reconciles to the decimal: Brazilian shares up 42.4% over twelve months, compounded with a 3.3% stronger real, produce the 47.1% the US line printed. No ratio change was detectable on any line across seventeen sampled dates in the past year.

The cost leg is trivial. Statutory German and Swiss withholding of 26.375% and 35% falls to 15% under treaty when the depositary files, and service fees run one to five cents per receipt per dividend — tens of basis points a year against moves of 30% to 47%.

What is broken is the arithmetic printed alongside the quote. Hitachi, the Tokyo conglomerate whose energy arm supplies transformers, breakers and grid management, shows a forward price-to-earnings ratio of 0.17; at 157.83 yen the real figure is about 26x. Mitsubishi Electric, whose infrastructure segment sells transmission and distribution systems, shows 0.13 and corrects to roughly 20x — the cheapest name here. Siemens corrects to about 21x, Prysmian, the Milan submarine-cable maker, to roughly 26x, and Schneider Electric — which reported record first-half results with 14% organic growth and raised guidance — to nearly 31x. ABB, the Zurich electrification and robotics group, needs no correction at all, at 29.6x, because it reports in dollars even though it lists in francs. Its second-quarter electrification orders rose 58% on data-center demand.

One line really is plumbing

Siemens Energy's US quote, SMEGF, is not a receipt at all: it is the Frankfurt ordinary share traded in New York, one for one, with no depositary and no fee. The company's actual programs trade elsewhere, under separate symbols. The line turns over 7,130 shares and $1.19m a day, against $22.4m for Siemens' receipt, and eleven of its last sixty-four sessions printed under a thousand shares. Its October 2 close of $171.09 was set by a 100-share trade implying a euro of 1.1765 — 4.6% above spot. Correct that print and the 90-day anchor behind it, and the twelve-month return is about 28.8%, not the 36.1% on the screen. Nothing was stale: there were zero unchanged closes in those sixty-four sessions. The defect is a thin, discontinuous book whose closing trade can land anywhere inside a wide band.

Strip the noise and the underlying business is the opposite of the chart. Siemens Energy fell 13.5% in Frankfurt over 90 days, including an 8.0% Xetra slide to €132.66 on September 14 as reports of a Siemens AG stake sale landed on top of a selloff in anything selling into data centers, after Microsoft joined OpenAI and Anthropic in endorsing slower frontier-model development. Over those same months it posted record orders of €17.9bn and confirmed 14–16% revenue growth with margin at the upper end of 10–12%. Quarterly operating margin has gone from 2.7% a year ago to 13.0%. "Execution of all that backlog will be a significant stretch to the industry," chief executive Christian Bruch said in February.

WEG runs the other way. Second-quarter revenue of R$10.14bn was down 0.6% year on year after a 6.1% decline in the first quarter, and operating margin compressed from 19.8% to 17.1%, with EBITDA down 2.1% even as return on invested capital improved to 33.6%. Management attributes the softness partly to a stronger real reducing reported results — the same appreciation that added three points to the American holder's year.

The verdict

Equal-weighted, these eight lines rose about 32% over twelve months, of which four to seven points was currency: a sixth of the move, not the move. So the trend reversals across the group are mostly business, and the September de-rating is a dated sector shock rather than a wrapper artifact. But the relative-value ranking every US screen prints is unusable: it makes the one name with shrinking revenue and compressing margins look like the bargain, and the one with 63% operating-income growth look expensive. Siemens Energy's cheapness is real and its chart is partly fiction; WEG's advance is real and its six-times valuation is fiction.

The practical test is smaller than the thesis. Before trusting a high or a low on one of these New York lines, ask how many shares made it — on October 2, for Europe's biggest turbine and transformer maker, the answer was one hundred.

The only line that moves with new reactors

Three companies were pitched as the clean way to own an American reactor buildout, and only one item in their combined accounts actually tracks new reactors: the Westinghouse Electric stake that Cameco and Brookfield Renewable hold between them. In the June quarter, Cameco's 49% share of Westinghouse carried a $10m loss, against $126m of earnings a year earlier.

That matters because Westinghouse is the only place an AP1000 reactor sale can appear in any of the three sets of books. Cameco, the largest listed uranium producer, carries it as equity income rather than revenue. Everything else on this shelf is pounds sold under utility contracts signed years ago, United States Navy propulsion reactors on government contracts, and hydroelectric dams. Since early September all three have been marked down together — roughly 11% on an equal-weighted basis in thirty days, with BWX Technologies closing within half a percent of its twelve-month low — as though one story were unwinding. The accounts describe three.

Cameco: the mine got better, the proxy got worse

Cameco's share of Westinghouse adjusted earnings before interest, tax, depreciation and amortization fell to $163m from $352m, which the company attributes mainly to the absence of roughly US$170m of Dukovany project revenue booked a year earlier; full-year guidance is US$370–430m. More telling, Cameco has eliminated the five-year 6% to 10% adjusted-EBITDA growth outlook it previously attached to Westinghouse's core business plus contracted new-build — the one explicit financial expression of the reactor thesis, withdrawn. What remains is conditional: the Department of Energy's US$17.5bn loan commitment of 23 June 2026 for long-lead items on up to ten AP1000 units, and an $80bn figure that is a vesting condition on the Commerce Department's participation interest rather than an order book.

The uranium business did the opposite. Realized price reached US$67.79 a pound, up 15% from a year earlier; full-year realized-price guidance was raised to C$91–96 from C$85–89 and production held at 19.5–21.5 million pounds, with only 0.2 million third-party pounds bought in the first quarter. The gap to a spot indicator near US$89.60 and a record US$96 long-term indicator is the contract book working as designed. "We are in the business of being disciplined and looking forward to capture that long-term value with those utilities who have come to realize that security of supply is important," chief executive Tim Gitzel told investors on 31 July. Group net income still fell 92% to $25.2m, and at 57.9x forward earnings — against roughly 65x in May, and consensus 2026 revenue growth of 1.2% — the de-rating is earned by the equity line, not the mine.

BWXT: the order book never heard about it

BWX Technologies, which builds naval reactors and fuel for the Navy's propulsion program alongside commercial steam generators and small-reactor components, has no commercial new-build problem. Government Operations was $601.3m of $901.6m of June-quarter revenue and grew 2%; Commercial Operations was $302.5m and grew 72%. Backlog reached $8.4bn, up 40%, on a 1.7 times book-to-bill, and all four 2026 guidance lines were raised on 3 August, with commercial growth lifted to about 45%.

The honest complaint is margin. Government work earned a 20.9% segment margin last quarter against 11.9% commercial, so the mix that produces the growth compresses the group: operating income fell 12.2%, a fourth straight quarterly decline, and annual operating margin has ground from 16.3% in 2021 to 10.1% in 2025. The company is also selling its medical-isotope business to Nordic Capital for up to $800m, where chief executive Rex Geveden said BWXT had "roughly tripled revenue, improved profitability and significantly expanded the product portfolio, including novel therapeutic isotopes" since 2018. At 28.4x forward earnings against roughly 46x in May, with consensus 2026 earnings up 24%, the shares have given back a third of their multiple while the order book grew.

Brookfield Renewable was never the nuclear name

Brookfield Renewable's Westinghouse exposure is roughly 11% net to the partnership; the 51% is held with institutional partners. The segment housing it contributed $44m of a record $421m of quarterly funds from operations and fell from $74m, while hydro delivered $336m. Group funds from operations rose 13.5%, though roughly $175m came from asset-sale gains, and the new capital is going to batteries — a $3bn purchase of the Aypa storage platform, about $420m net. Its break is a discount-rate event: a 5.26% distribution yield against a 30-year Treasury that reached 5.61% on 29 September, the highest since 2002. At 9.55x trailing enterprise value to EBITDA and 1.94x book, with consensus expecting reported losses through 2027, there is no forward earnings anchor to cheapen.

What the sell-off actually priced

Cameco's markdown is defensible on its own numbers, and it is Westinghouse that does the work. BWXT's is not explained by anything in its order book; margin mix and a withdrawn growth premium are the candidates, and the likelier reading of a 40% backlog gain meeting a one-third multiple cut is that buyers stopped paying for reactor optionality. Brookfield Renewable belongs to the bond market this quarter. Treating the three as one position was the error the drawdown exposed: over twelve months this corner held up better than the rest of the complex, where NuScale fell about 80% and Oklo about 72%.

The reactor case always rested on units in service around 2033. Until a utility signs, the shares are paid for pounds, submarines and water.

Within three weeks this summer, each of Europe's three listed high-voltage cable specialists told investors the year would be better than the spring plan. Nexans raised its earnings range on 29 July, Prysmian the next day, NKT on 13 August. Since then all three have traded between 17% and 20% below their 2026 highs.

What links them is a product almost nothing else sells: the submarine and onshore cable for the direct-current links that move power between countries and bring offshore wind ashore. The order books are long — NKT's transmission book runs to €13.0bn, Prysmian's to €17bn — and that length is what is being questioned. A high-voltage contract is priced years before the copper is bought, built against factory capacity that may not exist yet, laid by a vessel that can be late, and collected from a public counterparty against progress. Signed is not collected.

The one with a real problem

Nexans, the €6.0bn French group reshaping itself around electrification, is the member whose numbers match its share price. Its adjusted backlog was €7.7bn at 30 June, unchanged from December, with visibility only to 2028, and group adjusted EBITDA margin was flat at 11.9% of standard sales against 12.0% a year earlier — the only one of the three that did not expand. First-half net income fell 40.8% to €103m as disposals and impairments from the exit of non-electrification businesses ran through. "Our first-half performance reflects the continued disciplined execution of our strategy in an environment where the structural drivers of electrification remain stronger than ever," chief executive Julien Hueber said on 29 July.

Inside that flat book sits €1.2bn tied to the Great Sea Interconnector, the Cyprus–Crete–Greece link Nexans was to supply and install. The company has withdrawn tenders on the project, and Greece's transmission operator IPTO has said it would stop paying absent remaining decrees from the Cyprus Energy Regulatory Authority. Consensus now models Nexans' 2026 EBITDA down 16.1% to €716m — the only down year in the group. It is also the cheapest, at 12.4x trailing EV/EBITDA and 18.4x forward earnings.

The two without one

NKT, the Danish pure-play that makes both cable and the ships that lay it, did the opposite. Its Transmission operational EBITDA margin hit a record 20.2% and the group margin rose to 15.7% from 14.5%, and on 13 August it lifted guidance to revenue of €2.65–2.75bn at standard metal prices and operational EBITDA of €400–430m. Backlog did slip, to €13.0bn from a record €13.5bn three months earlier — because work was delivered. "The Champlain Hudson Power Express project in North America reached commercial operation, a historic milestone for NKT," chief executive Claes Westerlind said the same day.

That delivery also shows why the revenue line misleads. NKT's reported second-quarter revenue was €946.9m, flat year on year; at standard metal prices it was €657m, down 9%. The €290m difference — roughly a third of the reported top line — is copper and aluminium passed to customers at no margin, and London Metal Exchange copper set a record $14,455 a tonne in August.

Prysmian, at €37.9bn of market value far the largest, carries the same arithmetic plus acquisitions: second-quarter revenue rose 24.1% to €6.06bn against 9.4% organic growth, the gap being metal and the Encore Wire and Channell deals, which add short-cycle American building wire and connectivity with no grid content. Transmission grew 7.3% organically in the half on a €17bn backlog, and chief executive Massimo Battaini called the quarter, with its record €730m of adjusted EBITDA, "the best quarter ever of Prysmian" on 30 July.

Both ends of the cable

If interconnection demand were cooling, the suppliers at either end would know first. Hitachi Energy builds the converter stations and transformers; inside its Tokyo parent, group orders rose 87% to ¥1,906.3bn in the June quarter and energy revenue grew 37% to ¥911.9bn with adjusted EBITA margin at 14.2%. In June it won €770m of converter-station work on the 600MW Italy–Tunisia Elmed link. Hitachi sits 5.1% below its 2026 peak, the one name here that has not been marked down, on 13.6x trailing EV/EBITDA.

What the shares did

The drawdowns are broad, and the cable makers are the mild end: Prysmian 17.3% below its 2026 high, Nexans 18.9%, NKT 19.7%, against Siemens Energy at 22.5%, Sumitomo Electric at 29.8% and Korea's HD Hyundai Electric at 52.3%. Twelve-month returns remain large — NKT 39.1%, Prysmian 47.5% — which is what the unwind of a crowded position off an extended base looks like. Kepler Cheuvreux's upgrade of NKT to Hold, with a DKK 895 target, cited the roughly 23% fall since May rather than any change to estimates.

The verdict

The markdown is uniform; the businesses are not. NKT and Prysmian have handed buyers nothing to mark down — records on margin, books near highs, guidance raised, demand corroborated by the equipment at both ends — so what their shares have lost is the premium of early 2026, and NKT still carries the group's highest valuation at 20.4x trailing EV/EBITDA and about 29x consensus 2026 earnings, converting euro estimates at roughly 7.46 kroner. Nexans has earned its discount, and its Cyprus line is the live demonstration of what the whole group is being priced for: an order book is a claim on a regulator's signature and a ship's schedule, not cash. The usable consequence is that these three can no longer be read off one revenue line.

NKT is spending €2bn so that more of the chain belongs to it — the Karlskrona plant, the Cologne expansion and its own cable-laying vessel, the NKT Eleonora, now being outfitted in Norway. All three are due next year, which is also the first year consensus expects NKT's earnings to step up.

Two companies own overlapping dirt in and around the Delaware Basin of West Texas and New Mexico, pitch the same power developers and hyperscalers, and reported their second quarters on the same day. They are paid in almost opposite currencies.

Texas Pacific Land, the Dallas owner of roughly 880,000 West Texas acres that collects perpetual oil-and-gas royalties and sells water to Permian operators, took 59.2 cents of every second-quarter revenue dollar as a royalty on produced barrels, according to its earnings release. LandBridge, the four-employee Houston owner of about 325,000 surface acres, took 78.1 cents of its dollar as surface-use royalties and fees, and only 5.4 cents as an oil-and-gas royalty.

That split decides what each company is actually selling into the power and data-center buildout. A royalty owner is levered to barrels priced every day. A surface owner is paid for right-of-way, brackish water, produced-water handling and well pads, fees that track activity rather than price. The buildout is sold as a single claim on West Texas land. It is two different cash machines, and the market is currently paying more for the smaller one.

A record royalty quarter, marked down 12% in two sessions

Texas Pacific's barrel went the right way. Its oil-and-gas royalty line rose 53% year on year to $145.6m, on royalty production up 20% to 39.7 thousand barrels of oil equivalent a day and a realized oil price of $97.55. Water sales rose 55% to $39.7m and produced-water royalties 21% to $37.1m. Total revenue reached a record $246.1m, up 31%, at an 88% adjusted EBITDA margin, with free cash flow up 20%. Chief executive Tyler Glover said the quarter "delivered record results across major financial and operating metrics and achieved significant milestones within key growth initiatives."

The shares fell 8.2% the next session and 12.0% over two, on a revenue figure that missed consensus despite the record. The stock is 28.3% below its 31 March high, and its 50-day average has sat below its 200-day since mid-August. A second identifiable weight landed in the spring: Murray Stahl, chief executive of Horizon Kinetics, which reports a 14.5% stake, died on 7 April, and the stock dropped 15.7% the following session.

The third weight is the forecast. West Texas Intermediate closed at $91.11 on 2 October, up roughly 50% in a year on supply loss, with reduced Strait of Hormuz flows and regional production cuts driving third-quarter inventory draws of about 3 million barrels a day, per the Energy Information Administration's September outlook. That same agency forecast WTI averaging $51 a barrel for 2026. Nobody capitalizes a royalty stream at a price the official forecast disbelieves.

What a surface fee is worth when crude is $91

LandBridge's surface-use royalties rose 52.6% year on year to $52.2m and 41% sequentially, on higher produced-water handling volumes, at an 89% adjusted EBITDA margin. Its oil royalty line moved $0.6m, to $3.6m. Chief executive Jason Long told the 6 August call the company has "access to approximately 13.4 million acre-feet of brackish groundwater today, which is more than sufficient to meet long-term water needs for multi-gigawatt data center projects," and said LandBridge "uniquely aggregates the critical elements of data center development." Seven counterparties representing more than 10 gigawatts are under letter of intent or in late-stage talks. One is signed: PowerBridge holds an option on about 3,400 acres in Reeves County for a campus with up to 2 GW of co-located generation, bought with a $2.6m non-refundable fee and expiring in March 2027. A September 2025 agreement with NRG Energy contemplates 1,100 MW of gas-fired power, contingent on NRG landing a power purchase agreement.

Texas Pacific is on the same ground chasing the same buyers. Management described advanced talks on 25 gigawatts of projects, and Glover said he would be "very disappointed if we don't have a definitive agreement to announce in the very near term." It already has one: Project Kilby, where it supplies land and brackish water to Chevron for a plant serving a Microsoft data center of nearly 2.7 gigawatts, first power expected in 2028. It has also bought land outside its royalty acreage, in Shackelford and Jones Counties.

The premium buys indifference, not contracts

LandBridge is the dearer asset on the comparison that survives their different balance sheets, at 46.6x trailing enterprise value to EBITDA against 31.1x, with net leverage targeted at 2.0 to 2.5 times and a revolver recently lifted to $375m, against a Texas Pacific balance sheet carrying no debt. It is also the only name here whose forward price-to-earnings, 44.5x, sits above its trailing 37.1x. Its sponsor holds 57.5m Class B shares and a 79.9% voting interest, so the traded stock is a minority of the economics. Texas Pacific's 36.6x forward earnings capitalizes consensus 2026 earnings per share of $9.30, up a third, an estimate resting on a single analyst. Against its own history the de-rating is real: 26.1x price-to-gross-profit now, against 39.9x in early May.

The premium on the fee stream is defensible, because a surface dollar does not care whether the barrel is $91 or $51, and that indifference is worth paying for. What the premium does not yet buy is contracted cash. Option fees and letters of intent sit on one side, a signed Chevron supply deal on the other, at a 50% higher multiple. And the part of Texas Pacific's markdown that nothing explains is the water and surface lines, the ones crude does not touch, growing more than half and being de-rated alongside the royalty.

The cleanest test on the calendar belongs to the cheaper half's rival. PowerBridge's initial option on those Reeves County acres lapses in March 2027, and the market is already paying for the version where it does not.

Bloom Energy's fuel cells were selected in late September for Aligned Data Centers' 2-gigawatt "Project Phoenix" campus at Shippingport, Pennsylvania, and days before that the company replaced Molson Coors in the S&P 500. Fluence Energy, which integrates grid-scale battery systems, spent the same month cutting fiscal-2026 revenue guidance to roughly $2.4bn from a midpoint near $3.0bn, blaming delays ramping its Houston factory. Caterpillar and Cummins announced nothing.

Underneath the month sits a question of attribution: which of these companies is actually paid for on-site data-center power, and how much of each one's money comes from it. The answer nearly inverts the share prices. The two incumbent engine makers have the fastest-growing power businesses on the shelf and the worst-behaved shares; the fuel-cell company, the only one of the six with no construction, mining or truck-engine core, got the index seat and the order.

What Bloom actually sells twice

Bloom Energy builds solid-oxide fuel cells that convert natural gas or hydrogen into electricity without combustion, then sells a long-dated service contract on the same hardware. The service line was historically the loss-maker, because the stacks wore out. Service margin reached 22% in the June quarter, up 977 basis points year on year, the fifth straight quarter in double digits, because stack life went from nine months in 2018 to roughly five years, with annual price escalators written into the contracts.

The product side scaled at the same time. June-quarter revenue was $1.065bn, up 166%, with product revenue up 215% and full-year guidance raised to $3.9-4.2bn. Net income was $196.3m against a $42.6m loss a year earlier; operating margin reached 17.1%, almost entirely through leverage over fixed cost, since gross margin slipped to 33.4% from 36.1%. Product backlog stands near $6bn, about two and a half times a year ago, inside a roughly $20bn total. "We have visibility currently based on what we are working for 25 gigawatts of deployments," founder-chief executive K. R. Sridhar told investors on the July 28 call.

Two things qualify that. Diluted shares rose 39% year on year, so the ramp was partly equity-funded. And the stock now carries 107.1x forward earnings and 87.6x trailing gross profit, against 60.3x in late August — consensus needs revenue to reach $4.11bn this year and $6.91bn next for that to compress. Fuel is the physical limit: Matthew Smith of Chronometer Partners, speaking on a July 21 podcast, put a latest-generation Bloom unit's appetite at roughly 150 million cubic feet of gas a day per gigawatt and judged that at 2 to 5 gigawatts a year of capacity "there isn't the gas for that unless you take it from something else."

The incumbents were marked down on politics

Caterpillar's June quarter was its first above $20bn, revenue up 24%, operating margin 20.9% against 17.6%, and a record $72.1bn backlog, up about 92% year on year, some Power & Energy customers ordering into 2030. Power generation revenue grew 72%. The company cut its full-year tariff estimate to about $2.2bn from $2.5bn. "No one is slowing down at the moment," chief executive Joe Creed said on the call.

What broke was the multiple it started from. David Giroux told Barron's in July that Caterpillar averaged about 16x earnings over the past decade before trading near 38x on data-center enthusiasm; at 31.0x forward it is still roughly double the decade norm. The trigger was permitting. Baird downgraded the stock in July, citing New York's moratoria as "a recipe for further multiple compression." New York's governor had barred construction of hyperscaler sites drawing 50 megawatts or more for up to a year — the first statewide pause — after residential power prices rose nearly 68% since 2019. Local opposition blocked or delayed at least 75 projects worth nearly $130bn in the first quarter alone.

Cummins is harder to explain. Power Systems, its genset arm, booked $2.3bn in the quarter, up 19%, at 24.5% segment EBITDA against 22.8% — the highest-margin line in the company, and about a quarter of revenue against the Engine segment's third. There is no freight downturn in the print: the company raised group guidance to 10-13% growth and Engine guidance to 9-14% on improving North American truck demand. The blemish was cost, not orders — EBITDA margin fell to 17.5% from 18.4% on incentive compensation, freight and research spending, and earnings per share came in at $6.73 against $7.26 expected. The shares sit 27.4% below their high at 17.9x forward earnings, with consensus 2026 earnings up about 30%.

The storage leg is a margin story

Fluence holds a record backlog of about $6.4bn while reported gross margin fell to 5.1% from 14.8%, and now guides to an adjusted loss near $200m. Eos Energy, which makes zinc-based grid batteries, holds a record $807m backlog across about 3.4 gigawatt-hours and has never sold a battery above cost — gross margin was -71.0% last quarter, improved from -203.2% a year earlier, against $305.5m of unrestricted cash and $191.8m of first-half operating outflow. Tesla, the only profitable storage comparison, deployed 13.5 gigawatt-hours, up 41%, while energy gross margin fell to 20.4% from 30.3% on a vendor cell warranty charge; at $3.1bn of $28.2bn in revenue the energy line is 11% of a company valued at 221.9x forward earnings for automotive reasons.

So the verdict splits three ways. Bloom's revenue, service margin and converting backlog earn most of its re-rating; the final stretch, from about $206 to near $290 in under a month, coincides with mechanical index buying and leaves a multiple that only the 2027 consensus justifies. Caterpillar's de-rating is paid for by where its multiple began, not by its order book, which is growing faster than it has in years. Cummins is the one nothing in the current numbers accounts for: its data-center business is its best, its guidance went up, and it trades near half Caterpillar's forward earnings multiple. And Fluence and Eos de-rated on the cost of making the product, with their order books at records.

New York's pause applies to sites pulling 50 megawatts or more from the grid. A campus that generates its own power is not one of those — which is exactly what Bloom is being paid for, until somebody has to permit the pipeline that feeds it.

onsemi's factories ran fuller in the June quarter than in March, and the company told investors there is room to run them fuller still without buying another tool. Utilization rose to 83% from 77% as the company ramped against a rising backlog, and management put the ceiling at 92% to 93% on equipment already installed.

That is the whole argument in power semiconductors. onsemi — a maker of power switches, energy-conversion parts, sensors and analog chips sold into cars, factories, solar and, increasingly, AI data centers — does not get paid for electrification in any way its revenue line makes legible. It earns the spread between a fixed base of fabs and silicon-carbide substrate operations committed years ago and the volumes customers actually pull today. Utilization, capital intensity and channel inventory decide whether a falling share price means lost demand or unabsorbed capacity. From its July 2 high of $91.22 the stock lost about 26% into a September 17 low; every one of those three operating measures moved the other way.

The absorption math

June-quarter revenue was $1,604m, up 9%, with GAAP gross margin of 38.4% and a non-GAAP figure of 39.3% — up eight-tenths of a point and a fourth consecutive quarter of expansion. Operating margin reached 16.1% and diluted earnings were $0.56 a share. Operating income grew 33.7%, more than three times the rate of sales. Capital spending was $34m, or 2.1% of revenue, with full-year intensity now guided below 5%. Distributor inventory drained to 10.1 weeks from 10.8 as sell-through outran sell-in, and company inventory fell nine days to 192.

"We delivered revenue, gross margin and earnings per share above the midpoint of guidance, reflecting strengthening demand, particularly across AI-driven applications," chief executive Hassane El-Khoury said in the August 3 results release. Silicon-carbide power-device revenue rose 12.5% sequentially to $215.7m, with China automotive silicon carbide guided up 60% to 70% this year. The base all of this is measured from was grim: 2025 revenue fell 15.3% to $5.995bn and gross margin collapsed to 33.1%. onsemi is the number-two merchant supplier of power silicon carbide, roughly 20% to 25% of the market behind STMicroelectronics, and vertically integrated into substrates since its 2021 purchase of GT Advanced Technologies.

The broad-line maker says the channel is empty too

Vishay Intertechnology, which sells MOSFETs, diodes, resistors, inductors and capacitors mostly through distributors, is the natural check: capacity expansion, no silicon-carbide story. Its June quarter produced a book-to-bill of 1.32 — 1.23 in semiconductors, 1.40 in passives — and backlog up 18% to $1.9bn, equal to 6.1 months of shipments, on revenue of $888.6m at a 23.3% gross margin. Distributor inventory fell to 18 weeks from 20 while point-of-sale rose 20.5% year on year. "The weeks of inventory has gone down each of the quarters," chief executive Joel Smejkal told investors on the August 5 call. "So the distributors, even though they speak about the need to replenish, they haven't been able to do it." Vishay's shares still fell about 17% over three months, and its 2026 capital budget of $400-440m — half of it a 12-inch German fab now in installation — is near 11% of consensus revenue.

Where the empty fab actually is

Wolfspeed, the pure-play wide-bandgap maker with 2,371 employees, is the genuine case. Fiscal fourth-quarter revenue was $149.6m, down 24.1%, with GAAP gross margin at -25% attributed explicitly to underutilization. It needs roughly $800m of annual revenue to break even and is running about $200m short; the materials business meant to be the moat shrank 45% to $43.3m. It emerged from a prepackaged Chapter 11 on September 29, 2025, cutting debt from about $6.7bn to roughly $2bn on a share count near 48m that could dilute past 90m. Its shares trade at 2.76x trailing sales and 2.85x forward — the forward figure is higher because the sell side models fiscal 2027 revenue falling to $643m. Industry-wide, front-end silicon-carbide utilization fell to roughly 50% in the 2024 correction while Chinese 6-inch substrate prices dropped from about $1,500 a wafer to about $500.

The verdict

All three peaked within a day of each other on July 2 and troughed in mid-September; the October 2 rebound was complex-wide, with Texas Instruments up 4.4% and STMicroelectronics 7.0% on a Mizuho note arguing Nvidia's Vera Rubin ramp arrives sooner and lifts power content per rack from about $15,000 to about $115,000 under 800-volt distribution. onsemi's analyst day, where it set 2030 targets of about $11bn revenue and 53% gross margin, landed hours before the Federal Reserve's first rate increase since 2023, to 3.75%-4.00%, and investors marked the targets down.

Wolfspeed's de-rating is earned by its own income statement. onsemi's and Vishay's are not explained by anything in their order books, and the likelier reading is rotation out of long-dated automotive and industrial earnings at a higher discount rate. But the cheapness at both is forward-dated and conditional: onsemi's 53.7x trailing earnings, down from 340.9x in May as trough quarters roll off, becomes 26.5x forward and about 18.9x on 2027 consensus of $4.50 a share only if earnings roughly double, and Vishay's 21.4x on 2027 estimates requires the German fab to convert on schedule.

The idle-fab story in power semiconductors is real; it is just that the idle fabs belong to Wolfspeed. onsemi's third-quarter report, expected around November 2, is where the rest of the premise gets a number attached to it.

A dead smelter's interconnect, repriced

Century Aluminum, which runs primary aluminum smelters in the United States and Iceland, fully curtailed its Hawesville, Kentucky plant in August 2022 and never found power cheap enough to restart it. In February it sold the site — more than 750 acres with roughly 480 megawatts available through existing high-voltage transmission and an energized on-site substation — to TeraWulf for $200m in cash plus a 6.8% minority interest in the affiliate developing a data center there. On July 6 TeraWulf disclosed that a subsidiary had signed a 20-year lease with Anthropic for about 401 megawatts of critical computing load at the same address.

That is the clearest price yet on the input that decides whether a US smelter exists at all, and it is why the metals shelf has stopped trading as one thing. The grid buildout is a customer for copper and electrical steel and a competing bidder for the electricity aluminum needs. Alcoa reports third-quarter results on October 15 and Cleveland-Cliffs on October 19, which is when the two sides of that arithmetic next get marked.

The cost side

A single smelter draws roughly 11 terawatt-hours a year, about what a city the size of Boston uses. Century's own slides put power at the Indiana hub serving its Sebree smelter at $57 per megawatt-hour so far in 2026 against a $58 spot, up from $31 in 2024 and $44 last year, and the company guided to $10m to $15m of energy headwind in the third quarter. The Aluminum Association reckons a new smelter needs a 20-year contract at no more than $40 per megawatt-hour; power in the four states holding idle smelter capacity averaged $73.42. Microsoft conceded $115 per megawatt-hour to restart Three Mile Island.

And yet the quarter was the best in Century's history. June-quarter revenue rose 19.7% to $752.1m, gross margin reached 30.3% against 5.8% a year earlier, adjusted EBITDA hit $327m and cash now exceeds total debt. Mt. Holly returned to full capacity for the first time since 2015 on power secured through 2031 from Santee Cooper. Chief executive Jesse Gary said the new tonnes were arriving "into as strong a market as this industry has seen in a very long time," on the second-quarter call.

Alcoa's quarter was the same story at scale: its highest quarterly revenue in a ten-year corporate history and record aluminum-segment EBITDA of $1.1bn at a 32.3% margin. Alcoa is also insulated from the power squeeze for now — under 1% of its electricity is spot-priced, with the rest contracted or hedged — and the cost pressure that did show up was diesel and instability at its Punjab refinery, not electrons.

Both shares peaked on June 2 and both closed at 2026 lows on October 2, down 49.9% and 48.2% from that shared high. London aluminum is 18% off its June four-year peak at $3,115 a tonne, while the Midwest premium that earlier coverage tracked is back near $1.09 a pound, exactly Century's guidance assumption. Century now trades at 3.5x forward earnings against 5.8x trailing, on consensus 2026 earnings of $10.04 a share versus $0.42 reported for 2025. Alcoa's 6.5x forward against 8.5x trailing is harder to call dislocation: it is funding South32's bauxite, alumina and aluminum assets for $4.1bn upfront with $2.6bn of 6.63% and 6.88% notes plus 17m new shares, roughly 6% dilution, and fell 9% the day it was announced.

The demand side

Copper reached an all-time $14,617 a tonne in early September on grid and data-center demand against constrained mines. Southern Copper, which mines and refines copper and molybdenum in Peru and Mexico, ran a 62% gross margin on revenue up 40.6%, and by-product credits took its first-half net cash cost to minus three cents a pound — it is the most expensive name here at 26.8x forward earnings. Teck Resources grew revenue 78.3% with gross margin widening 21 points, and is the cheapest of the three at 11.4x forward against 19.1x trailing. Freeport-McMoRan is the exception: June-quarter revenue fell 7.3% and gross margin slipped, Grasberg milled about 67% of pre-incident rates last quarter, and the shares carry 24.2x forward on a 2027 recovery.

Cleveland-Cliffs, the only US producer of grain-oriented electrical steel and the flat-rolled supplier to Detroit, turned a positive gross margin for the first time in six quarters, with $286m of adjusted EBITDA, selling prices up $76 a ton and automotive shipments the highest in two years. It trades at 1.15x book because consensus still has it loss-making this year. The transformer-assembly leg of its grid story is gone — the Weirton plant was shelved — leaving the electrical-steel franchise to carry it.

What the power channel does and does not explain

Electricity does not explain this quarter's aluminum numbers; the margins were records and Alcoa's hedges hold. What it explains is the terminal value a buyer will pay for a smelter: every megawatt-hour the data centers bid up raises the cost of the next restart and the price of the next contract renewal, and Hawesville showed the land is worth more without pot lines on it. The halving is the metal price plus, at Alcoa, a genuine change in the equity's risk from near-7% debt and fresh shares. Nothing in either company's operating record explains the rest of it. On the other side, the grid bill of materials is being paid for in cash at Teck and Southern Copper and borrowed against the future at Freeport.

Century once employed hundreds of people at Hawesville turning electricity into metal. The same wires now carry a 20-year lease.