DK Street Journal

Halliburton booked more revenue outside North America in the June quarter than in any second quarter in more than a decade, and its shares still fell about 12% over the thirty days to 6 October. Liberty Energy, which pumps only American wells, fell under 4% in the same month.

That ordering sits awkwardly with the explanation the market has used since July, when the pressure pumpers dropped out of their uptrends and the complex began trading as though North American completion work were shrinking. Primary Vision counted 196 active frac spreads on 2 October, seventeen more than a year earlier, and expects the count to hold near 200 through at least November. The Dallas Fed's third-quarter energy survey showed its index of prices received for oilfield services swing from -30.0 to 9.3, meaning services pricing turned up in the quarter. The work grew and its price firmed; what moved was the barrel these companies are paid in.

Halliburton's two halves get one multiple

North America revenue of $2.3bn was roughly flat year on year and up 7% sequentially on more US land stimulation, while international revenue of $3.4bn rose 6% to the highest second-quarter figure in more than a decade. "Our international business delivered its highest second quarter revenue in more than a decade, despite the disruption in the Middle East," chief executive Jeff Miller said on the 21 July earnings call, adding that "our North America business delivered sequential improvement, and my outlook for our business is positive." Completion and Production, the stimulation arm directly comparable to the independent pumpers, earned $474m on $3.2bn of revenue, a 15% segment operating margin.

Group operating income rose 7% on 3.7% revenue growth. The shares now change hands at 14.0x forward earnings against 17.1x trailing and 23.1x in May, on a 6.3% trailing free cash flow yield, with consensus earnings per share going from $2.34 this year to $2.88 next.

The month a Gulf premium came out of the barrel

Brent slid from above $105 in mid-September to $97.44 on 30 September as Saudi exports recovered. The United States announced an exchange of 40m barrels from the Strategic Petroleum Reserve on 29 September; the G7 agreed to release 100m barrels of crude and diesel on 2 October, and OPEC+ left November targets unchanged. By 6 October Brent held about $13.80 over WTI on record Persian Gulf tanker freight near $33 a barrel, with the inland US benchmark, the one that actually prices an American stage, down over the month.

The Gulf-exposed names took the worst of it. NESR, the Middle East pumper, fell 29.5% over the thirty days and 3.9% on 30 September alone with nothing company-specific disclosed, UBS trimming its target to $46 from $47; SLB fell 13.1%. The likelier reading is a war premium deflating rather than Middle East activity turning down.

Liberty's cash is pre-paying for turbines

Liberty's June quarter was strong on every line a pumper is judged by: revenue of $1.189bn up 14%, gross margin 17.5% against 9.7%, gross profit more than doubled, records for pump hours, horsepower hours and utilisation and $151m of adjusted EBITDA. The shares fell 13.7% on the day of that beat.

At 0.78x trailing sales against 1.36x in May, and 7.0x EV/EBITDA, the de-rating is attached to the cash statement: trailing free cash flow yield is -9.6% as roughly $1.5bn of 2026 capital spending goes largely to deposits on long-lead power equipment, with $1.1bn of supply contracts still committed at 30 June and a further order worth $801m signed with Caterpillar on 22 July. "We have line of sight to that inbound 3 gigawatts," chief executive Ron Gusek said on the 23 July call. First power at the planned 2GW West Texas data-center campus Liberty is backing through a joint venture is anticipated in the fourth quarter of 2027.

ProPetro earned its discount

The Midland pumper is the one name where the share price is tracking the business. Revenue of $305.8m was down 6.2% year on year though up 13% sequentially, and the company lost $8.1m, an operating loss in five of the last six reported quarters. Price-to-gross-profit has risen from 7.1x in May to 12.2x even as the stock halved, because gross profit fell faster than the price. Its PROPWR book reached roughly 350MW of contracted capacity from about 240MW, and $47m of the quarter's $71m of capital spending went to power equipment inside a $525-595m full-year budget set against roughly $1.16bn of trailing revenue. Twelve fleets ran in the quarter, with a thirteenth scheduled for the end of the third. "We're beginning to see positive pricing momentum," chief executive Sam Sledge told investors, saying the industry has "consolidated through attrition."

What the quarters do and do not account for

Very little of the month. Halliburton grew revenue and operating income and shed about a quarter of its multiple since May; Liberty doubled its gross profit and lost more than a third of its price-to-sales. Each name has its own reason for being sold: Halliburton and NESR for where their barrels come from, Liberty for what it has contracted to pay suppliers years before a megawatt is delivered, ProPetro for losing money with twelve fleets working. Nor has the oilfield dollar left services — Select Water Infrastructure, the produced-water handler, has held its uptrend throughout and is up 84% on twelve months.

The frac count says the work is there and the Dallas Fed says its price is firming. What the past month repriced was the crude benchmark, and that is decided a long way from Midland.

Schneider Electric agreed on 4 October to pay $205 a share in cash for PTC, the Boston software house whose Creo design and Windchill lifecycle-management products sit inside manufacturers' engineering change processes. The deal values PTC's equity at about $22.6bn and implies a $23.7bn enterprise value, a 42.3% premium to the last close before the announcement of $144.03 on 2 October. The shares gapped 33.5% to $192.26 the next session on 29.5m shares against 1.1m the day before, and closed at $193.00 on 6 October.

That is a 5.9% discount to the cash price, and it is not a view on design-software demand. It is the spread the market charges for a closing the companies anticipate by the third quarter of 2027, subject to a majority shareholder vote and regulatory clearances. The belief being tested is larger: roughly $2 trillion of enterprise-software market value has gone since early 2026 on the thesis that AI agents would eat per-seat licences. An industrial buyer has now put $23.7bn of cash behind the opposite reading of one such installed base, backed by a debt commitment letter for up to $25bn from Morgan Stanley Europe SE and Société Générale, with a $700m break fee.

What the revenue line hides

PTC's reported revenue fell 6.8% year over year in the June quarter, to $600.0m, because it had sold businesses. The meter that matters ran the other way: constant-currency annual recurring revenue excluding divestitures reached $2.448bn, up 9.1%, above the top of guidance, and full-year growth guidance went to 9-9.5%. Free cash flow, the number management guides to, came in at $249m against a $240-245m range. On a midpoint exiting ARR near $2.534bn, the $23.7bn enterprise value works out to roughly 9.4 times recurring revenue by our calculation, and about 16.7 times consensus EBITDA of $1.418bn, against 14.1 times trailing where the stock sat on its own.

Both chief executives described the asset in the same terms. "PTC provides the software the world's leading manufacturers and product companies rely on to design, build and maintain great products and unlock more value from their product data in an increasingly AI-driven world," PTC's Neil Barua said. Schneider's Olivier Blum called the combination "the industry's most complete Software & AI powerhouse". The buyer put numbers on it: EUR 250m of cost synergies by year three and about EUR 800m of revenue synergies from cross-selling and joint development.

Three names made the month

Over the thirty days to 6 October, PTC gained 36.9%, Synopsys 28.3% and Cadence 22.8%; Autodesk added 6.1%, GitLab 3.8%, Atlassian 2.0%, while Figma lost 9.1% and Adobe 10.7%. The three leaders supply essentially the whole eight-name average of +10.0%. Synopsys's gain has its own dated cause in a different end market, chip design: fiscal third-quarter revenue up roughly 42% to $2.48bn including about $711m from Ansys, raised full-year guidance, and a Morgan Stanley upgrade to Overweight on 8 September.

Autodesk is the engineering-software name re-rating on its own results rather than on a buyer. June-quarter revenue grew 16.1% to $2.046bn, free cash flow was $561m, and full-year free-cash-flow guidance went to $2.725-2.75bn. It trades at 16.8 times forward earnings against 27.3 times trailing, the compression sizing the growth the market expects. On the 27 August call Andrew Anagnost framed the moat the way the Schneider deal does: "The future of AI won't belong to the company with the best single model. It will belong to the platform that combines the richest context with the right models." Autodesk is also cutting the price of AI rather than metering it, taking its Flex token minimum from $300 to $99 while warning that gross margins will face compression from AI workloads over time.

The seat-priced pair nobody bid for

GitLab, which sells a single-application software-development platform, and Figma, the browser-based design tool, now cost the same per revenue dollar: 8.3 times trailing sales each. GitLab's growth is the slower of the two at 21.3%, with a full-year guide of 18-19% and a multiple that has expanded from 5.5 times sales in late July; its net ARR still grew 42% and dollar-based net retention accelerated to 117%. Figma grew 48.2%, with net dollar retention of 136% and 15,964 customers above $10,000 of ARR, its operating margin improving from -64.4% to -31.7% over three quarters, and it is down 61.5% over twelve months. Consensus has its growth halving to 23.8% next year, which is the whole argument.

So the sector's good month decomposes into three company-specific events and five names going nowhere. PTC's price is now a cash number with a vote and an antitrust file attached; Synopsys earned its re-rating on Ansys; Atlassian's three-month gain traces largely to one 35.8% session after its fourth-quarter shareholder letter. Nothing in the week changed what a developer seat is worth. What changed is that an industrial buyer, not a software one, was willing to pay nine times recurring revenue for product data embedded in change-control workflows, and the market has not yet decided whether any seat-licence business qualifies for the same compliment.

PTC holders are now long a date rather than a business. GitLab reports next on 7 December, and that print, not the merger, is where the seat question gets answered.

Every major containerboard mill in the United States took market-related downtime in October, and the price of the waste paper they buy as raw material collapsed with them. Fastmarkets put US old corrugated container pricing at a $38-a-ton average for the month, down $39 from September's $77 and $88 over three months, blaming "weak bulk grade demand, high mill inventories, and market-related downtime at major paper and board mills in the USA." Mixed paper reached a high side of zero. Recovered paper is being landfilled in Florida.

Falling fiber has been read all autumn as the input relief that would finally let a round of announced price increases reach the operating line. The October print inverts that reading. Cheap fiber of this kind is the fingerprint of mills that cannot sell board at September's price, not of supply discipline tightening a market.

The index only ever confirmed half the increase

The revenue side of a corrugated dollar resets off a published third-party benchmark, with weeks to months of lag before an announced increase reaches an invoice, and only if the publisher validates it. September's Fastmarkets RISI settlement recognized linerboard up $70 a ton and medium up $100 against announced increases of $140 from Packaging Corporation of America, $100 from Smurfit Westrock and $80 from International Paper, all effective 1 September. No October settlement existed as of 6 October, so the only dated index evidence is September's.

Mark Kowlzan, Packaging Corp's chief executive, described the mechanics on the company's second-quarter call: "We are seeing the majority of the first increase rolling in during July and we'll see the beginning of the second increase in August with realizations split between Q3 and Q4."

The costs that are not indexed do not wait. US average diesel hit a record $6.53 a gallon on 22 September, roughly 77% above a year earlier. Andy Silvernail, International Paper's chief executive, sized a $2-a-gallon move at $150m to $200m for his company at a Jefferies conference on 10 September, and had already told investors on 30 July: "The pricing up to now has really been eaten by inflation... If you look at what's happened with OCC, energy, diesel, freight, you name it... it's unfortunately really eaten every bit of that pricing up until today."

What Packaging Corp has to prove on 21 October

Packaging Corp, which makes containerboard and converts nearly all of it into shipping boxes and retail displays, is the group's best operator and its most exposed valuation. Its revenue growth accelerated for four consecutive quarters, from 6.0% to 14.7% year on year, and operating income fell year on year in all four. June-quarter operating margin was 13.38% against 15.37%. Total corrugated shipments per day rose 24.3% on the Greif containerboard assets bought for $1.8bn, and legacy shipments still set a quarterly record, none of which reached the operating line.

The shares are up 7.7% over twelve months at 29.7x trailing and 22.0x forward earnings, which is 26.8x the $8.58 the company actually reported for 2025 and 20.8x peak-cycle 2022 earnings of $11.02. Third-quarter results come after the close on 21 October, guided to $2.91 a share excluding special items, against $2.51 reported a year earlier.

International Paper has already been marked to its arithmetic. June-quarter revenue fell 11.3% to $6.004bn and operating income was $45m, a margin of 0.75%. Its 2025 net loss of $3.52bn makes trailing earnings multiples meaningless; at $32.14 it is on 24.6x the 2026 consensus of $1.31 and 10.6x the 2027 consensus of $3.03, a 132% one-year jump that consensus builds out of margin rather than volume, on revenue up 6.1%. The company is closing five sites for an expected $230m earnings improvement, and its North American box volumes rose 1.7% a day while industry shipments fell 1.9%.

Graphic Packaging has the cost leg and none of the index

Graphic Packaging sells folding cartons, cups and food containers to consumer-goods and restaurant buyers and makes no containerboard at all. Management told investors it is "taking pricing actions on the approximately $1 billion of revenue where pricing is not determined by a contract" — roughly 12% of about $8.6bn of sales. It re-guided 2026 inflation to at least $150m from $60m-65m. June-quarter gross margin was 13.35% against 19.19%, operating income fell 56%, and 2026 consensus earnings of $0.70 a share sit 53% below the $1.48 delivered in 2025. The shares trade at 0.79x book and 12.3x forward earnings because the estimate underneath them fell further than the price. The boxboard operating rate it runs against was 88.8% in the second quarter, while containerboard ran near 95%.

So the twelve-month tape splits along earnings lines, and only one name is unexplained. International Paper down 31.1% and Graphic Packaging down 54.9% match businesses whose margins and estimates have gone backwards. Packaging Corp's gain does not: it is priced for 2027 earnings of $13.63 a share, a 30% step-up, by a company that has not grown operating income in a year. Since 4 September, shares of the two corrugated names fell 13.7% and 3.2% while the broad market rose, which is the market beginning to ask the same question.

The box buyers are asking it out loud. The Association of Independent Corrugated Converters publicly opposed the third increase of the year on 10 August, arguing raw-material costs did not justify it and noting that producers announce large numbers and secure smaller ones. October's fiber price, set by machines that stopped running, is the strongest argument they have been handed yet.

At Banco Galicia the only credit still expanding is written in dollars and lent to the companies drilling Vaca Muerta. Dollar loans grew 19% sequentially and 48.8% over the year to June, management told investors on 26 August, while the peso loan book shrank 4% in the quarter and full-year loan growth was guided to 10-15% with the majority of it in the dollar segment. Argentina's largest New York-listed bank by market value, at $6.13bn against Banco Macro's $4.13bn, now has its growth leg underwritten by the same shale exports that YPF and Vista Energy sell directly.

That shared exposure was invisible while the two legs moved apart. September's bank sell-off was read as a domestic credit accident, and it had the evidence for it: household loan delinquency across the Argentine system reached 12.9% in July, the highest since 2004 and more than double a year earlier, against 3.6% for companies. Then on 5 October the country-risk spread fell 49 basis points to 599, back below the 600 mark at which Argentina stays locked out of international debt markets, and everything rose at once.

The carry that paid for June is nearly spent

Galicia's June-quarter net income of ARS 258.3bn was up 49.6% year on year, and almost none of it came from lending. Net interest income fell 3% sequentially; income from financial instruments rose 275% on derivative gains and securities sales. Asset yields dropped 190 basis points to 21.1%, split 34.8% on pesos and 7.4% on dollars, and the group guides net interest margin down to about 16% for the year from 17.0%, with explicit compression in the second half.

The funding relief that produced the quarter is close to exhausted. Argentine deposit rates fell from 28.09% in January to 21.10% in July, which cannot repeat against a 29% policy rate and management's own assumption of roughly 29% inflation, leaving the real policy rate near zero. The central bank has meanwhile cut the daily minimum reserve requirement from 100% to 65% as part of what it calls gradual normalization.

The consumer book is where the damage shows. Bank non-performing loans reached 8.3%, up 60 basis points in the quarter, cost of risk ran 9.3%, and the Naranja X consumer-credit platform sat at 19.7%. "we see our ROE around 10% for the year... we see that that will continue improving, I would say, around 12%, try to end the year with something around 12%," chief operating and financial officer Gonzalo Covaro said on the 26 August call, against a medium-term aim he put at 15% to 20%.

Nothing in the barrels deteriorated

Vista Energy, a Mexico City-headquartered pure-play Vaca Muerta producer with 584 employees, grew June-quarter revenue 102.3% to $1.235bn, widened gross margin to 57.4% from 46.7% and earned $333.0m. Production reached 156.1 thousand barrels of oil equivalent a day and guidance was raised to about 158, with 72% of oil sales exported at export parity, after a $712m purchase of Equinor's Bandurria Sur and Bajo del Toro interests. "The recent addition of assets has given Vista greater scale: it allows us to make a significant leap in production, exports and development capacity in Vaca Muerta," chief executive Miguel Galuccio said of the deal.

At state-controlled YPF, shale output of 170,000 barrels a day was up 35.2% year on year inside total production of 523,000 barrels of oil equivalent, and shale lifting cost $4.0 a barrel of oil equivalent against $23.6 for mature conventional fields. The 440-km Vaca Muerta South line from Allen to Punta Colorada is planned to begin export operations in the second half of 2026 at roughly 390,000 barrels a day. Brent held at $100.84 on 6 October, supported by the 4 March closure of the Strait of Hormuz that removed nearly a fifth of global crude flows — the same closure that led OPEC and the International Energy Agency to cut 2026 demand forecasts.

What 5 October settled

Over the thirty days to 6 October, Supervielle fell 14.6%, Galicia 13.3%, Macro 12.2% and Vista 10.4%, with YPF down 2.8%. From the 2 October close, when the spread finished at 646 after an intraday 655, Galicia rose 7.1% and Macro 6.9% over two sessions against YPF's 3.0% and Vista's 2.8%.

The results account for the banks and not for the oil. A guided 10-12% inflation-adjusted peso return on equity at roughly 1.1x book, with Macro at 1.07x against 1.23x in mid-September, is thin against about 11.7% in dollars on Argentine sovereign paper, and the banks themselves withdrew the private-credit expansion that justified a premium: Macro cut its real loan growth target to 2-5% from 15-20%. Nothing comparable happened in the energy leg, where Vista trades at 4.59x trailing enterprise value to EBITDA and YPF at 7.95x with a 9.5% trailing free-cash-flow yield, and where basin output set a record near 400,000 barrels a day in the third quarter. Its thirty-day decline is a discount rate, and when the discount rate moved back the shares followed within two sessions.

Which leaves Galicia holding both ends of the same bet. The peso side of its balance sheet is a shrinking book funded by depositors whose rates can no longer fall; the growing side is credit to exporters whose dollars depend on a pipeline that has not yet started shipping.

e.l.f. Beauty's gross margin reached 83% in the quarter to 30 June, about 14 percentage points better than a year earlier, and the company has already announced what it will do with the gain: give it away. Roughly 10.5 of those 14 points were refunds of tariffs the company had previously paid, about $50m of them, run back through cost of goods sold. "Our plan is to fully reinvest that money in both pricing, to have a superior value proposition, as well as increased marketing across our entire portfolio of brands," chairman and chief executive Tarang Amin told CNBC on 5 August, with price cuts planned on about a tenth of the assortment.

That matters beyond one quarter because the August prints across beauty were read as proof that a three-year de-rating had ended. e.l.f., Coty and Estée Lauder all crossed from downtrends into uptrends between late July and early September, within days of their earnings. But the three dollars are earned in three unrelated ways, and only one of the turns is a consumer buying more.

The year e.l.f. ate the duty

e.l.f., which designs cosmetics, skin and hair products in Oakland and sells them through national retailers, owns no factories and employs 849 people. It sources roughly 75% of products from China, having lifted non-China manufacturing from 1% to more than 45% over three years, and its average tariff rate rose to about 55% in fiscal 2026 from 25%. It put a $1 increase across the global assortment on 1 August 2025, then said in May it would walk some of it back as unit demand slid. The absorption shows in the annual accounts: fiscal 2026 gross margin slipped to 70.7% from 71.2% while operating margin fell to 8.0% from 12.0%, on revenue up nearly a quarter.

Then the Supreme Court ruled on 20 February that the emergency-powers statute does not authorise tariffs of indefinite scope, and the refunds began. e.l.f. raised fiscal 2027 net sales guidance to 18-20% growth from 12-14%. Marketing and digital spend ran at 22% of sales in the June quarter and the company expects the full year at the high end of 23-25% — the advertising line is going up, not down. Of the quarter's $479.4m of revenue, $160m came from Rhode, the skincare brand acquired for about $1bn and absent from the prior-year base. Amin told the call e.l.f. continues to gain shelf space, at Ulta and Walmart in particular, and is entering hair care at Target.

The shares carry 28.6x forward earnings against 103.6x trailing — the gap is the market paying for a recovery that consensus has decelerating to 8.5% revenue growth by fiscal 2028. Over six months to 5 October the stock rose 73%, Estée Lauder 36% and Coty 33%; all three slipped over the past month, and all three remain far below prior highs, e.l.f. by about half.

Estée Lauder's turn is the one with shoppers in it

Estée Lauder sells prestige brands including Clinique, M·A·C and La Mer heavily through distributors, airport concessions and duty free, so reported sales have tracked what the channel orders. That reversed. "For the first time in three years, for the month of June and the month of July we are back into positive territory for travel retail globally, led by Hainan that was in double-digit growth in the fourth quarter," chief executive Stéphane de la Faverie said on 19 August, adding of inventory: "We're shipping to the demand." Fiscal 2026 adjusted operating margin widened 320 basis points to 11.2% and the company guides 3-5% organic growth this year. The cost of getting there keeps rising: cumulative restructuring charges reached $1.748bn, against annual savings tracking to the high end of $0.8bn-$1.0bn. At 28.1x forward earnings it is priced on profit still roughly half the $6.55 a share it earned in fiscal 2022.

Coty's bounce sits on the balance sheet

Coty mostly rents its prestige brands, paying royalties to fashion houses on licensed fragrance. In July it agreed to hand Gucci Beauty back to Kering a year early for about $400m. Fiscal 2026 gross margin fell 195 basis points to 62.9%, the June quarter's to 55.1%, and operating income swung to -$81.5m. What improved was the financing: the remaining Wella stake went to KKR for $750m, cutting leverage to about 3.4x against a 2x target. Net debt of roughly $2.91bn still exceeds the $2.39bn equity value, which is why 0.80x book exists beside 68.8x trailing EV/EBITDA. Interim chief executive Markus Strobel said the priority is now sell-out rather than sell-in, calling it "new thinking for the organization". Coty's uptrend was the first of the three to weaken, in late September.

What the summer actually proved

One of the three re-ratings is consumption: Estée Lauder's channel is reordering and its margin is widening without a one-off. e.l.f.'s is a legal refund it is deliberately spending down, which means the fiscal 2027 margin to watch is the one after refunds lapse. Coty's is a cheaper balance sheet with every operating meter still falling. The category explanation fails its own control: Procter & Gamble never left its downtrend, and its fiscal 2026 gross margin fell to 50.2% from 51.2% as productivity savings were offset by mix, reinvestment and commodities. Nothing relieved the input line across beauty. Demand, meanwhile, has levelled: Circana data show US prestige and mass beauty both up 7% in the first half, closing the premium gap that once separated the two ends of the aisle.

All three report within the same week, Estée Lauder on 2 November and the other two on the 4th. For e.l.f. that print is the first chance to see what the refund bought: shelf space and share won with lower prices, or a margin handed to shoppers and nothing left on the shelf to show for it.

Corning told customers in September that it will raise prices on its yen-denominated display glass substrates by 15% or more worldwide from the fourth quarter of 2026, citing persistent cost pressure across the electronics supply chain. Late in the same month it announced a fibre agreement with AT&T worth more than $3bn. Those two events belong to businesses that share a factory culture and nothing else.

The shares are priced as a data-center supplier: 48.5x forward earnings against 72.1x trailing, a compression that is the size of the growth being underwritten. What the June quarter actually shows is a company whose profit arrives from four mechanisms with four different clocks, only one of which a hyperscaler sets.

Slightly more than half the profit, and the AI part is smaller

On the Q2 2026 results, Optical Communications sales grew 32% to $2.07bn with segment net income up 77% to $438m, a 21% net margin the company called a record. Enterprise Networks, the data-center book, was $1.27bn of that, up 65%. Glass Innovations — display and specialty glass, merged into one segment effective the first quarter — earned $354m on sales up 1%. Automotive earned $82m; Life Sciences lost $21m.

Adding the five segment results gives $846m, which puts optical at 52% of segment profit and glass at 42%. Corning does not disclose profit below the segment line, so Enterprise Networks can only be estimated: at the optical segment's own 21% margin its sales imply roughly $267m, about a third of the segment total. Carrier sales, the other optical book, grew 1% in the quarter after a 17% first half, which management attributed to customer timing.

The glass dollar comes with its own exchange rate

Display substrates are sold in yen, and Corning reports them twice. Core sales of $4.74bn exceeded GAAP revenue of $4,505m, and core earnings of $0.78 a share exceeded reported diluted earnings of $0.64 — a 22% gap the company attributes principally to hedged currency exposures and largely non-cash discrete tax items. The constant-currency adjustment is the difference between spot yen and a long-term management-determined core rate aligned to its hedges, set at 120 to the dollar for 2025-2026. A cost-driven price increase announced for the fourth quarter is the clearest evidence yet that this dollar is not simply stable.

The third mechanism answers to Washington. Solar, which houses the Hemlock polysilicon joint venture, grew sales 90% to $438m and still posted a $7m loss after an extended maintenance shutdown at the wafer plant. From 4 December 2026 every imported solar component containing polysilicon faces a 15% Section 232 tariff alongside a minimum-import-price regime; Corning is vertically integrated behind that wall, from Michigan polysilicon to panel assembly in Phoenix.

A de-rating from an extreme, with the meters still rising

The stock fell 17.8% over three months to 5 October, with its worst session following the $2bn at-the-market equity programme filed on 11 September. Price-to-gross-profit is now 22.3x trailing against 26.3x in May. So the de-rating has taken a premium down from an extreme rather than repriced a deteriorating business: revenue grew 16.6%, the Verizon agreement signed on 8 September covers more than 80 million miles of fibre from 2027 to 2032, and third-quarter guidance of $4.9-5.0bn in core sales implies about 16% growth.

"We are entering a new phase of accelerating growth," chairman and chief executive Wendell Weeks told investors on 28 July, alongside a plan targeting roughly $40bn of annualised sales by the end of 2030. Sell-side opinion splits on exactly the composition question: Bernstein initiated at Market Perform with a $140 target, arguing the price already discounts optimistic data-center interconnect growth, while BofA lifted its target to $243 on optical strength.

Both can be right about different dollars. The optical book is compounding and contracted; the glass book is repricing to cover costs; the solar book is waiting on a tariff; and the lab-equipment book is losing money. A multiple set on the first of those has to absorb the other three.

The October print will say which. Guidance of $0.85-0.89 in core earnings per share is a group number, and the group is four businesses wearing one ticker.

AST SpaceMobile, which builds phased-array satellites designed to talk straight to ordinary unmodified handsets, booked $31.5m of revenue in the June quarter against $1.2m a year earlier. It kept about a quarter of that as gross profit, because what it sold was gateway hardware and work on US government contracts: three new awards with funded near-term value above $100m, on top of more than $125m of cumulative government funding.

The jump in the revenue line has been read as the moment direct-to-device connectivity — satellite service to handsets nobody has to modify, sold through mobile carriers — began earning money. The margin says something narrower. Gross margin was 100% in the June quarter of 2025 and 25.2% this year, the signature of a line that has started carrying the cost of goods it ships. Commercial service revenue is expected to ramp in 2027.

Twelve satellites against forty-five

The company's own meter is the constellation. AST says continuous service across the United States, Europe and Japan takes roughly 45 to 60 satellites, and about 90 for continuous global coverage. As of October it has around twelve in orbit: five Block 1 units from September 2024, one deployed in February, three launched in June and three in August. Another was placed in a lower orbit than planned on 19 April and de-orbited, costing the June quarter a $125.9m loss on involuntary conversion. Initial US service with AT&T and Verizon will be non-continuous.

The demand side is contracted. Definitive agreements run with AT&T through 2030, Vodafone through 2034 and Verizon for coverage starting in 2026, inside partnerships with more than 60 operators, and the Federal Communications Commission authorized up to 248 satellites in April. Backlog is about $1.30bn, and full-year revenue guidance of $150m to $200m was held.

The spectrum bill that does not wait for a launch

One obligation runs whether or not anything flies. Under agreements signed on 22 March 2025, AST's spectrum vehicle owes at least $80m a year for 80-plus-year rights to up to 40 MHz of Ligado's L-band in the US and Canada, plus 5 MHz at 1670–1675 MHz, with roughly $550m more due at closing. Ligado filed for Chapter 11 in January 2025, the same month the deal was struck.

Capital spending was about $610m in the June quarter against $257m in the March quarter, roughly 19 dollars of capex for every dollar of revenue. Funding it has cost ownership: diluted shares rose 23.6% year over year to 299.1m, and July brought $1.15bn of 1.625% convertible notes, leaving pro forma cash above $3.7bn.

The shares sit 56.2% below the $133.09 close of 28 May and are down 27.6% over three months. The sharpest day of the year was 16 July, a 17% fall the day after a $1bn convertible offering was announced. B. Riley's Mike Crawford cut the stock to Neutral with a $65 target from $85. At 205.7x trailing sales against 387x in early May, the de-rating arrived while revenue grew.

"With the largest phased arrays ever deployed in low Earth orbit… we believe we are uniquely positioned to deliver scalable direct-to-device connectivity for both commercial and government customers around the world," chairman and chief executive Abel Avellan said on 10 August.

Globalstar's dollar is a reimbursement and its price is a contract

Globalstar, which runs a 24-satellite low-earth-orbit fleet plus SPOT trackers and commercial internet-of-things airtime, is the one listed operator already paid at scale for satellite-to-phone work. Apple was 63% of revenue in the June quarter, deferred revenue stood at $869m of largely Apple prepayments, and $550m of capex is largely reimbursable under the services agreements. Service revenue fell 5% on the timing of reimbursement service fees while commercial internet-of-things activations hit a record: the pass-through is what moved the line.

None of that sets the price. Since 13 April, Globalstar has been under a definitive merger agreement with Amazon at $90 a share in cash or 0.3210 Amazon shares capped at $90, about $11.57bn, with holders of some 58% of voting power already consenting and closing expected in 2027. Consideration can fall by up to $110m if operational milestones are missed. Closes have stayed inside a $78.57–$84.43 range for five months, and the 5 October close of $83.62 is 7.6% under the cash terms. The company has said it does not intend to hold further earnings calls or update guidance.

So a twelve-month gain of 98.5% in Globalstar rests on a contract while operating income went from $6.15m to -$4.78m, and AST's three-month decline came with revenue up and guidance intact. The split is funding, not demand. AST buys satellites, spectrum and launches years before any airtime bill arrives, and each raise reprices the equity; Globalstar's buyer has fixed what its shares are worth, and the customer behind 63% of its revenue is migrating iPhone Emergency SOS to Amazon Leo as part of the same arrangement.

The live competitor is further along than either and sells less than both plan to. SpaceX has launched more than 650 direct-to-cell satellites and T-Mobile's T-Satellite went commercial in July 2025 as a $10-a-month add-on, but voice service remains in testing through 2026, and Amazon's own direct-to-device service is not due to start until 2028.

For now the whole category's revenue comes from three places: a defense budget, an iPhone maker's prepayment and a capex invoice sent back to the customer. The first phone bill is still ahead of all of it.

Two companies rent the same thing, pallet positions in refrigerated rooms, and last month the market stopped treating them alike. Both are filling more space while handling fewer pallets: Americold's physical occupancy reached 67.4% against economic occupancy of 76.0%, and Lineage's same-store throughput fell 1.8% while the price per pallet handled rose. The improvement is coming from rate and cost removal, with physical volume still shrinking at both.

What separated the shares was the balance sheet. Americold closed a joint venture with EQT, put the proceeds against maturities and raised cash-flow guidance; Lineage carries $7.8bn of net debt, leverage near 6.0x against its own 5.0-5.5x target, and a dividend taking 73% of guided cash flow that is itself declining. When the ten-year Treasury yield passed 5%, Americold was the only industrial landlord in its set still trending up, and Lineage absorbed three analyst target cuts inside a week.

Americold spent the summer selling buildings to pay down debt. Twelve of its refrigerated facilities went into a joint venture with EQT's Active Core Infrastructure fund that closed on 31 August, freeing roughly $1.1bn of net proceeds, with about $895m of maturities through 2028 earmarked for repayment against total net debt of some $4bn. Lineage, the larger of the two operators with 498 facilities against Americold's 185, spent the same months committing capital to warehouse automation on top of $7.8bn of net debt.

Then the discount rate moved. The Federal Reserve raised rates on 16 September for the first time in three years and the ten-year Treasury yield touched 5.041%, and the sector's working assumption, that cold storage's problem is food destocking and the cure is customers restocking, turned out to be the wrong meter to watch. Neither landlord is handling more pallets. Both are earning more per pallet position. The difference in what their shares did is sitting on the liability side.

Rooms filling up, volumes still falling

Americold, which rents pallet positions and sells pallet-handling labour to food producers and distributors, reported second-quarter physical occupancy of 67.4%, up 300 basis points year on year, against economic occupancy of 76.0%. The gap is positions paid for under fixed commitments whether or not pallets are in them, and it runs to nearly nine points. Fixed commitments were 58% of storage revenue, with the top 25 customers on committed structures carrying an eight-year weighted-average stated term. Meanwhile the company handled 8.9 million throughput pallets, down 1.0%.

The two dollars inside one building behave nothing alike. Rent-and-storage margin was 61.2%, down 90 basis points; warehouse-services margin was 13.3%, up 70. That second line is a spread between the price per pallet handled and hourly wages plus the power to hold a room cold, and power is going the wrong way: US average commercial electricity prices rose 6.1% to 13.86 cents per kilowatt-hour in the first half of 2026.

"We are not relying on a recovery in demand to create value," chief financial officer Christopher Papa told investors on 6 August. "Instead, we are laser focused on executing against the priorities that are within our control."

Lineage shows the same shape with a narrower wedge: same-warehouse physical occupancy of 75.8% against 81.5% economic, same-store throughput down 1.8%, services revenue per throughput pallet up 2.1% to $32.18, and same-store net operating income down 2.9%. Its contract conversion is running backwards: the share of rent-and-storage revenue carrying minimum guarantees slipped to 44.5% from a 46.7% peak.

The glut is in old buildings, and the restocking check splits

The market absorbed negative 56 million cubic feet in the first half against 41 million of deliveries, the first negative first-half absorption since 2007, taking vacancy to a record 7.7%; pre-2006 buildings hold 68% of all vacant cube while 2006-2019 vintage runs 3.4% vacant. Replacement cost of $130-$350 a square foot, against $85-$150 for dry storage, is why nobody prices new space down to clear the old. Lineage's own estimate is that the market is about 10% overbuilt.

The independent read on inventories is mixed. USDA's September report put end-August red meat holdings up 8% year on year, pork up 12%, with poultry down 4%. On the customer side, Conagra reported Frozen & Refrigerated volumes down about 10% on its first-quarter call of 30 September, the result of its own pricing actions.

What the two balance sheets bought

Americold raised full-year cash-flow guidance to $1.26-$1.32 a share, up four cents at the midpoint after absorbing roughly five cents of dilution from the EQT venture; at $14.26 that is about 11.1x the midpoint, against an average close of $25.72 in 2024. JPMorgan upgraded the shares to Neutral on 24 September with a $16 target, citing a stabilised operating environment.

Lineage's cash flow is going the other way: adjusted funds from operations of $0.76 a share in the June quarter, down 6.2%, after $0.78 and down 8.2% in March, with leverage near 6.0x against a stated 5.0-5.5x target. The quarterly dividend declared on 15 September, $0.5325, annualises to $2.13, about 73% of the $2.925 midpoint of guidance. Competing for the same cash is the automation programme, which targets $110m of additional EBITDA from a 10% cut to roughly $1bn of annual labour spend, with $250m invested and $200m more committed through 2030. The company has idled 15 facilities, about 1% of its US portfolio, while carrying 20 under construction that it says add $134m of net operating income. Goldman Sachs cut its target to $48 from $53 on 28 September, Scotiabank to $39 from $44 on 2 October, Evercore ISI to $43 from $45 the same day.

How much the numbers explain

For Lineage, almost all of it. A 19.4% fall over three months sits against declining cash flow per share, negative same-store income, a dividend covered by a shrinking number and the heaviest debt load in the sub-sector when long yields hit their highest level since 2007. For Americold the accounting is incomplete. Revenue growth inflected from -1.6% a year ago to +1.9% in the June quarter and operating income grew 23.8%, mostly from cost removal; over 30 days the shares added 0.3% while every industrial landlord around them fell, Prologis by 6.2%. Over three months they still fell 11.5%. What looks like recovery at 11.1x guided cash flow is a de-rated price meeting a slightly better business, and consensus still models revenue down 2.6% this year.

Neither company reports again until November, so every pallet count above predates the rate move that split them. The thing worth watching is who stops competing. "We wouldn't be surprised at all, and we're certainly hearing on the street, that there'll be a couple of competitor exits in the coming quarters," Lineage chief executive Greg Lehmkuhl said on 5 August. Two-thirds of the empty cubic feet in America sit in buildings put up before 2006, owned by someone. Whether those owners leave before customers refill their rooms is the question the occupancy line cannot answer.

Monday's price letter, and the settlement that didn't follow

Nucor's price sheet moved again on Monday. The largest US electric-arc-furnace steelmaker, which melts recycled scrap into sheet, plate, bar and structural steel, lifted its spot hot-rolled coil offer by $10 to $1,230 a short ton, an eighth consecutive weekly increase that adds $30 over the month and more than $300 since October last year. The scrap side went nowhere: October's ferrous settlement came in "strong sideways", busheling near $450 a gross ton, with no one in the market calling a downward cycle. Nucor raised tubular prices on 2 October as well.

The companies grouped as data-center building materials were bought on the premise that the buildout pays whoever supplies the site. What the past twelve months actually paid for is the opposite of pricing power. Nucor is up 81.8% over that stretch and Steel Dynamics 60.7%, while Vulcan Materials is down 19.6%, Martin Marietta 23.5%, Eagle Materials 26.7% and CRH 32.4%. The two aggregates franchises whose quarries cannot be replicated are the de-rated half; the steelmaker whose realized price is reset weekly by spot indices, under an import floor set by a 50% Section 232 tariff on steel articles, is the half being paid up for.

What the spread is worth when scrap sits still

Nucor's June quarter shows the arithmetic. Its mills sold to outside customers at $1,145 a net ton, up 7% sequentially, while scrap and scrap substitutes cost $422 a gross ton; a gross ton is 2,240 pounds against a net ton's 2,000, so about $377 on the same basis, leaving a spread near $768 a ton on a record 7.1m tons shipped. Revenue rose 23.0% to $10.4bn, gross margin widened to 19.6% from 14.5%, and operating income rose 77%. That margin has now expanded four quarters running from 11.2%.

"The demand drivers across the spectrum are incredible," chief executive Leon Topalian told investors on the 28 July call. He has also described non-residential customers as "busier than anything" in his thirty-year career, with Nucor supplying roughly 95% of the steel content of a data center. The genuine data-center line is also the squeezed one: in the Steel Products segment, which fabricates joists, deck and insulated metal panels, earnings fell year on year because the rising mill price is an input cost there, and those plants ran near 65% utilization against 88% in the mills. The downstream backlog grew 10% sequentially.

At 13.3x forward earnings against 20.0x trailing, Nucor carries the largest compression on the shelf. The warning is in the estimates rather than the price: consensus has 2026 earnings per share at $18.88, up 137%, then $19.68 in 2027 and $18.68 in 2028.

Vulcan's lever still works; its operating leverage ran backwards

Vulcan's problem is not volume. It shipped 59.9m tons in the June quarter, up 1%, and public infrastructure awards in its markets were up 20%. Freight-adjusted price reached $22.97 a ton, yet cash gross profit per ton rose only to $12.02 from $11.88, because a $40m energy headwind and unit cash costs up 3% excluding diesel took the rest. Revenue growth has decelerated across four quarters, 13.9% to 3.2% to 7.4% to 2.5%, and in the June quarter operating income fell 9.7% to $425.3m on revenue up 2.5%.

"Price is our biggest lever when it comes to overcoming headwinds like this and inflationary pressures," chief executive Ronnie Pruitt said on the second-quarter call, with pricing expected to exit 2026 at the upper end of a 4-6% range. The lever is real and so is the geology: Vulcan is the largest US aggregates producer, first or second in markets worth about 90% of revenue, and freight costs make each pit a local franchise. Consensus still models 2026 EBITDA down 3.8% to $2.57bn, with the step-up arriving in 2027 and 2028. The shares hold 27.1x forward against 28.8x trailing, and the market capitalization is $31.7bn against $38.58bn and a 35.1x multiple in early May.

Martin Marietta, the number-two producer, is the harder read. Reported gross margin fell to 25.4% from 30.0%, but a $52m purchase-accounting inventory step-up on acquired assets drove most of that while adjusted cash gross profit rose 15% to $636m and organic shipments grew for a fourth straight quarter, up 2.3%. "When we go into 2027, we're going to be through all the inventory issues on purchase price accounting with Quikrete," chair and chief executive Ward Nye said on 30 July. It has since combined with Lhoist North America for $13.5bn, adding a lime book that generated $786m of adjusted EBITDA in the twelve months to December 2025 — industrial cost pass-throughs, not quarry pricing. At 27.0x forward and 16.8x trailing enterprise value to EBITDA it is the dearest name here; its 11.9x trailing price-to-earnings is unusable, inflated by a first-quarter divestiture gain. It set a 52-week low on 17 September and Wells Fargo cut its target to $585 from $609 five days later.

The calendar, not the construction site

End demand offers nothing to blame. Nonresidential construction spending ran at a $773.0bn annual rate in August, up 1.0% and a fifth consecutive monthly increase; data-center spending rose 73.2% year on year with year-to-date starts of $84.1bn, nearly three times a year earlier; single-family starts rose 7.6% to 918,000. Highway and transit authorities were extended at fiscal 2026 levels to 11 December rather than cut. Data centers are 3-5% of Vulcan's volumes; Martin Marietta reports its data-center backlog up 90% and says 70% of US data-center and manufacturing square footage sits within 55 miles of its operations.

So the results explain part of each move and the clock explains the rest. Vulcan's de-rating tracks its own income statement, where price gains are being eaten by energy and cash costs. Martin Marietta's does not: its operating numbers are better than its reported margin, and what has fallen is the price attached to a 2027 story. Nucor's advance is underwritten quarter by quarter, by a spread that widened again this week, and discounted as a peak that consensus expects to flatten. One clause finishes it: with the 10-year Treasury yield at its highest since 2002 and traders pricing roughly a 71% chance of an October rate rise, earnings dated 2028 cost more to hold than earnings banked this quarter. Steel Dynamics, whose revenue rose 33.4% and which trades at 13.3x forward, sits on the same side for the same reason; CRH at 14.2x forward and Eagle Materials at 13.7x are already priced for declining earnings.

The last thirty days hit everything — Nucor fell 3.9%, Martin Marietta 5.7%, Vulcan 6.8%, Eagle Materials 12.0% — which is what a discount-rate move looks like before the quarter sorts it out. All three report in late October.

Vulcan can put through 4-6% price again next year, and probably will. What no quarry franchise can do is move the year its earnings land.

Norwegian Cruise Line Holdings, which sells voyages across the Norwegian, Oceania and Regent Seven Seas brands, went to the bond market on 1 October to push its nearest maturity out three years and paid a coupon 262.5 basis points higher to do it. The money it holds against voyages already sold has meanwhile stopped growing: advance ticket sales including the long-term portion came to $4.0bn at 30 June against $3.9bn a year earlier, on a fleet whose capacity days keep rising.

Late September read this industry as intact. Carnival reported on 29 September with third-quarter-record customer deposits of $7.6bn, half a billion dollars above the prior-year record, and the whole shelf gapped up that session, Royal Caribbean 5.7% and Norwegian 4.2%. But a cruise line collects its money when the ticket is sold, up to two years early, and holds it free of interest until the passenger sails. That balance is the forward demand meter, and on it Norwegian is standing still while two peers are still filling up.

The float is flat while the fleet grows

The current advance-ticket line was $3,651.2m at 30 June against $3,200.6m at the end of December, a seasonal build that leaves the year-on-year total roughly where it was. Inside the quarter, the second dollar carried the first: passenger ticket revenue rose 1.2% to $1.73bn while onboard and other revenue — beverage packages, excursions, casino — rose 12.6% to $910.7m, taking onboard to 34.5% of revenue from 32.1%.

Net yield fell about 2.6% in constant currency, ahead of the company's own guidance, with the full year guided down roughly 5% and the third quarter guided down 8.9%. Net cruise cost excluding fuel per capacity day was $168.52 as reported, against guidance of about 1% growth. What compressed was the price.

Chief executive John Chidsey was blunt about the cause on the 30 July call. "The vast majority of our problems are self-inflicted execution issues, not macro," he said, calling them "totally in our control to fix" alongside $225m of identified savings. The company also said first-half 2027 net yields will be negative.

Berths and interest arrive on fixed dates

Net leverage was 5.3x at 30 June on roughly $16bn of debt against a $6.83bn equity value. NCL Corporation priced an upsized $950m of senior notes due 2031 at 8.750% to redeem the 6.125% notes due 2028, repay about $376.3m of revolver borrowings and prepay about $42.2m of export-credit-backed facilities, closing 15 October. Adjusted for that, 2027 net interest is guided to $860m-$880m: by arithmetic, about 8.6% of the revenue analysts expect that year, before a ship arrives.

Ships do arrive. The orderbook runs to 14 vessels valued at over $19bn and some 45,000 berths, including the 3,650-guest Norwegian Luna in 2027.

At 8.3x trailing enterprise value to EBITDA, Norwegian trades well below Royal Caribbean's 13.2x and Viking's 19.2x. Its forward price-to-earnings of 9.7x sits above its trailing 8.9x, which is the market pricing a smaller profit; consensus has 2026 earnings per share 26% under 2025, and trailing free cash flow is negative 16.8% of market value.

Royal Caribbean collected faster, Viking collected more per day

Royal Caribbean, with 61 ships across Royal Caribbean International, Celebrity, Azamara and Silversea, is the mirror image on the float: deposits reached $6,736m at 30 June and built $996m during the quarter against a build of $883m a year earlier. Operating income slipped 1.7% and gross margin gave up 2.4 points, yet net yields rose 1.2% in constant currency and full-year adjusted earnings guidance went up to $17.73-$17.87, roughly 14% above 2025. The shares are down 14.3% over twelve months, taking the multiple to 15.3x forward from about 20x realised 2025 earnings, while the company funds about $4.7bn of 2026 capital spending including $1.6bn outside new ships, much of it on private destinations whose yields land years later.

Viking Holdings, whose 92-vessel fleet is 81 riverboats against nine ocean liners, is the one whose price per day is rising: net yield of $645 per passenger cruise day, up 6.2% on revenue up 16.5%, with $4,711m of 2027 advance bookings, 21% ahead of 2026 at the same point. As of 9 August it had 53% of 2027 core capacity sold at $958 per passenger cruise day, up 10%, on 15% more capacity. Its shares still fell 19.2% over three months, from the most expensive multiple on the shelf, and it carries a cost shock the ocean operators cannot have: more than half its European river-cruise days from mid-July to mid-August were hit by low water, with 10-12% of affected passengers cancelling.

What the autumn selling actually described

Three of these names fell over the past three months: Norwegian 22.7%, Viking 19.2%, Royal Caribbean 5.6%. Only Norwegian's decline is matched by deteriorating results — falling yields, operating income down 14.3%, a flat float and an interest bill fixed by contract. Royal Caribbean's de-rating has no earnings decline underneath it, and Viking's followed a weather event from a starting multiple of 19.2x EBITDA. Norwegian's shares have been in a downtrend since August and touched $14.18 on 18 September, before the company said on 1 October that third-quarter results would exceed July guidance.

The ceiling over all of it is supply. The global orderbook stands at 78 ships and more than 206,600 berths, with 15 ships and about 27,860 berths due in 2027, roughly 4% more capacity. Carnival's Josh Weinstein told investors on 29 September that "for full year 2027, we are already half booked with both occupancy and pricing at record levels" — on planned capacity growth of 0.5%. Norwegian is adding berths into the same water at a pace its forward book has not yet matched.

Norwegian reports third-quarter results on 4 November. Its yield is already written down for that quarter; the company has said so. What the market has not seen is the deposit line, the money collected for voyages nobody has sailed, and the only figure that says whether the next two years have been sold.