DK Street Journal

GoDaddy spent early September telling investors that the way it gets paid is changing. At Citi's Global Technology, Media and Telecommunications conference on 8 September, chief financial officer Mark McCaffrey said the financial model is shifting toward a mix of subscription and token-based usage as its Airo artificial-intelligence bundle spreads through the customer base, and that the resulting impact on revenue timing would be addressed more fully when the company guides 2027. He reiterated free cash flow of about $1.8bn and a normalized margin on adjusted earnings before interest, taxes, depreciation and amortization of 33% for this year.

That is a longer path to the same destination, and it lands on a company the market has already stopped dating. GoDaddy — which registers domains, hosts and builds websites, and sells payments and point-of-sale systems to small businesses — has seen its forward earnings multiple fall to 13.2x from roughly 30.6x at the end of 2024, with no deterioration in earnings whatsoever. The shares fell 7.1% on the day of the Citi appearance, but the whole small-business software complex fell with it: Shopify dropped 7.5% and Wix 5.9% against a broad market off 0.3%, and reporting that session turned up no company-specific announcement from the other names. The likelier reading is a sector repricing that GoDaddy's own language amplified rather than caused.

The earnings are not the problem

Second-quarter revenue reached $1.298bn, up 6.6%, while operating income rose 31.7% to $351.2m and operating margin widened from 21.9% to 27.1%. Free cash flow rose 13% to $443m. Applications and Commerce, the attach engine, grew 11% to $515m at a segment margin near 47%. Airo is selling: "Customers are choosing Airo at a rate that is exceeding our expectations. Annualized bookings run rate has increased 5x to $50 million, versus the $10 million we shared just 1 quarter ago," chief executive Amanpal Bhutani told investors on the 30 July call.

The problem is arithmetic one layer down. The customer count was 20.5m, up 22,000 in three months, while revenue per customer rose 9% to $250 — the basket is growing, the funnel is not. And Bhutani conceded the direction of travel on the same call: "as AI reshapes how small businesses create and manage their online presence, we expect the need for traditional products like do-it-for-you services and template-based website builders to narrow and evolve over time." That sentence is the de-rating in the company's own words, and it is the same fear that has taken Wix down 54.9% in a year.

The supplier is paid by contract

One layer up sits VeriSign, which operates the .com and .net registries and never meets a customer — GoDaddy buys its wholesale domain-years and resells them at retail near $22.99 a name. VeriSign's problem is the mirror image: its earnings multiple has gone up while its reported growth has gone down, to 6.0% in the second quarter from 7.6% two quarters earlier, because the frozen wholesale price stopped contributing. Its units, however, turned: the base reached 179.1m names, up 5.1%, on what chairman and chief executive D. James Bidzos called on the 23 July call "the largest we have seen for any quarter in our history" at 12.7m new registrations, and full-year base guidance was doubled. The renewal rate slipped to 75.2% from 75.5%.

The wholesale .com fee rises 7% to $10.97 on 1 November, the first of four permitted annual steps under a registry agreement whose cap is policed through the Commerce Department's cooperative agreement. Finance chief John Calys told investors on 23 July the company models "about 50% of November's 7% price increase on .com to be recognized in 2027 revenues." The extra 71 cents does not meaningfully squeeze a registrar charging double.

The verdict

The registry trades at 29.9x forward earnings — against 25.9x at the end of 2024 — while growing revenue more slowly than the registrar at less than half the registrar's multiple. Both facts are defensible and neither is an accident. VeriSign is paid an administered fee on a base that is growing again, plus an uncapped .web launch expected to reach general availability by early 2027; its multiple is paying in advance for a price cycle that has not started billing. GoDaddy is judged on whether a bundle sold by usage can outrun the erosion of products its own chief executive expects to narrow — a question no quarter of 31% operating-income growth answers.

Management will try to answer it on 1 December, at an investor night in Tempe. Until then the registrar's revenue increasingly arrives when a customer calls an agent, while the registry's arrives the moment a name renews.

Brookfield Corporation touched a 52-week low of $37.87 in early September, closing at $38.22 on 11 September — down about 17% over twelve months. The business underneath it reported its largest fundraising quarter ever six weeks earlier. Its shorter-dated leg down is new: the shares flipped into a downtrend on 1 September, after the listed affiliates had already done so, and short interest rose 51.6% during September to 20.34m shares.

What makes this worth untangling is that the whole complex has been sold as one payout story, and only one member of it is actually a payout story in trouble. Brookfield Corporation is the parent — an alternative asset manager that runs client money across real estate, renewable power, infrastructure, private equity and credit while investing its own balance sheet alongside it. It pays a dividend of $0.07 a quarter. Whatever repriced it, a distribution yield measured against a 10-year Treasury that reached 4.818% in early September is not the mechanism.

What the parent actually earns

Second-quarter distributable earnings before realizations came to $1.4bn, up 15% year on year, or $0.61 a share; the last twelve months produced $5.7bn, $2.39 a share. Fee-related earnings rose 20%. Fee-bearing capital reached $672bn, up 19%, on a record $77bn raised in the quarter. The wealth arm's distributable earnings grew 23%, with insurance assets of $190bn following the Just Group acquisition.

None of that reaches reported earnings in a usable form. Under international accounting rules, depreciation on consolidated assets and the share of profit owned by minority partners crush the bottom line — full-year 2025 diluted earnings were $0.50 a share against $2.39 of distributable earnings over the trailing year. The trailing price-to-earnings ratio of 68x measures the accounting, not the business. At $38.22 the shares are near 16 times last-twelve-month distributable earnings and 1.94 times book value.

Management publishes its own estimate of what the company is worth — a "plan value" of $66.37 a share at 31 March 2026, trimmed from $68 at year-end — and has repurchased stock this year at an average $41, which it described as roughly a 40% discount to that figure. The estimate is the company's own, produced by an interested party. The buyback is the one place it is being tested with cash.

The one vehicle with a real problem

Brookfield Renewable, which owns hydro, wind, solar and storage plants across the Americas and sells power under long contracts, reported a record quarter that was partly a sales event. "We delivered another record quarter, generating FFO of $421 million, or $0.62 per unit, up 13% or 11% per unit year-over-year," chief executive Connor Teskey told investors on 31 July. About $175m of that sat in the hydro segment as other income from developed-asset and non-core sales. Strip it and per-unit funds from operations fall near $0.36 against a declared distribution of $0.392 — coverage of roughly 0.92 times. Accounting revenue fell 1.8% in the quarter. The units yield 5.17%, barely clear of the risk-free curve, and trade at 9.69 times trailing enterprise value to earnings before interest, taxes, depreciation and amortization.

Brookfield Infrastructure is the opposite case and gets measured differently. Its regulated utilities, rail and motorways, gas pipelines and roughly 148,000 telecom towers produced second-quarter funds from operations of $0.89 a unit, up 10%, with the distribution raised 6% to $0.455 — a 66% payout, inside the 60-70% target. Crucially, its definition of funds from operations excludes gains on disposals, so that payout is struck on operating cash alone. Revenue growth accelerated to 19.4%, and at 7.05 times enterprise value to EBITDA it is the cheapest of the listed vehicles. Its units were still in an uptrend as of 10 September.

Brookfield Asset Management, the fee collector, earned $808m of fee-related earnings in the quarter, up 20%, and has bought back $575m of its own shares this year. It has held an uptrend since mid-August despite falling 9.5% in a fortnight. It also showed what has changed: it issued $1bn of notes in the second quarter at 4.832% for five years and 5.298% for ten.

The verdict

Brookfield Renewable's de-rating is earned — a distribution that needs asset sales to clear one times coverage deserves a lower multiple when money costs more. Nothing in the reported results of the parent, the manager or the infrastructure partnership does the same work. Their cost of capital rose, their duration lengthened, and holders of long-dated cash flows repriced them together; the likelier reading of the parent's September leg is rate duration plus positioning rather than franchise damage. The Westinghouse stake the company holds alongside Cameco is the clearest example of what the market is declining to pay for: a reactor order cycle whose units would not run before the 2030s.

On 14 October, holders vote on folding the paired corporations into their partnerships one-for-one, ending the last structural quirk in the stack. After that there is nothing left to simplify, and the discount to management's own number will have to be argued on the earnings — or closed by the buyback that is currently the only visible buyer at these prices.

Lamb Weston sold materially more frozen potatoes in its fiscal fourth quarter and made less money doing it. Net sales rose 6%, led by a 7% increase in sales volume against a 3% decline in price/mix, and gross margin came in at 20.4% — the same figure as the prior-year quarter. Operating income fell 9.1%.

That is the whole question in front of the company, which manufactures fries, commercial ingredients and appetizers in Eagle, Idaho and sells them mostly business-to-business, to restaurant chains and foodservice distributors, under multi-year contracts. Its earnings are a spread between a potato cost contracted once a year with growers and a price it cannot reset mid-term, with utilization of its processing lines as the swing factor. Fiscal 2027 is supposed to be the year the spread widens. Management has already told investors it will not.

The tailwind is spent before it arrives

Lamb Weston's own fiscal 2027 outlook says the benefits of lower raw potato costs, supply-chain savings, better fixed-cost absorption and the lapping of fiscal 2026 potato write-offs are "anticipated to be mostly offset by inflation in essentially all other input cost areas." Adjusted EBITDA is guided to $1.1–1.2bn against $1,147m delivered, on sales guided flat to up 1%.

The crop cuts both ways. North American contracted acreage is expected to fall at least 10% for 2026, and the fry industry has signaled intent to lower contract prices by 3 to 5 cents per kilogram — a cheaper ton bought, and a cheaper contract sold. Underneath that, the end market is soft: US frozen potato exports fell 12.8% in the first quarter, and quick-service restaurant foot traffic was down 1.6% year on year in May on a higher average check. The National Restaurant Association counted 45% of operators reporting lower customer traffic that month, the fifteenth net decline in sixteen months.

So the company is managing absorption directly: it is closing its Munro plant in Argentina and curtailing a Netherlands line, affecting about 110 employees at $80m–$110m of pre-tax charges, alongside a roughly 4% workforce cut. Its Focus to Win program targets at least $250m of annualized run-rate savings by the end of fiscal 2028 and cleared its $100m milestone this year. The pressure is external: the board was expanded to 13 seats under an agreement with Jana Partners and Continental Grain, and Starboard Value has pressed for 25% adjusted EBITDA margins by fiscal 2029. "This past year marked an important inflection point for our Company," chief executive Mike Smith said on July 24. North America delivered segment EBITDA up 17% at a 26% margin; International EBITDA fell 55% on European price/mix and a $33.1m potato write-off.

Conagra's problem is metal and meat

Conagra — Birds Eye, Healthy Choice, Marie Callender's, Slim Jim, sold at retail shelf price through a handful of grocers — shares the frozen-foods label and none of the economics. Its cost shock is tariffs on the tinplate steel and aluminum in its cans, which the company put at 3% of cost of goods sold, more than $200m a year, plus beef inflation. Fiscal 2026 revenue fell 2.9% to $11.28bn and gross margin slid to 23.9% from 27.7% two years ago; a roughly $2bn goodwill and brand impairment produced a net loss. Net debt ended at $7.1bn, or 3.83x, against a 3.0x target, and the quarterly dividend was halved to $0.175. "I really believe a balanced approach to capital allocation is critical to the long-term success of the company," chief executive John Brase said on the July 15 call. Fiscal 2027 organic sales are guided down 1–3% with volumes down mid-single digits, and an incremental $125m goes into supply-chain resilience.

What the last two weeks actually were

Both shares fell hard in early September, and neither announced anything. The trigger was Campbell's September 3 results: a sales miss, adjusted gross margin down 190 basis points to 28.6%, fiscal 2027 earnings guided 17–24% lower and a 36% dividend cut. From September 2 to September 11, Campbell's fell 10.9%, Conagra 10.8% and Lamb Weston 9.1% — and so did General Mills, down 11.7%, J.M. Smucker, Kraft Heinz and Post, against roughly 1% for the broad market. Structural pressure sits behind it: GLP-1 medications are now used by about 18% of US adults, up from roughly 14% a year earlier, with users spending less per grocery trip.

The September leg neither protagonist earned. The twelve-month de-rating they did: Lamb Weston's diluted earnings of $2.08 are less than half the $4.98 of fiscal 2024, and it trades at 15.8x forward earnings against 22.8x trailing and 10.7x trailing enterprise value to EBITDA — inexpensive against profit guided to stand still. Conagra's trailing multiple is meaningless after the impairment; its 10.1x forward earnings applies to a consensus that has earnings falling another 14.6% before any recovery in fiscal 2028.

The useful meter at Lamb Weston is no longer revenue, which has now grown four quarters running. It is gross margin and factory absorption, because the company has told everyone what the cheaper crop is worth net of everything else. Volume that arrives at a 3% lower price into flat profit guidance is share bought with margin.

Lamb Weston reports on October 1, into a crop whose savings it has already promised away.

Maximus runs the eligibility files and phone lines behind American public health programs — Medicaid and Affordable Care Act marketplace enrollment, multilingual contact centers for government health and human-services agencies, independent medical reviews and appeals — and it has now reported three consecutive quarters of shrinking revenue. Analysts still expect its earnings per share to rise this year.

The shares have lost more than a third of their value over twelve months, and profits did almost none of that work. A year ago the company was worth about 16.4 times the earnings it went on to report for fiscal 2025; it now trades near 8.3 times trailing earnings, with free cash flow equal to roughly 14.5% of its market value. Underneath the halved multiple is a federal civilian services market that stopped buying on schedule — and a company whose largest earner is rebid every year.

The hole in this year's earnings

The Department of Veterans Affairs paused performance incentives and disincentives on its Medical Disability Exam program effective 1 July 2026, expected to run through the end of December. Those incentives had contributed about $0.35 of earnings per share in each of the first three quarters of fiscal 2026. Maximus cut adjusted earnings guidance to $7.90–8.20 from $8.25–8.55 and left revenue guidance of $5.2–5.35bn untouched — an earnings event rather than a demand one.

"These incentives have become a bigger contribution for us in our fiscal year '26 than they have been in prior years, which is really a testament to the investments we've made into the program over the past several years that have brought us to the high level of performance across the incentive metrics," chief financial officer David Mutryn told investors on August 6.

That program is also the recompete. Maximus holds one of four positions on the follow-on veterans exam contract vehicle, with a $13.2bn ceiling, alongside Leidos subsidiary QTC Medical Services, Optumserve Health Services and Loyal Source Government Services. The regional awards carry a one-year base with a single option year, so the risk recurs annually rather than on a comfortable multi-year cycle. Chief executive Bruce Caswell said preliminary analysis of the draft performance work statement indicates it covers all six regions the company serves today.

A budget-line split, not a company one

Maximus was not singled out. Over the same twelve months Booz Allen Hamilton fell 27.8%, Leidos 31.5% and ICF International 13.6%, while defense- and intelligence-weighted CACI rose 24.3% and SAIC 20.1%. The line runs through the appropriation, not through execution quality, and Maximus is trying to cross it: it has identified a $47bn addressable market in defense and national security, with wins at the Air Force and the Transportation Security Administration.

The bear's evidence is conversion. The pipeline is $50.4bn, more than half of it federal, but year-to-date signed awards were $1.25bn — a book-to-bill near half — with another $1.35bn awarded and unsigned as long recompetes cleared protest. Management described civilian-agency opportunities delayed, rescoped or canceled, with more bridge contracts and short extensions.

What comes next is dated. The 2025 reconciliation law requires 44 states and the District of Columbia to condition Medicaid eligibility for the expansion population on work requirements from 1 January 2027, with 34 of them verifying compliance every six months at renewal. That is 21 million beneficiaries to be redetermined semiannually — the caseload Maximus is paid per case to process. A federal court rejected the challenge to the requirements on 29 July, and the company expects its U.S. Services segment back to mid-single-digit organic growth in the September quarter.

The label's other tenants

Two companies share Maximus's industry heading and share nothing else with it. Target Hospitality owns 27 temporary-accommodation communities holding roughly 15,500 beds, and has more than doubled over the year on work unrelated to any appropriation: since January it has signed over 9,000 contracted beds representing more than $1.4bn of multiyear contracts for AI data-center construction crews, including a 48-month agreement expected to exceed $750m. Its government segment — the part the shared label implies is the engine — shrank to $13m in the June quarter, with $5–7m of transitional costs booked to redeploy those assets. The business is still loss-making, at an operating margin of -8.8% in the quarter and 60.4 times trailing enterprise value to EBITDA, which is a price for 2027, when consensus has revenue at $731m.

First Advantage, which screens job applicants and bills per screen, is the read on hiring, and it is soft. Base revenue grew 6.7% in the June quarter; management attributed about half of that to episodic enterprise-wide rescreening and labor restructuring, implying underlying volume growth near 3%, and guided the fourth quarter to a neutral base contribution. It raised full-year revenue guidance to $1.67–1.71bn on pricing, attach rates and Sterling synergies, and at 16.5 times forward earnings against 26.9% expected earnings growth this year, the price is not the stretch here. US payrolls rose 162,000 in August with unemployment at 4.1%; nothing in the screening book contradicts that.

The verdict

Four percent less revenue and a $0.35 incentive hole earn a lower price for Maximus. They do not obviously earn a multiple half what it was, on a company consensus still has growing earnings every year through fiscal 2028. The market is pricing permanence — that civilian agencies keep buying less, that recompetes keep landing as bridges, and that the veterans exam work is lost rather than repriced. The January redeterminations are the first thing on the calendar capable of testing that, and they are a policy, not a forecast. Federal funding itself runs to 11 December 2026 under the latest extension.

The rebid is decided a year at a time; the caseload arrives on a date already written into law. The current price assumes neither goes Maximus's way.

Entergy reported higher net income for the June quarter and lower earnings per share, and the gap between those two facts is the equity it is issuing to build power plants for Meta. Net income rose 3.4% to $487.8m. Diluted shares rose 4.6%, to 466.3m, and adjusted earnings per share came in at $1.03 against $1.05 a year earlier.

That is the whole mechanism of a regulated utility compressed into one quarter. Entergy — the New Orleans company that serves about 3 million electricity customers across Arkansas, Louisiana, Mississippi and Texas — is not paid for megawatt-hours in any way its revenue line makes legible. Fuel and purchased power are largely passed through at no margin; earnings come from an allowed return on rate base, the depreciated capital a state commission has approved it to own. Data-center load reaches net income only after it becomes approved capital spending — and capital spending on this scale has to be funded. Entergy's five-year plan runs to $67bn, with $13.2bn of investment in 2026 rising to $16.8bn in 2027, supported by a $7bn equity program. About $4.1bn of that equity was already settled or contracted under forward sale agreements as of 30 June, against a consolidated debt-to-capital ratio of 65.2%.

Rate base grows; the denominator grows with it

The dilution is not new and not an accident. Full-year 2025 diluted shares averaged 450.2m against 431.6m in 2024. Industrial sales grew 10% year on year in the June quarter as new projects ramped, revenue growth nonetheless decelerated from 12.0% in the March quarter to 5.9%, and operating income was slightly lower than a year ago at $834.7m. Entergy affirmed adjusted earnings guidance of $4.25 to $4.45 for 2026 and told investors the majority of the year-on-year increase lands in the fourth quarter. "We remain firmly on track to meet our 2026 adjusted EPS guidance and longer-term outlooks," chair and chief executive Drew Marsh said on the July 29 call.

What Louisiana decides in December

The demand side is the least contested part. Marsh put the pipeline at "seven to 12 gigawatts of hyperscale data center potential," plus three to five gigawatts of traditional industrial interest — a range, not a contract book. What is contracted is Meta's Hyperion campus in Richland Parish, under an electric service agreement currently running 15 years with a proposal to extend to 20, which Entergy says is structured so that Meta pays full cost of service and delivers roughly $2bn of customer savings over twenty years on top of $650m announced earlier.

Who carries the risk if the load does not show up is what the Louisiana Public Service Commission votes on in December. It agreed 4-1 in April to fast-track seven more gas plants for the Meta campus — the first application under the state's new "Lightning Amendment" — on top of three units totalling 2.26 gigawatts approved in August 2025. Critics, including the Union of Concerned Scientists, argue the framework leaves ratepayers exposed to large portions of data-center-triggered system costs. Entergy's answer is a voluntary standard: "The Fair Share Plus pledge really kind of originated with our first customer in Mississippi, with AWS," Marsh said on the same call. Virginia codified the equivalent instead — its new large-load class, effective 1 January 2027, requires customers above 25 megawatts to pay at least 85% of contracted transmission and distribution demand and 60% of generation demand. Dominion, the incumbent there, no longer offers an independent read on any of it: since NextEra agreed in May to buy it all-stock at 0.8138 NextEra shares, it trades within about 3% of deal terms.

What the de-rating is doing

Entergy's shares are flat over thirty days and down 3.8% over three months, up 17.7% over twelve; the 50-day average crossed below the 200-day on 21 August. The multiple has come in from 29.9x trailing earnings in May to 26.7x trailing and 24.0x forward — still the dearest of the large regulated names, against Dominion's 18.2x forward. With the 30-year Treasury at 5.25% against a 12-month average of 4.89% and authorized returns near 9.8%, the spread that makes a utility a bond substitute has narrowed for everyone.

The group-wide bearish reading is mostly one name and one state. Across the eight large load-growth utilities the equal-weight thirty-day decline is 4.5%, but PG&E fell 19.0% in the 31 August session alone — its worst since March 2020 — after California lawmakers amended Senate Bill 492 without the wildfire liability protections investors expected. Excluding it, the group is down 2.5% and one member is higher.

So the de-rating is doing rational work rather than opening a gap. Entergy's business is delivering what it promised on the load side and converting it into rate base through a formula rate plan that admits each new generator into rates on entering service. What nothing in the quarter yet establishes is that the conversion reaches a shareholder: net income is growing at low single digits while the share count grows faster, and consensus asks for 16% per-share growth in 2027 to make the current price work. One number that would settle it — Entergy Louisiana's currently authorized return on equity — is not stated in the commission's published formula-rate schedules.

The Cottonwood plant Entergy acquired is available now; the customer arrival window management describes is years out. December decides seven more plants on the same bet.

Hallador Energy owns a one-gigawatt grid connection in western Indiana, the coal plant behind it and the underground mines that feed it. In the June quarter it sold the right to call on that plant for most of the next fourteen years.

The company's forward book of contracted energy and capacity revenue stood at $1.6bn on June 30 against $571.2m three months earlier. Add third-party coal commitments of $236.5m and the coal Hallador sells to its own plant, and the segment-level figure is $2.4bn. That is the whole investment case in one line: a company with a market value of roughly $734m has booked years of revenue it has not yet recognized, on paper signed with counterparties who want a generator kept alive.

"Six years ago, we were an underground coal mining company. We acquired a 1-gigawatt interconnection, then the plant that utilizes it," chairman and chief executive Brent Bilsland told investors on the August 10 call. The agreements, he said, place Hallador "in a substantially sold forward position on accredited capacity for approximately the next 14 consecutive years with commitments extending through 2040."

What the contracts leave out

They do not sell electricity. The 12-year agreement signed May 1, covering planning years 2028 through 2040 and worth with an earlier three-year deal roughly $1.1bn, is capacity-only and excludes energy sales. The counterparty is a subsidiary of a utility; no data-center offtake has been disclosed, and the deal still needs regulatory approvals expected in the second half of 2026. The data centers are adjacent rather than contractual — one project has broken ground on a 1,000-acre site next to Merom, another developer has bought land on the far side of the plant.

Leaving the energy position open is deliberate, and in the June quarter it hurt. A 60-day planned outage on Merom Unit 1 and unplanned Unit 2 downtime landed in high-price hours, forcing Hallador to buy replacement power; adjusted EBITDA came in at -$2.9m, and the realized energy price fell to $41.69 per megawatt-hour from $52.66. Third-party coal sales actually rose 6.6% to $40.6m, still about 40% of revenue. Nothing here was caused by the coal price.

The price of the thing Hallador has sold is administratively pinned. PJM's auction for 2028/2029, announced July 14, cleared at the federally approved cap of $325 per megawatt-day for a third consecutive year, slightly below each of the two prior caps. The grid operator bought 138,318 megawatts, 5.6% short of its own reliability requirement, and reckons the auction would have cleared at $555 without the cap. The scarcity is genuine; most of its value accrues to buyers.

The other way to be paid

Alliance Resource Partners, the Tulsa partnership mining the Illinois Basin and Northern Appalachia with an oil and gas mineral royalty book attached, sells tons years forward. In the quarter it added 21.2m committed-and-priced tons and now has 29.4m priced for 2027 delivery. "Our expected 2026 coal sales tons are essentially fully committed at the midpoint of guidance," chief executive Joseph W. Craft III said on July 27.

Spot moves reach that book only as contracts roll, and when Appalachian realized price rolled down to $63.57 a ton on legacy Tunnel Ridge tonnage expiring, cost per ton fell 29.7% to $46.22 — the margin per ton widened. Net income rose 33.9% to $79.6m, royalty segment EBITDA set a record at $38.0m, about a fifth of the total, and distribution coverage reached 1.39x on the maintained $0.60 quarterly payout. The units trade at 11.0x forward earnings against 12.9x trailing, 6.0x trailing EV/EBITDA and a 10.2% free cash flow yield, on earnings consensus has near a cycle low before a rebound of roughly 22% in 2027.

The twelve-month gain in this corner of energy belongs largely to Peabody Energy, up 55%, and it is earned on something neither of the others touches: seaborne metallurgical coal priced off indexes, with coking coal futures lately near $275 a ton. Peabody lost $90.6m in the June quarter on revenue of $1.003bn and is left ramping its Centurion mine after terminating the $2.32bn purchase of Anglo American's Australian coking coal assets.

The verdict

One heading, three unrelated payment mechanisms. Alliance's advance is earned in the accounts — cost per ton, not price per ton, and a royalty stream with no mine attached. Hallador's de-rating is also earned: two straight quarters of negative operating income, consensus at a 2026 loss of $0.25 a share, and 12.4x trailing EV/EBITDA on cash earnings that have collapsed. What is not in those numbers is the backlog, the sub-$800m Turtle Creek gas peaker at roughly $1,700 per kilowatt, and $84.2m of liquidity that has to fund it.

Hallador has sold the next fourteen years of standing ready, at a price a federal cap holds down. What it kept for itself is the power the plant actually generates — the part that requires the plant to run.

nCino, which sells loan-origination, account-opening and compliance software to banks and credit unions, has spent a year rewriting how it charges. Instead of counting seats and modules, renewals now take a one-time price step of about 10% and then scale with the client's asset growth. Just under half the company's annual contract value has crossed over — 48%, up from 40% three months earlier, with 12 of its 20 largest American enterprise customers converted.

Procore, the construction-management platform that competes with Autodesk Construction Cloud and Oracle Aconex, never sold seats at all: it charges an upfront annual fee per product based on the customer's Annual Construction Volume — the aggregate dollar value of work across that customer's projects — with unlimited users, support and storage included. Both companies have therefore handed their growth rate to someone else's activity: nCino's to bank balance sheets and mortgage volumes, Procore's to the dollar value of buildings going up. The meters that matter are construction planning and weekly mortgage applications, and they currently point in opposite directions.

The bank index is working; the mortgage index is not

nCino's July-quarter revenue was $161.0m, up 8.2%, with subscription revenue of $143.5m up 10%. Strip out mortgage and subscription growth accelerated to 12%; United States mortgage revenue of $20.6m fell 1%, and management guided it down again, to $20m this quarter and $18.5m next, blaming higher-for-longer rates and churn among independent mortgage banks. That guidance has since hardened into fact: the Mortgage Bankers Association reported applications down 2.7% in the week ending 4 September, with its refinance index 25% below a year earlier and the 30-year fixed rate at 6.85%, the highest since June 2025.

The rest of the business is where the repricing lands. Gross margin reached 62.6% from 59.2% a year earlier, operating margin swung to +8.5% from −6.2%, and free cash flow of $34.0m rose 170%. "We are no longer trying to convince prospective customers that we can lead the transition to agentic-AI-powered banking. We're doing it," chief executive Sean Desmond told investors on the August 25 call. Full-year revenue guidance of $644m–$647m still implies single-digit growth, and consensus carries 8.5% into the following year.

The construction index is contracting

Procore's June-quarter revenue was $375.2m, up 15.8%, the fourth straight quarter in a narrow band from 14.5%. Gross margin held near 80%, operating margin turned positive at 1.2% against −9.3% a year earlier, and net income of $16.9m replaced a $21.1m loss. "Our outstanding Q2 results demonstrate the continued value our platform provides to the construction industry," chief executive Ajei Gopal said on 29 July, alongside full-year guidance of $1,510m–$1,514m — about 14% growth, slower than the quarter just reported. The order book grows faster than revenue, but management told analysts the two rates will converge over three to four quarters as average contract duration stabilises, so part of that gap is contract length rather than demand.

Underneath sits the index itself. Private nonresidential construction spending fell 4.7% year over year in June and 7.9% excluding data centers, with manufacturing down 21.4%. The forward planning gauge is no better: the Dodge Momentum Index slipped 0.4% in August to 282.0, commercial planning down 3.0%, and it leads nonresidential spending by twelve to eighteen months.

What the shares did

Procore rose 67% from its 25 June low to a 27 August peak of $64.05 and nCino 65% into early September; both have since given back about 12% in six sessions, with no company news at either. They fell in step with BlackLine, Freshworks and Appian — close-automation, seat-priced support software and low-code, none of them indexed to a customer's volume — after a poor 30-year Treasury auction pushed yields up across the curve on 10 September. Rotation out of long-duration software is the likelier reading, not five company stories. Over twelve months Procore is down 25% and nCino 31%.

The verdict

Procore earned its summer on margins, not on its index: with earnings still thin, price-to-gross-profit of 7.05x trailing and 6.62x forward sits below the 7.96x of early May, so the advance restored a multiple rather than created one. Its uplift lever requires customers to build more in a market whose non-data-center spending is shrinking. nCino's lever is self-help — a renewal step that arrives whether or not the customer grows — which is why its two valuation lenses disagree so violently: trailing price-to-earnings has collapsed from 348x in May to 67x as profits inflected, while price-to-gross-profit has risen to 5.97x from 5.52x over the same stretch. Earnings say it is re-rating downward; gross profit says it has re-rated up about a tenth while growing 8%. Both are true, and which one governs depends on whether the repricing outruns the mortgage drag.

More than half of nCino's contract value has yet to convert. After that, the rest of the story belongs to the banks and the builders.

At the Jefferies Global Industrials Conference on September 10, International Paper's chief executive put a number on the week's oil move: diesel, Andy Silvernail said, has risen about $2.00 a gallon in six months. For a company that makes linerboard and medium and converts them into corrugated boxes, that is a headwind of $150m to $200m. He was speaking on the day his shares bottomed, after a 6.0% single-session fall on September 9 — the session Brent crude jumped to $100.72 on Middle East fighting and US diesel set a record at $5.94 a gallon, on its way past $6 by September 11.

That is the whole of the story, and it is a timing story. A containerboard maker's revenue is set by contracts indexed to a published third-party benchmark, with one to three months of lag before an announced increase reaches an invoice — and only if the publisher validates it. Diesel has no lag. So the industry's strongest pricing year in memory and its worst input shock in years are settling on two different clocks, and the shares are marking the fast one.

The slow leg

The price side is real and documented. Roughly 3.9m tons — about 10% of US containerboard capacity — was permanently retired between February 2025 and March 2026, International Paper's Red River, Riceboro and Savannah mills a large share of it. The industry operating rate reached nearly 95% in the second quarter against 93% in the first. Producers opened a third increase in July, the largest on record: PCA at $140 a ton effective September 1, International Paper at $80 and Smurfit Westrock at $100.

Collecting it is another matter. Fastmarkets RISI has recognized a net increase of only $100 a ton across the year's first two rounds, and North American containerboard was flat in August for the second consecutive month. Buyers are contesting the third: the Association of Independent Corrugated Converters opposed it on August 10, arguing three hikes in five months are not justified by raw-material costs. Anthony Smurfit, chief executive of Smurfit Westrock, the third large North American producer, described the lag plainly on the July 29 call: "the full effect of that is going to be felt in quarter 3 and quarter 4 and any other pricing initiatives will be felt either very late quarter 4 or into quarter 1 of next year."

The fast leg, and what it has already eaten

"The pricing up to now has really been eaten by inflation," Silvernail told investors on July 30. "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." Smurfit Westrock's finance chief Ken Bowles said the same thing on July 29 — "freight costs globally represented a significant headwind, driven largely by higher fuel costs and shipping rates" — as the company's freight estimate climbed to $300m from $50m in April.

International Paper's June quarter is what that arithmetic produces: revenue down 11.3% to $6.00bn and operating income of $45m, a margin of 0.7% against 3.0% a year earlier. Full-year North American adjusted earnings before interest, tax, depreciation and amortization guidance came down to $2.35–2.45bn, the macro headwind estimate was tripled to about $150m, and the second-half volume assumption was cut to flat from up 1%. Demand is not the culprit: box volumes still grew 1.7% a day year over year.

Packaging Corporation of America, the Illinois producer that converts most of its own tons into boxes, sits on the other side. It told the same Jefferies conference that containerboard markets remain "exceptionally tight," that it is running full and cannot produce additional tons, and guided the third quarter to higher price and mix as it implements the first increase and begins the second. It fell 10.8% between August 11 and September 10; International Paper fell 17.0% and Smurfit Westrock 13.0%, against 1.7% for the S&P 500 tracker. The September 9 session itself was broad — the index lost 0.48% — so the packaging complex was singled out.

What the businesses earn

International Paper's de-rating is earned on the earnings currently visible. Trailing multiples are unusable after 2025's $3.52bn net loss; on consensus the shares trade at 25.0x this year's $1.37 and 11.4x next year's $3.00. The fall is a vote against the 2027 line, which requires profit to more than double while the company spins off its European packaging business, still targeted for early 2027.

PCA's is harder to justify from the business. It is de-rating from a full price — 30.5x trailing earnings against roughly 25x at the end of 2024 and 24x at the end of 2025, on lower earnings, and 22.5x forward, the most expensive of the three. Smurfit Westrock is the cheapest, at 8.7x trailing enterprise value to EBITDA and 1.4x book, having cut North American corrugated volumes 4.5% deliberately and shrunk its loss-making plant count from about 80 to about 20. None of the three reported a demand collapse. What they reported was a cost line that moved before the price line could.

The industry's decade-long argument is that price increases pay for input inflation. This year has tested the sequencing rather than the logic, and the sequencing failed: the third round is the largest ever announced, and the only meter that decides whether it happened has now printed no change for two straight months.

BWX Technologies sells most of what it makes to one customer under contracts that reimburse its costs, and in August it told investors that segment's revenue would grow more slowly this year — because it had found ways to spend less. On the August 3 second-quarter call, management cut 2026 Government Operations revenue-growth guidance to high single digits from low teens on stronger cost performance, and in the same breath raised that segment's adjusted earnings-before-interest-taxes-depreciation-and-amortization margin guide to roughly 20.5% from above 19%.

That is the mechanism a nuclear-labeled stock chart cannot show. BWXT, a Lynchburg, Virginia manufacturer of naval reactors, nuclear fuel and missile launch tubes for the US Navy's propulsion program, books cost-type work: lower reimbursed cost means a smaller revenue line and a better fee margin. Investors reading the government revenue line as a demand meter are reading an accounting artifact — and it is the line that matters most, because Government Operations was $601.3m of the $901.6m BWXT reported for the June quarter, growing 2% year over year at a 17.6% segment operating margin.

The growth is in the lower-margin half

Everything that grew fast sits in Commercial Operations — Canadian reactor components, fuel handling, and, since the Precision Components Group deal closed in early July, US commercial manufacturing including AP1000 content. That segment produced $302.5m of revenue, up 72% (33% of it organic), and $24.3m of operating income: an 8.0% margin, less than half the government business earns.

So the company grew revenue 18% and reported operating income of $89.9m, down 12.2%. Gross margin fell to 22.44% from 25.05% a year earlier. That pattern is not new — reported operating income has declined year over year in each of the last four quarters, and full-year 2025 operating income fell 15.1% on an 18.3% revenue gain. Acquired, capacity-hungry commercial work is diluting a franchise margin. Management also trimmed the Commercial Operations margin guide to about 13% from about 14%, citing capacity expansion and hiring, with recovery expected in 2027.

Demand is funded; the build rate is the constraint

Backlog ended the quarter at $8.4bn, up 40%, on trailing book-to-bill of 1.7x — though that is a step down from the $8.7bn peak the prior quarter, when a $1.4bn Naval Reactors pricing tranche landed. On July 29 the Navy awarded $76.6bn to General Dynamics and HII for nine Virginia-class and five Columbia-class submarines. The money is there; the boats are late. The first Columbia boat is running roughly 12 to 17 months behind, and Virginia-class construction runs near one boat a year against a two-a-year target the Navy's own chief has told Congress will not arrive until about 2032.

All four 2026 guidance lines went up on August 3: revenue to about $3.8bn, adjusted EBITDA to $662–672m, adjusted earnings per share to $4.70–4.80, and free cash flow to $345–360m from an initial $315–330m — the opposite of a capacity program strangling cash. BWXT is also selling its medical business to Nordic Capital for up to $800m, roughly $130m of revenue, keeping a 20% stake. On August 26 the Army paired its BANR microreactor with Fort Campbell under the Janus program. The commercial order the bull case wants is still pending: "Although award timing can be difficult to predict our customer discussions are advancing," chief executive Rex Geveden said on the August 3 call, guiding to at least one new-build equipment order by year-end without committing.

What the multiple did

Shares closed at $152.49 on September 10, 36% below the April 15 high of $238.42, having fallen hardest between August 17 and 25 on unusually heavy volume. Forward earnings guidance rose across that span; the forward price-to-earnings multiple fell to 32.2x on consensus 2026 earnings of $4.74, against roughly 46x in early May. The de-rating is almost entirely multiple compression, and it is sector-wide — on September 10 Piper Sandler split its advanced-nuclear coverage, rating Oklo a buy and X-Energy a sell, and Oklo, NuScale and X-Energy each fell about 5%.

The other company filed under the same fuel-and-fabrication heading shows what the market is actually repricing. Lightbridge, a 13-person Reston, Virginia developer of helically twisted metallic reactor fuel, has reported no revenue in any quarter since 2021. Its first-half net loss widened to $12.1m from $8.3m as research spending doubled, diluted shares rose 43.3% year over year to 31.9m, and it held $237.5m of cash against a $258m market value — about $20m of implied value for the entire fuel program, and 0.98x book. Its calendar belongs to Idaho National Laboratory: samples came out of the Advanced Test Reactor on May 6 and must cool before examination begins later this year. Selection for the Department of Energy's Launch Pad program on August 31 did not stop the slide.

The verdict splits cleanly. BWXT's earned deterioration is margin mix — a lower-margin acquired segment carrying the growth — and that is worth something, but not a third of the company's value. The rest is the unwind of a narrative premium that priced microreactors and small modular reactors, sold off now on the same impulse that is marking down pre-revenue developers. What nothing in the numbers explains is why a business whose cost discipline improves its fee margin trades on the sentiment of companies that have no fees at all.

About 55% of second-half earnings fall in the December quarter, management said. A naval contractor spends the year proving it can spend less; the proof arrives in the last three months.

Venture Global exported 127 cargoes from its Gulf Coast terminals in the June quarter at a 54% EBITDA margin, and then raised its earnings forecast on the assumption it keeps getting paid that way. The company — an Arlington, Virginia developer that builds and runs modular liquefaction plants at Calcasieu Pass, Plaquemines and the unfinished CP2 — lifted 2026 consolidated adjusted earnings before interest, taxes, depreciation and amortization guidance to $8.7–9.1bn from $8.2–8.5bn, struck against a market liquefaction fee of $12.50–13.50 per million British thermal units.

That number is a spread, not a fee schedule. Venture Global disclosed that a $1 change in it moves 2026 EBITDA by $180–210m and 2027 EBITDA by $650–700m — the second figure larger because the volume behind it is larger. Cheniere, the Houston operator of Sabine Pass and Corpus Christi and the only US exporter with a longer record, disclosed the mirror image on its August 6 call: under 1 million tonnes of 2026 output is unsold, and a $1 move in market margins is worth less than $50m of full-year EBITDA. One company rents machines at $2.50–3.00 per million British thermal units of fixed liquefaction fee for twenty years. The other sells the gas.

Why the spread is open

The reason Venture Global's realized fee is quadruple Cheniere's contracted one is physical. Iranian strikes on Qatar's Ras Laffan complex in March 2026 destroyed two trains totaling 12.8 million tonnes a year, and Qatar extended force majeure on European and Asian deliveries again in late August as the Strait of Hormuz stayed shut. The feedgas side is cheap at the same time: the Energy Information Administration cut its 2026 Henry Hub forecast to $3.44 per million British thermal units in its August outlook, from $3.67.

Selling into that gap is policy, not accident. Chief executive Michael Sabel told investors the portfolio aims at a median liquefaction fee that historical data suggests is "nearly twice that of a 20-year contract price," and said the company has "several years of commissioning cargoes from CP2 and bolt-on projects coming." Venture Global's own filings put Plaquemines Phase 1 in service in the fourth quarter and asked regulators to extend its in-service deadline to the end of 2027, which lengthens the commissioning window.

The bill for the first round of this arrived last October, when an arbitration panel found the company had failed to declare commercial operations at Calcasieu Pass in timely fashion. BP is now seeking at least $3.7bn and potentially more than $6bn; a Venture Global spokesperson called the claim "unserious and not supported by evidence or controlling law." A separate panel went the company's way against Shell, and a New York court refused to overturn that award in March. The next Calcasieu Pass hearing is in late November.

What conversion costs

The estimates already answer the question the model poses. Consensus has Venture Global's 2027 revenue falling 13.8% to $16.02bn and earnings per share falling 52% to $0.807, while Plaquemines Phase 2 and CP2 roughly double what it can produce. On that 2027 number the $15.50 share is 19.2x, against a forward 9.18x on 2026 that flatters a single year. Trailing free cash flow yield is -24.9% — the arithmetic of a construction program.

Cheniere is the opposite balance sheet and the duller story: guidance raised twice, to $7.9–8.4bn, June-quarter revenue up 26.3%, and a price-to-gross-profit of 4.94x against 5.37x in early May, so the shares got cheaper on that measure as they rose. Its headline earnings are unreadable — a $3,502m net loss in the March quarter sat above $1,973m of operating income — because gas-supply contracts were marked to market; three-quarters of those volumes moved to normal-purchase accounting in mid-June.

Cheniere Energy Partners, the partnership that owns Sabine Pass, is the group's best twelve-month performer at +29.5% on the weakest business: June-quarter revenue growth decelerated to 5.2%, consensus has 2026 EBITDA down 1.5%, and distribution guidance of $3.10–3.40 per unit was reconfirmed rather than raised. It is the only one of the three whose forward earnings multiple, 14.67x, sits above its trailing 11.51x.

The verdict divides cleanly. Cheniere's advance is earned by contracted volume and is indifferent to the Qatari outage. Venture Global's is earned too — but by the outage itself, and the same delay that produced the margin produced the damages claim. Leadership inside the group flipped in a month: the partnership led thirty days of trading in mid-August and now trails it, up 2.3% against Venture Global's 16.1%.

The hearing in November and the Plaquemines commercial operations date fall within weeks of each other. One of them ends the commissioning window; the other prices what it already cost.