DK Street Journal

A price cut with no competitor

Amphastar Pharmaceuticals, the Rancho Cucamonga maker of generic and proprietary injectables, inhalation and intranasal medicines, sells the only intranasal glucagon rescue treatment on the US market. In the June quarter it collected less for it than a year earlier: BAQSIMI net sales of $45.5m, down 3%. No generic version has been approved. The price came down through rebates and discounts.

The complex-generic argument rests on manufacturing difficulty — build something few rivals can replicate and the price holds. BAQSIMI shows the buyer can take the price anyway, and that matters because Amphastar's margin has been going one direction for three years: 54.5% gross margin in 2023, 51.1% in 2024, 49.5% in 2025. Operating income fell 31.7% last year to $140.4m and diluted earnings went from $3.06 a share to $2.03.

Margin that has to be re-bought each year

The March quarter was the trough, at a 41.1% gross margin against 50.0% a year earlier on flat revenue. June recovered to 50.8% on revenue of $183.9m, up 5.4% — and the company located that recovery in recently launched ipratropium bromide, teriparatide and iron sucrose, partly offset by lower average selling prices on BAQSIMI, glucagon and epinephrine multidose vials. Operating income still fell 5.8%. Meanwhile the legacy glucagon emergency kit, which BAQSIMI replaces, fell 45% to $14.1m in the December 2025 quarter.

Two bills land on top. On 2 July the Food and Drug Administration sent Amphastar's IMS subsidiary a warning letter over manufacturing-practice violations at South El Monte, California, following a December inspection; distribution was not restricted, but remediation adds $2m to $3m of expense a quarter for several quarters. And BAQSIMI's own success triggered a $100m milestone payment owed to Eli Lilly in the third quarter, for clearing $175m of annual net sales.

The reshoring option nobody is paying for

Amphastar is, on paper, the cleanest domestic beneficiary of US trade policy: all finished products are made in the United States, and in July 2025 it announced a plan to quadruple capacity at Rancho Cucamonga. Imported generics face zero tariffs until August 2028, then 100%, then 200% a year later, while India supplies close to half of American generic medicines. "Patients pay the tariff," Sandoz chief executive Richard Saynor said on 8 September. Amphastar still imports active ingredients and starting materials, so the hedge is partial — and over twelve months the shares fell 2.7% regardless.

That decline did rational work. Consensus has 2026 earnings at $2.97 a share, down 13.6% from the $3.44 that stood for 2025, leaving the stock at 8.8x forward earnings against 15.0x trailing, with a 13.3% trailing free-cash-flow yield — the highest of the generic makers. Diluted share count has fallen 15% since late 2024, to 44.2m, so the reported earnings decline understates the drop in absolute profit.

What the past month does not explain is the rally: up 22.1%, including 11.2% across four sessions into 24 September, with no company announcement discoverable. Barclays lifted its price target to $23 from $20 while keeping an Equal Weight rating — a target below the $26.13 close. Management reaffirmed mid- to high-single-digit sales growth for 2026.

The business earns the cheap multiple and the cash yield; it does not yet earn the month. A margin that must be repurchased with fresh approvals every year is a different asset from one protected by manufacturing barriers, and BAQSIMI is the proof — the price fell while the barrier held.

The $100m check to Lilly clears this quarter, paid on a product already being discounted with no competitor anywhere in sight.

Alignment Healthcare told investors on 15 September that its third-quarter medical costs were running hot, and named the reasons: hospitals billing more aggressively, a heavier load of disputes and appeals, and longer skilled-nursing stays. At Baird's Global Healthcare Conference in New York the Orange, California insurer — which owns Medicare Advantage health plans in California, North Carolina and Nevada and also delivers care to members of unaffiliated plans — also disclosed $10m to $11m of unplanned investment earmarked for the second half. Full-year guidance was left unchanged. The shares fell 19.9% that session, from $12.94 to $10.37, on 28.5m shares against 5.7m the day before.

That matters beyond one session because of what the company says is causing it. The mechanism management gave is not utilization running above the bid it filed in June but a claims-system transition that left 2025 hospital bills landing in 2026 — a timing problem inside a business whose revenue is a per-member rate set annually in Washington and paid monthly regardless. "Whenever you go through a claims application implementation, those are messy and noisy... We are pretty much through the end of that, and there is a lot of data points that would suggest … all that noise is starting to settle down," chief executive John Kao said at the conference.

The meters went the other way

June-quarter revenue rose 31.6% to $1,336m, operating income rose 85.1% to $42.1m and net income more than doubled to $36.6m. Reported gross margin was 13.55% against 13.15% a year earlier, implying a medical benefit ratio — the share of premium consumed by claims — of 86.45%. The company's own adjusted figure was 86.3%, about 0.4 points better year over year, with the midpoint of membership, revenue and adjusted EBITDA guidance raised on 30 July. Guidance has been lifted or reaffirmed at every point in this decline, never trimmed.

Nor does an industry cost shock fit the window. UnitedHealth Group said its 2026 Medicare cost trend is running below its initial estimate of about 10%, with care patterns becoming more predictable. Humana is up 56.4% over twelve months and CVS Health 19.5%. Alignment is down 52.6%.

Clover, same book, opposite direction

Clover Health runs a capitated Medicare Advantage book roughly half Alignment's size out of Franklin, Tennessee, built around physician software it also licenses to outside providers through its Counterpart Health arm. June-quarter revenue rose 56% to $743m and Medicare Advantage membership 48% to 157,309, guidance went to $2.92–3.0bn of revenue, and it closed the quarter with $443m of cash and no debt. It carries a $2.33bn market value against Alignment's $1.70bn on half the sales.

On price to gross profit — the measure that survives a 13.5% margin sitting beside a 20.6% one, and Clover's trailing loss — Alignment now fetches 2.60x forward against 4.24x for Clover. On 10 September the two were 4.13x and 4.10x. Seventeen sessions opened the entire gap.

What the fall earns

The part the business explains is thin. Nothing in the reported numbers deteriorated; what changed is the market's willingness to underwrite a company still spending for growth while the largest carriers deliberately trade enrollment for margin. The genuinely unresolved part is legal and regulatory: a securities investigation opened after the Baird disclosures, and TD Cowen's estimate that losing the fourth star on Alignment's H3815 contract would cost roughly $175m of bonus revenue annually. Clover's own advantage is no safer — its half-star court win, worth an estimated $120m in 2027, went to the Eleventh Circuit on appeal on 21 July. And on earnings Alignment is not cheap at all: 45.2x forward against 41.2x trailing, because consensus 2026 earnings of $0.18 a share sit below the trailing base.

The Centers for Medicare and Medicaid Services publishes the 2027 star ratings in early October, and about half of the cut points got harder. Alignment's claim that essentially all its members will again sit in four-star-or-better plans rests on a preliminary read of that data. The confirmation arrives days before seniors start choosing.

A downgrade about someone else's product

Twilio, whose roughly 1,800 programming interfaces let developers embed text messages, voice calls and email inside their own applications and which bills per message sent and per minute of voice, closed at $275.80 on 25 September after falling 7.96% in a session. The cause was a rating change. HSBC's Sameer Lam moved the stock to Reduce from Hold while keeping a $211 price target, arguing that rising AI-agent activity "does not necessarily translate into a larger share of the resulting revenue for Twilio."

That sentence is the argument of Twilio's entire September. The company has published no financial figure since 6 August. Everything that moved the shares since happened elsewhere: Meta launched Muse, its consumer AI agent, on 8 September, built so that talking to it works like messaging another person, either standalone or inside WhatsApp. Desks then repriced Twilio as the toll road for agent chatter — TD Cowen's Derrick Wood to $300, Wells Fargo to $275 and Rosenblatt's Catherine Trebnick to $290 — and the stock ran 22.9% in four sessions to a $299.66 close on 24 September. Price to trailing gross profit went from 12.97x on 22 August to 15.48x on 25 September, against 14.37x forward, on no company disclosure in between.

What Twilio can actually show

The business underneath is accelerating. Revenue growth ran 14.3% in the December quarter, 20.0% in March and 22.0% in June, to $1,499m. Dollar-based net expansion — what existing customers spend against a year ago — went 108%, then 114%, then 116%. Free cash flow grew 34% to $353m, and June-quarter messaging revenue grew 28%.

Two facts cut the other way. Gross profit grew 20.4% while revenue grew 22.0%, because US carriers raised application-to-person messaging fees three times this year and Twilio books those surcharges in revenue at no margin: roughly $190m of incremental pass-through in 2026, of which $71m explained about a quarter of the June-quarter revenue increase. Gross margin slipped to 48.42% from 49.07%. And Twilio's own guide for the September quarter is organic growth of 11–12%, after posting 17%. Consensus has revenue growth falling to 11.4% next year.

The seat-billed side rose too, on arithmetic

Salesforce, which sells sales, service, marketing and analytics software to enterprises on multi-year per-seat subscriptions, gained 13.8% over the same thirty days. Its June quarter grew revenue 10.8% to $11,345m — about 6% excluding Informatica, the data-management business absent from the year-ago base. Gross margin fell to 76.65% from 78.10% and GAAP operating income was flat at $2,331m. Diluted earnings per share nonetheless rose 119% to $4.29, because the share count fell from 962m to 821m under a $25bn accelerated repurchase struck at an average of about $176.

One meter genuinely confirms: current remaining performance obligation, the contracted work due within a year, reached $33.5bn, up 14% in constant currency and a point ahead of guidance. Marc Benioff, chief executive, spent the 26 August call on the partnership behind it: "The number one AI in the world, Anthropic, and the number one CRM, Salesforce, coming together for the first time in an incredibly powerful way to build a new product called Claudeforce." Agentforce annual recurring revenue is $1.5bn.

Investors gave much of it back anyway. Salesforce fell 8.5% from 15 to 25 September, across a global outage on 16 September that struck during its own Dreamforce conference, even as Guggenheim raised its target to $300 and Mizuho to $280. At 13.97x forward earnings against 21.27x trailing and a 7.9% free-cash-flow yield, it is the cheapest large-cap in software on earnings. Measured against gross profit it is 5.64x — dearer than HubSpot, the mid-market customer-management vendor, at 3.96x, and Adobe at 4.05x, while growing organically slower than either.

The per-seat control is not clean

Adobe, which sells Creative Cloud and marketing software by subscription, fell 14.0% in the month on a record $6.76bn quarter and raised guidance — undone by net new recurring revenue down 36–37% and remaining performance obligation up just 8%. Figma, the browser-based design subscription, grew 48.2% to $370.1m, the fastest in the group, and fell 23.5%, though 77.7m lock-up shares came free on 8 August. Braze, which bills on messages sent rather than seats, fell 20.1% with revenue accelerating to 30.2%.

The verdict

Consumption-billed vendors did beat per-seat ones this month, but not because the market found evidence of usage. Twilio's re-rating is an inference about a Meta product, and Twilio has disclosed nothing that tests it; HSBC's objection is that traffic and revenue share are separate questions. Salesforce's rise is accounting: flat operating income converted into 119% earnings growth by retiring stock. Both moves are financed by things outside the reported quarter.

Twilio's September quarter will be the first disclosure that can show whether agent traffic arrived. Management has already told investors to expect organic growth of 11–12%.

Clorox spent fiscal 2026 explaining an inventory swing it caused itself. Before replacing the US software that turns a retailer's purchase order into a truckload, its largest customers pre-bought stock to protect shelf availability, then ran that stock down instead of reordering. Reported organic sales fell 13% for the year to June, of which about 13.5 percentage points was the lapping of those pulled-forward shipments, the company said — and US consumption was back to flat by the June quarter.

That wedge has closed, and the shares have fallen roughly another tenth since late August to a 52-week low of $82.47, against a $128.90 high. What the market is pricing now is not shipments. It is the cost line and the balance sheet of a company whose entire shareholder case is a dividend it has raised for 49 consecutive years.

The shortfall was one quarter, the margin damage was all of them

The drawdown is localized: revenue fell 18.9% in the September 2025 quarter and then moved barely at all — down 0.8% in December, up 0.1% in March, down 2.0% in June. Profitability is the part that did not normalize. Gross margin fell 290 basis points for the full year, to 42.3%, and in the June quarter it fell 521 basis points to 41.3%. Diluted earnings came in at $4.81 a share against $6.52.

Fiscal 2027 guidance concedes the rest. Net sales are guided up 13% to 14%, but roughly 9.5 points of that is the GOJO acquisition and more than 3.5 points of the 3.5%-to-4.5% organic figure is pure arithmetic from lapping last year's drawdown; strip both and underlying growth is flat to slightly up. Gross margin is guided roughly flat at about 42% because more than $200m of inflation — "more than double our historical range" — eats the cost savings. "We expect the operating environment to remain challenging, with continued cost volatility and a value-seeking consumer," Chair and Chief Executive Linda Rendle said of fiscal 2027 in the 3 August results release. At the Boston Global Staples Conference, management said it will take a straight price increase on Glad trash bags, the part of the portfolio most exposed to private label, while conceding it does not know how competitors will respond.

The dividend and the cash that has to fund it

Clorox reported fiscal 2026 net cash from operations of $612m, 9.1% of sales, and an adjusted free cash flow of $881m that adds back a $476m venture-termination payment. On the reported basis, free cash flow was closer to $405m — against roughly $600m paid out in dividends. The board still raised the quarterly rate to $1.25, lifting the annualized payout to $5.00 and the yield to near 6% at the September price. Fiscal 2027 free cash flow is guided to 11%-13% of net sales, about $840m to $990m on guided sales, which would cover the payout roughly 1.4 to 1.6 times. It has to, because the April purchase of GOJO Industries for $2.25bn in cash left leverage at about 3.6x earnings before interest, taxes, depreciation and amortization, against a stated target near 2.5x by the end of 2027.

The shares now change hands at 14.5 times forward earnings against 17.4 times trailing and a history above 20x, and 16.8 times enterprise value to the same earnings measure with the acquisition debt aboard. Consensus fiscal 2027 earnings of $5.79 sit inside management's own $5.70-$6.00 range, so estimates have been cut to meet guidance rather than left hanging above it. Brokers have followed the price down: Wells Fargo cut its target to $90 from $102 on 11 September and Barclays moved to $84.

Not the category

Staples sold off generally in September — 14 of 18 large US names fell over the thirty days to 25 September while the S&P 500 barely moved, the likelier reading being rotation out of defensives as long Treasury yields pushed higher. But Clorox's 20% fall is about twelve points worse than the group's middle, and the two obvious comparisons refuse to cooperate. Procter & Gamble, which sells Tide and Dawn into the same aisles, grew revenue 3.3% to $87.0bn in the year to June and trades near 21 times forward earnings. Church & Dwight, owner of Arm & Hammer, OxiClean and TheraBreath, grew organic sales 5.8% in the June quarter on 4.3% volume, widened gross margin by 240 basis points and raised its 2026 outlook twice; it is up over the past twelve months and trades above 25 times forward earnings. Chief Executive Rick Dierker called the company's "evergreen model" intact at the Barclays staples conference on 9 September. Household cleaning demand is not contracting; Clorox's discount of roughly a third to Procter & Gamble is its own.

So the earnings decline is earned and largely explained: a self-inflicted shipment gap plus real input inflation, with cleaning holding share for an eighth consecutive quarter while Fresh Step, Glad and Kingsford do not. What the last month added is not an operating fact but a financing one — a levered balance sheet and a payout that reported cash flow did not fund last year. At 11% of sales, the guided cash conversion makes the September selloff look like an overshoot; at fiscal 2026's 9%, the yield is the market telling you the streak is the thing at risk.

The first quarter of the guided year is the one where last year's drawdown flatters the sales comparison automatically. The cash line gets no such help.

Ascendis Pharma got its obesity chemistry back on September 14, and its shares fell 9.5% the next session on 2.96m shares against roughly 200,000 in the days before. Novo Nordisk terminated the 2024 collaboration whose lead program was a once-monthly version of semaglutide built on the Danish company's TransCon carrier, handing the rights back across metabolic and cardiovascular disease. The pact had carried up to $285m in upfront, development and regulatory payments. Neither side now owes the other anything.

That one day is most of Ascendis's 10.1% decline over the past month, and it is a clean test of what the market pays for in mid-cap biotech — an option on a partnered obesity drug, or approved medicines already billing. Ascendis lost the first. The second is what the company actually is.

Paid by the patient, by the week

Ascendis, headquartered in Hellerup and worth $14.0bn, sells three rare-disease drugs built on one prodrug chemistry that tethers a hormone to an inert carrier so a daily injection becomes a weekly or monthly one: Skytrofa for growth hormone deficiency, Yorvipath for hypoparathyroidism, and Yuviwel for pediatric achondroplasia. Second-quarter revenue reached $344.9m, up 118.2% year over year, at a 90.6% gross margin. Operating margin came in at 18.4% against -33.5% a year earlier, the fifth consecutive quarterly improvement.

Reported in euros on the August 13 call, product revenue was EUR 315m, of which Yorvipath supplied EUR 252m on roughly 1,000 unique new US patients per quarter; Skytrofa contributed EUR 55m and Yuviwel EUR 8m in a first partial quarter with more than 220 enrollments. The economics are a subscription, and the leak is retention. "The majority of the drop-off is during that titration period in terms of when patients experience the most amount of change and where additional education and a higher touch support model makes sense," Jay Wu, president of Ascendis US, told investors on August 13.

The company closed the quarter with EUR 812m of cash, EUR 1.4bn of equity, no bank or convertible debt, and repurchased EUR 56m of its own shares while guiding to operating cash flow above EUR 500m for 2026. No guidance was trimmed. Annual revenue has gone from $7.8m in 2021 to $691.7m in 2025 while the annual operating loss narrowed from $451.8m to $130.9m.

The other value transfer inside the drawdown window went the other way in the market's reading. On August 30-31 Ascendis and BioMarin settled their patent disputes globally, with Ascendis paying 20% of US Yuviwel net sales retroactive to first sale and 18% in the EU, Brazil and South Korea until May 2030. The shares rose 5.8% that session: an import-ban risk before the US International Trade Commission was removed, and the royalty was the price.

What the launch costs

Cytokinetics is the nearer comparison and the sharper contrast. The South San Francisco company launched MYQORZO, an oral cardiac myosin inhibitor, in the US in January against Bristol Myers Squibb's incumbent Camzyos; product revenue went from $4.8m in the first quarter to $25m in the second. But total revenue fell 52.2% as collaboration income rolled off, the operating loss widened 57.7% to $175.9m, and consensus shows no net profit until 2029. It trades at 53.7x forward sales — roughly five times Ascendis's 10.4x, for a business three years from earning anything.

The premium is not what broke

Seven of the eight names on Barron's SMID-biotech takeout list fell over the past thirty days. Strip Dyne and MoonLake, each with its own datable blowup, and the remaining six averaged -9.6% against the equal-weight biotech benchmark XBI's roughly 8.3% decline. That residual is duration: the Federal Reserve's benchmark at 4.00% and a 10-year Treasury yield at its highest in nineteen years discount pre-profit biotech hardest. Acquisition appetite is not deflating — first-half 2026 biopharma deal value reached about $130bn across 42 deals, with average deal size up to $3.1bn from $2.7bn, and Novartis paid a 46% premium for Avidity. Alnylam, the one member too large to be bought, is the only one up over the period, having done its own damage on a guidance cut in August.

So the verdict on Ascendis splits cleanly. The market took roughly a tenth off a company whose operating line inflected, in exchange for one forgone partnership option and one royalty that bought away a trade-ban risk. Neither counterparty event touches Yorvipath's per-patient revenue, and the earnings multiple that would normally arbitrate is useless here — first-quarter net income of $639.7m sat on $25.2m of operating income, an artifact of one-off items.

The obesity program Novo returned had not yet earned a dollar. The thousand American patients starting Yorvipath each quarter already have.

Tandem Diabetes Care, which sells the t:slim X2 durable insulin pump to insulin-dependent patients, shipped a record 33,000 pumps worldwide in the second quarter and left its full-year outlook of $1.065–1.085bn alone. Its shares are down 27.7% over the past month and have surrendered the whole 19.1% gain they made the day after that report.

Dexcom, Insulet and Tandem sell the same patient the same promise — hardware or a disposable that must be replaced for life — and are paid for it in three incompatible ways. Dexcom bills per disposable glucose sensor. Insulet bills per Omnipod, pay-as-you-go through the pharmacy counter, with no durable device sold first. Tandem sells a pump up front against a roughly four-year replacement queue. In a final rule issued on 28 November 2025, Medicare decided that two of those three models get paid differently: continuous glucose monitors and insulin pumps are reclassified as "frequently and substantially serviced", moved onto continuous monthly rental and pushed into competitive bidding, with roughly ten national contract suppliers expected in the category. Bidder registration and the bid window open in late summer and early fall of this year; single payment amounts arrive in 2027, contracts on 1 January 2028. Congressional Diabetes Caucus leaders have formally objected that the change will complicate patient access, so the eventual price is unresolved — which is precisely the shape of a discount that accrues over a quarter instead of a morning.

The pump that loses its up-front payment

Tandem's revenue grew 5.8% in the quarter, a crawl beside its peers, but the operating line moved: gross margin of 56.9% against 52.3% a year earlier, and an operating loss narrowed 73% to -$13.8m. It also reaffirmed 94,000–95,000 US pump shipments for the year and filed its tubeless Mobi pump with the Food and Drug Administration. "Getting into that market … is gonna give us access to a significantly higher interest level and it is going to drive meaningful growth," chief executive John Sheridan said on the 6 August call, noting tubed pumps grow at mid-single digits while tubeless grows above 20%.

The shares now fetch 1.00x forward sales, down from roughly 1.66x in late February on essentially unchanged consensus revenue. Two identifiable reasons exist. Beta Bionics, whose iLet pump reached about 10% of new-patient share and grew second-quarter sales 38%, rose 42.8% over three months while Tandem fell — a share shift, not a shrinking market. And Tandem is the single name whose payment model Medicare is rewriting: the up-front durable-equipment sale becomes a bid-set monthly rental. Truist cut its price target to $28 from $31 in late September while keeping a buy rating, citing recalibrated medtech valuations rather than demand.

The pod Medicare cannot bid

Insulet's second quarter, reported 5 August, was $801.7m of revenue, up 23.5%, above its own guidance range, with international Omnipod up 35.5% and gross margin at 70.2%, a five-quarter high. The break was retention: it trimmed full-year US Omnipod growth to 17–19% from 20–22% because type 2 customers were staying and using less than expected, and the stock fell 20.9% that session. "While we are updating our outlook to reflect what we're learning as we scale in type 2, our conviction in the long-term opportunity remains unchanged," chief executive Ashley McEvoy said. The shares trade at 20.9x forward earnings against 25.4x trailing; a year ago the same FY2026 consensus of $6.50 a share implied close to 50x. Meanwhile Omnipod sits inside the Medicare Part D pharmacy benefit, which is not subject to competitive bidding.

The one that rose

Dexcom, the sensor maker, is the group's advancer, up 26.8% over twelve months. Its economics improved rather than eroded: gross margin of 63.4% versus 59.5%, adjusted gross margin up 400 basis points to 64.1% with full-year guidance raised, and its own filing crediting higher sensor volumes and favorable revenue per customer, partly offset by channel mix and rebate eligibility. It also won coverage across the four largest commercial pharmacy benefit managers. "We carried forward solid demand from the first quarter for Dexcom CGM globally as we benefited from broader access and share gains," chief executive Jake Leach told investors on 30 July. But revenue growth halved, from 21.6% a year ago to 13.1%, and the forward multiple went the other way: about 22x at the 30 April low, 32.5x now on an unchanged $2.67 of FY2026 consensus earnings, barely below its 33.4x trailing. Dexcom straddles both payment worlds, with a rental ceiling illustrated at $272.69 a month on the Medicare side.

What each part earns

The last month was not kind to devices generally — Stryker fell 16.9% and Abbott 9.6% while the S&P 500 was flat — so Dexcom's and Insulet's declines were milder than the peer median, and Tandem is the outlier that needs explaining. The most-cited structural bear case has meanwhile inverted: Dexcom data at this year's American Diabetes Association meeting showed sensor use nearly quadrupling among non-insulin users after starting a GLP-1 drug.

So the margin expansion Dexcom earns; the 46% expansion in its multiple while growth halved, nothing in the operating line explains. Insulet's guidance cut is worth about 300 basis points of US growth, which does not buy a 58% compression in its multiple, though retention is the first genuinely new fact anyone has about patient demand in this niche. And Tandem's discount rests on two things that are real but dated far out — a rival taking new starts today, and a payment change that sets prices in 2027.

The number nobody has yet is the single monthly amount Medicare will pay for a pump. Tandem trades as though it has already been published.

Florida's regulators set the profit Florida Power & Light is allowed to earn at 10.95%, on a 59.6% equity layer, and locked it there through 2029. The money that pays for the plant reprices every morning.

Over three sessions ending September 25 the 10-year Treasury yield reached about 5.20%, its highest in nearly two decades, on a hotter-than-expected September flash purchasing managers' survey and roughly two-to-one odds of an October Federal Reserve increase. Utilities and property trusts led the losses, the Utilities Select Sector SPDR fund falling 1.9% on September 24 alone. NextEra Energy shares fell 4.0% across the three days, Dominion Energy 2.8% and Entergy 2.6%, against 0.3% for the S&P 500 exchange-traded fund.

Why a yield move lands on the income statement

None of that is a demand event. All three spend more on construction than their operations generate — Dominion's trailing free cash flow is negative by 12.4% of its market value, NextEra's by 6.2% — so each dollar of the buildout arrives from the securities markets at whatever they charge and is repaid at a return a commission fixed years earlier. Duke Energy's August sale of $1.75bn of equity units priced at an all-in distribution rate of 7.75% against a 9.8% allowed return: roughly two points of spread, and narrowing. Virginia's commission set Dominion's authorized return at 9.8% against the 10.4% asked, and trimmed its 2026 base-rate increase to $565.7m from $822m sought.

The awkward case

NextEra owns Florida Power & Light's 5.7m customer accounts and the largest contracted clean-energy development book in the country. Its June quarter grew revenue 12.4% to $7.53bn and operating income 17.1% — faster than sales — on a diluted share count up 1.3%. "NextEra Energy delivered a strong second quarter, with adjusted earnings per share increasing by 9.5% year-over-year," chairman and chief executive John Ketchum said of results reported July 24; the September investor presentation aims at the high end of $3.92 to $4.02 in adjusted earnings a share. At the September 23 close the shares were about 22% below the $98.75 high set May 1. The price the market pays for each dollar of NextEra's gross profit has fallen to 7.93 from 11.73 in May.

Dominion is no longer setting its own price. Since May 18 it has been the target of an all-stock NextEra offer at a fixed 0.8138 shares, a $66.8bn combination both shareholder bases approved on September 3; at the September 25 closes the implied ratio was 0.7976, a 2.0% discount to terms. Its quarter — revenue up 19.6%, operating income down 2.4% — now feeds an arbitrage spread rather than a valuation. The live variable is Richmond: Virginia attorney general Jay Jones asked the State Corporation Commission on September 21 to restart the review clock, arguing the companies "have put a different deal on the table", which would push a January 11 decision deadline into March.

Entergy, the New Orleans utility serving about 3m customers across Arkansas, Louisiana, Mississippi and Texas, is where the "failing to earn it" case has teeth. June-quarter revenue rose 5.9% while operating income was flat, and diluted earnings slipped to $1.03 a share from $1.05 because the share count rose 4.6%. That dilution funds a $67bn five-year plan whose largest contracted block still has no approved rate behind it: Louisiana regulators decide in December on seven more gas plants for Meta's Hyperion campus. "We continue to have seven to 12 gigawatts of hyperscale data center potential in our pipeline," chair and chief executive Drew Marsh told investors on July 29. At 25.6 times trailing earnings, down from 29.9 in May, Entergy still sits well above the roughly 16.8 long-run median for vertically integrated utilities.

The verdict

The de-rating splits three ways. Entergy's is largely earned — an expensive starting point, flat operating income, real dilution. NextEra's is not explained by anything in its own results; the business accelerated through the four months its gross-profit multiple lost a third. And the sector averages that make the whole group look wrecked lean on PG&E, down 32.4% in thirty days on California wildfire-liability legislation, which has no load or allowed-return content whatever. What binds the three is the spread between the cost of the next dollar and a return already fixed by order. A demand failure would show up as cancelled contracts; this one shows up in the coupon.

Virginia's new rate class for customers above 25 megawatts, effective January 2027, obliges a data center to sign for 14 years and pay for 85% of its contracted delivery demand whether it draws the power or not. The load has been made contractual. The financing has not.

Hilton Grand Vacations sells timeshare weeks and then lends most buyers the purchase price, which makes it half resort developer and half consumer finance company. In the June quarter it ran 239,000 sales tours, up 6% and the fourth consecutive quarter of tour growth. Its shares have lost a third of their value since late June.

That pairing is the shape of the whole travel complex. Over the thirty days to September 25, casino, timeshare and aircraft-parts shares fell between 12% and 25% — MGM 24.9%, Hilton Grand Vacations 20.3%, Wynn 17.8%, Las Vegas Sands 17.5% — while the S&P 500 tracking fund rose 0.8%. These businesses are paid in incompatible ways for the same trip: per financed timeshare interval, per booked transaction, per unit of gaming win. Read one meter at a time, and only one of them shows a customer walking away.

Per interval: the credit line has not cracked

Hilton Grand Vacations' honest top line is contract sales, and they slipped $24m to $810m against the June 2025 quarter — with tours up 6.1% and volume per guest down 8.6%. That is a price-per-tour problem. "[S]ales execution fell short of our expectation, which weighed on overall sales productivity," chief executive Mark Wang told investors on the July 30 call, pointing at high-volume Bluegreen locations in Orlando and Myrtle Beach. Management trimmed full-year volume-per-guest guidance to a low-to-mid single-digit decline, from flat to slightly down, and missed on adjusted EBITDA, $265m against $291m expected.

The credit book, where a genuine consumer break would show first, did not cooperate with the bear case. The loan-loss provision ran near 17% of contract sales, the high end of the mid-teens target, but the company attributed it to a higher borrowing propensity and more trust-product mix, with delinquencies stable or improving at Bluegreen after underwriting changes. Shares trade at 8.0 times trailing earnings before interest, taxes, depreciation and amortization, a 15% trailing free-cash-flow yield, and 16.6 times trailing earnings against 7.4 times forward — a forward figure resting on consensus 2026 earnings of $4.85 a share versus $0.89 actually reported in 2025. What the company does pay for directly is the rate: it securitizes its own receivables, and the ten-year Treasury note closed September 25 at 5.17%, its highest since June 2007, with swaps pricing three more quarter-point increases.

Per transaction: volumes grew and the loss grew faster

The supposed collapse in corporate-travel software is one stock. Global Business Travel Group has been pinned in a 1.2% range since July because it sits under a definitive all-cash merger at $9.50 a share with Long Lake Management agreed May 4; its June-quarter transactions grew 45% with 95% customer retention.

The 33.6% faller is Navan, which sells artificial-intelligence-driven travel and expense software to corporate finance departments. "This quarter, total GBV, this is travel bookings on the platform, grew by 45% year-over-year to more than $3 billion," chief executive Ariel Cohen said on the September 9 call, adding that requests for proposals had tripled year on year in the first half. Revenue rose 35.4% to $232.8m, guidance went up for the second time, to $927–933m for the year — and the operating loss widened to $27.6m from $12.3m as costs outran the revenue beat. At about 5.0 times forward sales against 5.6 trailing, that is a margin verdict on a growing book of trips.

Per unit of win: the one real deceleration

Las Vegas Sands is the exception. Revenue fell 0.7% in the June quarter after four quarters of 24–26% growth, operating income dropped 22.8%, and the market it is paid by has turned: Macau gross gaming revenue fell 1.2% in August, a third straight monthly decline. Even here the quarter was partly luck — Sands China's mass gaming revenue grew 8% against a 4% market, and Macau earnings were dented by a record-low 1.35% VIP hold worth roughly $87m while rolling volumes rose 73%. "[T]he customers are there and the productivity is there if the product is right," chief executive Patrick Dumont said on the July 22 call, reaffirming a $700m quarterly Macau earnings target. Sell-side estimates have started to follow the regulator's prints down: Morgan Stanley cut its 2026 Macau forecast and Citi looks for September down about 2%, while consensus still carries 8.8% revenue growth for the year. At 15.1 times trailing earnings and 12.6 forward, most of that de-rating is done.

The parts business says seats are still full

If trips were disappearing, aircraft maintenance would show it first. AAR, which distributes aerospace parts and performs airframe and landing-gear overhauls, grew fiscal 2026 revenue 19.0% to $3.31bn with a 23.0% final quarter, and its decline is characterized as a valuation reset rather than deteriorating fundamentals — 20.2 times forward earnings against 23.8 trailing. VSE, the aviation-parts distributor and repair shop, grew June-quarter revenue 65.0% with gross margin up to 17.0% from 11.4%, and had the most multiple to surrender: 65.7 times trailing earnings compressing to 26.6 forward. Neither company's revenue estimates were cut during the slide. The threat to them is forward flight hours, priced by fuel: jet fuel near $4.56 a gallon is roughly double its pre-war level with the Strait of Hormuz effectively shut, and United alone guides to about $6bn of extra fuel cost this year.

What the businesses earn and what they do not

One of these five sell-offs is earned by the operating record. Las Vegas Sands is falling because the monthly number it is publicly graded against is going down, and a third of its profit shortfall was hold rather than volume. The other four fell while volumes compounded, credit held and guidance went up rather than down — which points at the discount rate on long-dated cyclical cash flow, and at a fuel bill that raises the price of every trip without yet reducing the number of them. For Hilton Grand Vacations, that rate is not an abstraction but the cost of funding its own receivables; for AAR and VSE it is simply the arithmetic of a high multiple meeting a 5% ten-year.

Macau publishes its September gaming revenue in the first days of October. It is the only meter in this group that can validate the market's verdict on schedule; the rest are waiting on the bond market to say something different.

Fervo Energy, a Houston developer that drills geothermal wells with the horizontal-drilling and fiber-optic sensing methods of the shale patch, put power on the grid at its Cape Station project in Beaver County, Utah on 24 September — the first utility-scale enhanced-geothermal plant to do so. The first of three 33-megawatt blocks in the roughly 100 MW first phase is due at its contractual commercial operations date by 1 October. The shares rose 2.2% on the announcement and fell 9.16% the next session.

What that milestone puts to the test is whether always-on clean generation actually collects a scarcity price. Geothermal's entire pitch to data centers is a capacity factor near 90%, and there are two listed ways to own it, paid in incompatible ways. Ormat Technologies sells megawatt-hours into twenty-year contracts signed years ago; Fervo sells capacity that bills only once wells are drilled, financed and energized. Ormat realized roughly $91.5 per megawatt-hour across its portfolio in the first half. Fervo has told investors its behind-the-meter economics stay competitive in a $100–130 range.

Ormat's growth came from batteries and a turbine

On the consolidated line Ormat is doing well. June-quarter revenue rose 10.6% to $258.8m, gross profit rose 20.7%, and on 5 August management raised full-year guidance to $1.15–1.20bn. On the line the clean-firm-power thesis rests on, it is not. First-half electricity revenue rose 3.1% to $350.9m as generation rose 2.0% to 3.83 million megawatt-hours — an implied price barely above last year's. Operating income fell 3.2% in the quarter.

The growth sat elsewhere. Merchant energy storage revenue rose 195.1% to $42.8m on favorable spot pricing, at a margin management guided down toward 40–50%, and the first quarter carried a $105m turbine delivery for the Topp 2 project at engineering-and-construction margins. That delivery is why the product backlog fell to about $203m. "Demand for reliable, carbon-free baseload power continues to strengthen, particularly from data centers," chief executive Doron Blachar told investors, citing a 15-year portfolio contract of up to 150 MW to serve Google through NV Energy — a deal whose deliveries begin in 2028 and whose price is not disclosed.

Ormat now trades at 17.3 times trailing gross profit, against 20.1x in August and 25.9x in May, while that gross profit grew. Sales multiples flatter a business shifting into third-party construction work, so cash profit is the comparison that holds. On earnings the compression is harder to call unearned: 36.8 times forward profit rests on a consensus that has 2027 revenue down 1.7% and earnings per share down 3.5%, because the lumpy turbine year does not repeat.

Fervo's meters all improved

Every disclosed operating number at Fervo moved the right way. Binding offtake reached 1,054 MW and a $7.2bn revenue backlog after Google contracted 396 MW on 1 September. Its Sawtooth 7 well hit 19,500 feet and 460°F in 21 days, against more than 70 days for the original Project Red wells at shallower depth and lower heat. The 2030 capacity target went up to 1.1 gigawatts. "This is a gamechanger for the geothermal industry," chief executive Tim Latimer said on 24 September.

The shares are 43.8% below the $27 May listing price anyway. The binding constraint is money, not contracting: $2.1bn of cash against $228.4m of debt, second-half capex guided to $850–900m, and a Phase II target of $5,500 per kilowatt of installed capacity. Book value is the only multiple that means much here — 1.60 times, with most of the book being IPO cash — since consensus puts revenue at $72.4m in 2027 and first profit in 2029. Bank of America's Ross Fowler cut his target on 25 September while keeping a Buy, citing a lower peer group multiple of 12.8 times against 14.2 times — the analyst's own words for a re-rating rather than a forecast cut. Wolfe Research upgraded the stock the same day.

What the discount is for

Neither de-rating is an operating failure. The whole capital-intensive power complex fell together as the 30-year Treasury yield reached 5.49%, its highest since 2004, and cash flows arriving in 2029 lose the most present value when the long end moves. What is specific to geothermal sits underneath that: Ormat's incremental dollar now comes from low-margin construction and merchant batteries rather than contracted firm power, and Fervo's revenue line cannot begin converting its backlog until Phase II wells are funded and drilled. A 90% capacity factor is not yet showing up in either company's realized price in a way an investor can read.

About one in ten of Cape Phase II's roughly 80 planned wells is finished. The next number that matters is not a multiple — it is the first invoice Fervo sends.