DK Street Journal

Planet Fitness added no members between March and June. It closed the second quarter with 21.5m, level with the previous quarter and up 3.6% on the year, and the 1.7% increase in system-wide same-club sales came entirely from rate rather than bodies — against an 8.2% increase in the same quarter of 2025. A week earlier, Life Time had reported average monthly dues of $245, up 12.3%, and $993 of revenue per membership.

The two sit under one industry heading and are not in the same business. Planet Fitness owns fewer than one club in ten; it collects a 7% royalty on franchisee membership dues plus a national advertising levy, and sells franchisees their treadmills. Life Time designs, builds and operates large-format athletic clubs itself. Each grows on capital it does not fully control — franchisee balance sheets and the securitization market on one side, the sale-leaseback bid on the other — and this year that distinction has decided everything.

The guidance cut that reset the royalty

Planet Fitness shares fell more than 31% in one session on 7 May, the steepest drop since the 2015 listing, after net new member adds fell 36% and the company cut its full-year outlook: same-club sales to about 1% from 4–5%, adjusted EBITDA growth to roughly 6% from 10%. Management blamed messaging, competition in select markets, weather and the macro backdrop, and shelved a planned nationwide Black Card price increase. Consensus now models a 2.0% decline in this year's adjusted EBITDA — the first forecast earnings fall since 2021.

The June quarter shows where the reported growth came from. Franchise revenue rose 13.5% to $135.8m, but $10.1m of the $16.1m increase was a one-point rise in the advertising fund contribution rate, money the company collects and spends. Equipment revenue rose 4.1%, with 21 placements at new franchisee clubs against 19 a year earlier — modest re-engagement, though replacement sales were 85% of the mix. Operating margin still widened to 33.9% from 30.0%, helped by $200m of buybacks at an average $50.44 that cut the diluted share count to 77.1m from 84.2m. "The benefit of lower share count is being partially offset by higher interest expense following the drawdown of our $75 million VFN," chief financial officer Sudhanshu Priyadarshi told investors on 6 August. The remaining $250m of authorization competes with a dated wall: the 2022-1 Class A-2-I notes carry an anticipated repayment date of December 2026.

Attrition ran at 3.5%, the middle of its historical range, with no gap between card tiers. "We are concentrating our efforts this year on two priorities that are central to reigniting net member growth, driving acquisition and reinforcing affordability," chief executive Colleen Keating said on the same call. The problem is joins, not leavers.

The clubs Life Time sells and leases back

Life Time grew second-quarter revenue 13.7% to $866.0m, lifted operating margin by three percentage points to 17.3%, and cut net debt to 1.4 times EBITDA from 1.8 a year ago and 9.0 in mid-2022. Comparable-centre revenue rose 9.1%, with in-club spending — dynamic personal training, the spa — contributing as much as list pricing. Against a US gym industry growing about 1.3% this year, that is share and price gain rather than a rising tide.

The cash tells a second story. Capital expenditure was $263.3m in the quarter against $200.2m of sale-leaseback proceeds, with the full-year target raised to roughly $400m from $300m; first-half free cash flow of $85.3m is inseparable from those sales, and on a trailing twelve-month basis the free cash flow yield is still negative. "We have tremendous amount of real estate deals in the pipeline that I am more excited than I've ever been," founder and chief executive Bahram Akradi said on 30 July.

What each price is paying for

Planet Fitness trades at 10.8 times trailing enterprise value to EBITDA and 16.9 times trailing earnings, with a 6.6% trailing free cash flow yield. Life Time trades at 14.3 times and 22.1 times; its forward price/earnings of 26.0 sits above trailing because consensus 2026 net income of $361m is below the $415m earned in the last four quarters — a modelling artifact rather than a growth discount.

Both companies' moving averages converged into neutral territory on 28 July from opposite directions, Planet Fitness rising out of a months-long downtrend and Life Time falling out of an uptrend. Life Time's slip over the past thirty days looks like supply rather than operations: Leonard Green affiliates sold just over 5m shares at $43.16 on 10 August, cutting the group below 5% ownership.

The de-rating at Planet Fitness is earned: a company whose comps now run at industry rate deserves less than one compounding at nine percent. What the discount does not obviously price is that the cheaper company is the one throwing off cash, while the dearer one is funding its largest club class yet by selling the buildings into a market that has so far bid for them. The split is the K-shaped consumer written into two income statements — the $245 member absorbs increases, the $15 member is not joining — and it leaves each company hostage to a different counterparty: Life Time to the sale-leaseback bid, Planet Fitness to the December refinancing and to franchisee nerve.

Planet Fitness still says the United States will one day hold 5,000 of its clubs. Almost every one of them is somebody else's decision to build.

Planet Labs' best quarter came out of a delivery bay. The company, which flies roughly 150 small Earth-imaging satellites and sells the resulting daily picture of the planet as data subscriptions to governments, farmers, insurers and mapmakers, handed the first of its new Pelican satellites to the Swedish Armed Forces in the quarter ended 31 July. That handover pulled about $30m of one-time revenue into the period from the quarter where it had been expected. Revenue of $116.1m, up 58%, was a record; one-time delivery revenue was 12% of it, against 1% a year earlier.

Composition is the whole story in an industry the market still prices as though it were paid per picture. Planet's subscription revenue is annual contract value recognized evenly across multi-year terms, so any quarter's imagery line is a backlog-conversion schedule rather than a reading on demand. Growth has to arrive from new agreements — or, increasingly, from selling the satellite itself. Planet has guided the October quarter to $101–105m, a sequential decline and about 27% year-on-year growth, because a delivery cannot be made twice.

The meters that sit outside the revenue line

Backlog ended the quarter at roughly $815m, up about 11%, with remaining performance obligations of $753m, up 9%. Two quarters earlier the backlog figure was near $900m and growing 79%. Existing customers are expanding spend slowly: net dollar retention on annual contract value was 109%, 110% including winbacks. And the growth sits where the deliveries are — defense and intelligence revenue rose 90% while civil government grew 5%, and Europe, the Middle East and Africa grew 130% on Swedish and German satellite-services work.

The franchise underneath is not in question. "No one has a sufficient number of Earth imaging satellites in the right kind of plane and all that to do a daily scan," chief executive Will Marshall told investors on the September 3 call. "And so if you want to monitor for new threats and monitor things consistently, we're the only game in town." That uniqueness won a sole-source $8m National Geospatial-Intelligence Agency monitoring contract in the quarter, alongside a five-year German government satellite-services deal and a first national program with the Rwanda Space Agency.

What it does not do is make the margin a pricing story. Non-GAAP gross margin fell from 61% to 59% on satellite-services mix, and the full-year guide is 55–57%. Capital spending was raised to $100–115m against guided revenue of $430–441m, roughly a quarter of the top line, spent rebuilding a fleet with a short useful life. Trailing free-cash-flow yield is 0.42%. The $865m of cash is real; $120m of it came from an at-the-market offering priced at $31.96 a share, 92% above the 10 September close.

The same machine, one-sixth the size

BlackSky Technology, which sells minutes-from-tasking imagery and analytics and separately builds, integrates and operates satellite and ground systems, reported June-quarter revenue of $33.3m, up 50%. Its space-based intelligence and analytics subscriptions were $24.5m, up 36%, leaving a program remainder that roughly doubled to about $8.8m. The quarter before, revenue fell 29.7%. Full-year 2025 revenue grew 4.4% and gross profit shrank. International contracts are now over 80% of funded backlog, US government subscription revenue was flat, and the company said visibility into the 2027 US budget is unclear. Adjusted earnings before interest, taxes, depreciation and amortization were positive at $4.7m.

The share prices moved as a group, not as verdicts on either business. Planet, BlackSky, Spire Global and Satellogic all peaked within three days of each other in late May and are down 67.5%, 59.8%, 55.5% and 53.2% since, as the 30-year Treasury yield hit a 19-year high above 5.33% — the kind of move that lands hardest on long-duration, cash-burning small caps. Planet now trades at 12.7x forward sales against a trailing 39.4x in early May, and is still the most expensive of the four on forward price-to-gross-profit at 22.9x, against BlackSky's 14.9x.

So the reported income statements do not justify the fall: Planet's operating margin improved from −24.5% to −11.6% while the stock halved. The forward meters partly do. Backlog growth in the teens, retention under 110% and a guided sequential decline are what a business looks like when a handover, rather than demand, made the quarter. BlackSky's headline is evidence for neither case until its subscription line, not its program line, carries the growth.

Behind both sits one budget. The National Reconnaissance Office spends about $400m a year on commercial imagery, and the White House budget office has already proposed cutting it; BlackSky's last disclosed extension of that vehicle covered Gen-2 services only into mid-2026. Selling satellites to Stockholm and Berlin has become the growth story precisely because Washington stopped being one.

Chemed's hospice arm is paid a flat Medicare rate for each day a patient is on service, whatever that day's care actually costs. In the June quarter its patients stayed less time — an average of 101.2 days against 137.1 a year earlier — and the accrual Chemed books against the aggregate per-beneficiary cap, Medicare's ceiling on average payment per patient, fell to $0.5m from $16.4m.

That swing is most of the reason operating income rose 31% on revenue up 8.8%, and it explains nothing whatsoever about the companies filed beside it. Four businesses still trade under the home health and hospice heading, and each is paid by a different machine: a hospice per-diem, state Medicaid personal-care rates, a California pediatric nursing rate, and a spread over the acquisition cost of infusion drugs. CMS's proposed calendar-2027 home health payment rule, issued 1 July with an estimated 2.4% aggregate increase, lifted all of them in late July. They have since gone separate ways: over the past month Aveanna Healthcare rose 52.7% while Option Care Health fell 2.7%, Chemed 5.2% and Addus HomeCare 1.1%. Over three months the spread runs from Aveanna's 99% to Option Care's 13.7%.

Chemed: shorter stays, and a plumbing arm Google is squeezing

VITAS, Chemed's hospice and palliative care business, admitted 19,125 patients in the quarter, up 9%, and carried an average daily census of 23,687, up 6.1%. Adjusted earnings of $6.06 a share beat consensus of $5.60, and management raised full-year census growth guidance to 5.75-6.25% and consolidated adjusted earnings to $25.00-25.75 a share, a midpoint 7.8% above last year. "VITAS has never been in a better position to take advantage of growth opportunities," VITAS president and chief executive Joel L. Wherley told investors on the 29 July call. "We have put the difficulties of 2025 behind us."

The buyback is not doing the work. Reported diluted earnings per share rose 43.7% to $5.13, but diluted shares fell 8.1% — roughly eight points of the gain — leaving the remainder to the 29% rise in net income.

Roto-Rooter, the plumbing and drain-cleaning franchise bolted to the hospice business, is the drag, and its problem is a search engine. Adjusted earnings before interest, taxes, depreciation and amortization were flat at $48.5m with margin down 77 basis points; total sales leads fell 1.6% and free leads fell 13.1%. "Google hates the idea of free leads," chief executive Kevin J. McNamara said on the same call. "They have systematically tried to drive their users away from the free aspects of service providers."

Option Care: paid on the spread, not the patient

Option Care Health, the largest national provider of home and alternate-site infusion therapy, delivers immunoglobulin, nutrition and chronic inflammatory therapies in patients' homes with nurses attached. Its economics are a distributor's: gross margin fell from 22.8% in 2023 to 18.1% last year, and 2025 revenue grew 13.0% to $5.65bn while gross profit grew 0.7%. Biosimilar conversion in the chronic inflammatory portfolio repriced the drug while volumes kept rising.

June quarter revenue rose 1.9% to $1.442bn and gross profit 5.7% to $267.3m, margin recovering to 18.53% from 17.85% — the first quarter in five in which profit dollars outgrew sales. "Stelara and related biosimilars will represent less than 1% of 2026 company net revenue and gross profit," chief financial officer Meenal Anil Sethna said on 29 July; the chronic inflammatory headwind still embedded in guidance is about $55m of gross profit. The May guidance cut that took 26% off the shares in a session has not been recovered, and the shares are the only ones of the four that have not re-established an uptrend, the 50-day and 200-day averages tangled since late July. "We are not satisfied with our performance," chief executive John Charles Rademacher told investors.

Where they trade

Gross margins differ too widely here for sales multiples to compare, so price against gross profit does the work: Option Care at 3.42x trailing is the cheapest of the four and Chemed at 8.51x the dearest, with Addus at 4.63x and Aveanna at 3.63x. Option Care fetches 12.6x forward earnings against 17.8x trailing and yields 9.8% on trailing free cash flow; Chemed 20.2x forward against 25.8x trailing, still short of the roughly 32.7x its 2024 high implied. Addus, whose personal-care revenue comes mostly from state Medicaid programs, grew 8.0% in the quarter after running near 25% through 2025 and trades at 16.8x forward. Aveanna's revenue rose 13.7% but gross margin fell to 32.6% from 35.8%; its 13 August guidance raise and California's first pediatric private-duty-nursing rate increase since July 2018, effective 1 January 2027, are what moved the stock.

What the businesses earn

Chemed's quarter is operating, not financial engineering — but the cap cushion came from a mix shift toward shorter stays, and mix reverses. Option Care's reacceleration is one quarter old and priced against a top line consensus has growing 1.9% this year; the cheapness is real and so is the flatness behind it. What July's rally assumed — that one federal rule moves this group — has not survived August. Anyone holding the four is holding four payers.

CMS finalized the 2027 hospice rate at a 2.3% update, trimmed from the 2.4% it proposed in April. VITAS's new per-diem takes effect on 1 October; in Florida, where the cap cushion it just rebuilt sits, management says the increase is 1.0%.

Everspin Technologies makes magnetoresistive memory — chips that keep their contents when the power is cut — and the fastest-growing line in its June quarter was not a chip at all. Non-product revenue reached $3.4m against $2.1m a year earlier, the first partial-quarter recognition of a $40m, 30-month subcontract to supply Toggle MRAM to a US prime contractor. Product sales still carried the quarter, at $15.3m, up 38% year over year, taking the company total to a record $18.7m, up 42%.

That record arrived into a share price that has fallen 30% over three months and sits 64% below its 12 May high of $44.01. For an 85-person company in Chandler, Arizona with a $374m market value, the question is whether the decline marks something breaking in a business selling into long design-in cycles, or the deflation of a spike. On the evidence, it is the second — and that matters because the stock is still not priced as a broken business.

What is actually growing

Revenue growth has accelerated for three straight quarters — 11.8% year over year in the December quarter, 13.2% in March, 42% in June — after sales fell 21% in full-year 2024. The June quarter's product strength came from industrial automation, helped by a recovery in Japan, from European energy management, and from aerospace and defense, including a first geostationary satellite win: Astro Digital selected the 64-megabit PERSYST magnetic memory for boot and telemetry on its Raven Bus.

"Driven by strong product revenue coupled with initial non-product revenue under our recently signed $40 million contract with a US prime contractor," chief executive Sanjeev Aggarwal said of the quarter on the 5 August call. Management cautioned that recognition on that contract follows a bell curve rather than a straight line. September-quarter revenue is guided to $19.5-20.5m, roughly 40% above the $14.06m of a year earlier.

The defense channel widened again on 2 September, when Everspin and Teledyne HiRel Semiconductors announced a partnership to push 64-, 128- and 256-megabit PERSYST parts into avionics, radar, electronic-warfare payloads and satellite electronics, with customer availability anticipated in the fourth quarter.

What is not growing is the thing the company's name is most often attached to. Its Compute Express Link work — the memory-pooling standard — is a proof-of-concept demonstration pairing externally developed controller intellectual property with an AMD field-programmable gate array, plus a memorandum of understanding with MaxLinear. It earns nothing today.

The loss is a legal bill

Operating expenses rose to $14.5m from $8.7m, including $4.0m of litigation and $1.1m of non-recurring engineering. Avalanche Technology sued in January over four patents on magnetic materials and filed a parallel complaint at the International Trade Commission; the claims target the spin-transfer-torque line only, not Toggle memory or sensors, and the import-ban proceeding is targeted for completion on 6 July 2027. Guidance bakes in about another $4m of it this quarter. Everspin ended June with $43.9m of cash, no debt, and $0.2m of operating cash flow — thin, but funded.

The reported gross margin of 53.9%, up from 51.3%, is mix rather than pricing power: management guided product gross margins down to the mid-to-upper 40s on packaging and test cost increases at outsourced assemblers and higher gold prices, with the higher-margin contract revenue holding the company total above 50%.

What the price knows

On 12 May, 14.0m shares changed hands at the peak; recent volume averages 0.86m a day. A year ago the stock was $6.80 on a similar revenue base, which means the sales multiple has roughly doubled even after the fall. At 5.98x trailing and 5.01x forward sales — earnings multiples are meaningless against a trailing loss — the shares carry about 4.3x 2027 consensus revenue of $86.2m, and 3.7x the company's own target of more than $100m by fiscal 2029. Thursday's 4.7% drop came alongside a broad memory selloff, with Western Digital down 7.0% and Micron 4.3%, against a 0.5% decline for the S&P 500.

So the business earns none of the de-rating and the valuation excuses most of it. Accelerating revenue, a widening defense book and a funded balance sheet are not the profile of a company losing its market; what has compressed is the speculative premium a May squeeze installed. The risks that remain are specific and dated: a legal bill that has to stop growing, and a product margin management has already guided lower.

Everspin's stated path past $100m of revenue runs through a 256-megabit part entering production in mid-2027 and a foundry partner, Taiwan Semiconductor, expected to join at the end of this year — neither of which has produced a shipped chip yet.

Icahn Enterprises' investment funds spent the June quarter positioned against the business that dominates its own income statement. The funds finished June net short by the equivalent of 30% of notional exposure once refining hedges are counted, and net long 23% once those hedges are stripped out — the short book is specifically a hedge against refining, not a general view on equities. The refining exposure being hedged is CVR Energy, the coking refiner at Coffeyville, Kansas and Wynnewood, Oklahoma that also makes nitrogen fertilizer, and of which Icahn Enterprises owns 66.3% after a January 2025 tender at $18.25 a share drew almost no takers.

So when US product margins went to records this summer, the parent did not collect. CVR closed at $49.08 on 10 September, up 52.4% in a month; Icahn Enterprises' units fell over the same stretch, though most of the visible drop was the $0.50 distribution that went ex on 17 August — adjusted for it, the month's move is roughly -1.8%. The stake in CVR is now worth about $3.2bn against a $4.62bn market value for the whole holding company. Whether the rest of the structure adds or subtracts is the question, and the answer is disclosed quarterly.

What the refinery actually earned

CVR's June quarter produced $209m of adjusted EBITDA and $0.34 of adjusted earnings per share, against a reported loss of three cents once a $73m unfavorable mark on its renewable-fuel obligation and $81m of realized crack-spread hedging losses are counted. Petroleum segment adjusted EBITDA rose 179% to $106m, on Group 3 two-to-one-to-one crack spreads of $44.91 a barrel against $24.02 a year earlier, with throughput of 213,000 barrels a day at 98% utilization. Fertilizer adjusted EBITDA rose 60% to $107m — slightly more than refining — on a spread between cheap US natural gas and scarce global nitrogen, with CVR Partners' realized ammonia gate price up 33% to $791 a ton while American producers pay under $3 per million British thermal units for gas against roughly $19 in Europe.

The drag on refining is compliance. CVR booked $216m of Renewable Identification Number expense in the quarter, $11.16 a barrel, cutting its capture of the posted margin by about a quarter. "RIN prices are adding nearly $0.40 a gallon to the price of gasoline in the U.S., and RFS compliance costs are more than twice all the other combined operating costs for many refineries," chief executive Dane Neumann told investors on the July 30 call. Wynnewood's pending 100% small-refinery exemption would lift consolidated capture by roughly nine percent; the precedent is the August 2025 decision that cut CVR's 2020-2024 obligation by more than 424 million RINs, booked as a roughly $488m gain. On August 31 the Environmental Protection Agency decided 34 petitions for the 2025 compliance year, granting 18 full exemptions and pushing the compliance deadline to October 1 — a dated regulatory event sitting inside CVR's month.

The margin itself has a cause outside anyone's control: roughly 40% of Russia's refining capacity is offline after drone strikes, a shock to product-making rather than to crude. The forward curve treats it as temporary — the December 2027 strip sits more than 35% below the September crack.

The parent's meters are not sales

Icahn Enterprises' consolidated revenue rose 27.4% last quarter, and it is almost entirely CVR's fuel and fertilizer sales; the line tells a unitholder nothing. The honest meter is the indicative net asset value the company publishes, $2.602bn at June 30, down $765m in three months. The breakdown: $435m off the CVR position and $243m off the funds on broad-market hedge losses, offset by $97m for the pending $700m Pep Boys sale.

Against 669m diluted units — up 22.75% in a year — that NAV is about $3.89 a unit, against $6.835 on September 10, a premium near 76%, corroborated by 2.54 times book on a balance sheet largely marked to market. The board declared the same $0.50 quarterly distribution on August 3, about $335m on that count, in a quarter when adjusted EBITDA attributable to Icahn Enterprises was minus $134m. What makes it survivable is ownership: Carl Icahn and affiliates held roughly 87% of the units at June 30 and take the payout largely in more units, so the cash out the door is far below the declared figure — and the unit count grows regardless. Energy contributed $102m of attributable EBITDA last quarter; CVR's ten-cent dividend sent about $7m of actual cash upstairs.

The verdict

CVR's advance is largely earned. At eight times trailing enterprise value to EBITDA and a 7.1% trailing free cash flow yield, with $737m of cash, it remains the laggard of a peer group that re-rated 114% to 175% over twelve months, PBF Energy at the top and CVR up 53.8%. What the business has not established is duration: trailing EBITDA is flattered by last year's one-off RIN gain, the forward crack curve is far lower, and the sell side carries four sell ratings, no buys and a $31.40 average target, 36% below spot.

Icahn Enterprises is the opposite case. Its moving-average trend rolled over in June while CVR made a new 52-week high, and the de-rating is doing rational work rather than opening a gap — the units sit well above the company's own published value, after five consecutive years of net losses. The structural point is the one to carry forward: the funds' hedge deliberately neutralizes much of the refining windfall in both directions, so a crack reversal that hurts the CVR mark should help the short book, and a further rally will keep doing what it did last quarter.

The EPA remains delinquent on Wynnewood's 2020 exemption petition, with the 2025 compliance deadline now pushed to October 1. A grant would widen the refinery's capture rate; whether any of it reaches Icahn Enterprises' unitholders depends on a hedge book disclosed once a quarter, in aggregate.

SailPoint reported its July quarter on 9 September, and the contracted book it steers by grew roughly eight points faster than the revenue on its income statement. The identity-governance company — Austin-founded, run by co-founder Mark McClain, selling software that decides which employees, contractors and machine accounts may reach which applications and data — put annual recurring revenue at $1.231bn, up 25%, against reported revenue of $308.8m, up 17%. It raised its full-year recurring-revenue target to $1.38bn. The shares closed lower that day.

The gap is not noise. SailPoint is still converting a book sold as up-front term licenses into cloud subscriptions recognized ratably, and management quantified the toll: every $5m of mix moved from term license to subscription costs roughly $10m of recognized revenue, about a $5m headwind in the quarter. Cloud mix is guided at 90-95% of net new recurring revenue for the fiscal year. Until that conversion finishes, reported revenue is a lagging, structurally understated read on what the company sold.

What the book says

Cloud recurring revenue reached $847m, up 36%, and accounted for 97% of net new recurring revenue against a guided 90-95%. Dollar-based net revenue retention was 113%. Remaining performance obligations rose 30% to $1.9bn and the portion due within a year 27% to $931m — a book growing faster than the revenue it will become, which management attributed to larger deals rather than longer contract terms. Average recurring revenue per cloud customer passed $400,000, up 17%.

The demand story management tells is agent governance. Recurring revenue tied to artificial-intelligence products passed $70m, ahead of a $100m full-year target and more than 30% of net new recurring revenue, with customers that adopt those products spending over 60% more. "The EU AI Act contains specific human oversight and audit logging requirements, which are scheduled to become enforceable as early as 2027," McClain told investors on 9 September. "We believe regulators will increasingly ask who is accountable for this agent and its actions."

What it costs

The blemish is in the cost line. Gross margin fell to 59.0% from 67.3% a year earlier, so gross profit grew 2.4% on 16.8% revenue growth — the disclosure does not explain the drop. The GAAP operating loss widened to $54.7m from $40.8m. Adjusted operating margin was 20.3% in the quarter but is guided down to 17.7% for the current one, carrying costs from the Entro Security acquisition, which extended discovery to more than 1,200 types of non-human identity. Free cash flow was $37m, with about $200m expected for the year.

The comparison doing the pricing

Okta, the independent single-sign-on and access-management vendor that competes with SailPoint in governance, is the reason this matters now. Its July quarter grew 10.6% — a third straight deceleration — and it guided current-quarter bookings growth back to 11-12% from 14%. Yet Okta trades at 11.84x trailing gross profit, against 5.97x in early May, and 43.6x forward earnings while consensus carries sub-11% revenue growth for two years. SailPoint, at a $10.0bn market value, is about 8.1x its recurring revenue base growing 25%, with no forward earnings multiple to quote because it loses money on a GAAP basis. Sell-side targets straddle the price: Cantor Fitzgerald at $25, Evercore ISI at $22, Mizuho at $19 and Neutral.

So the market is paying a premium for operating leverage it can see on Okta's income statement and discounting growth it cannot see on SailPoint's. Some of that discount is earned — the gross-margin fall is real, the losses are widening, and Thoma Bravo held roughly 86% of the stock after the February 2025 relisting, a float that can expand at the sponsor's choosing and has no dated trigger. But the part of the discount attributable to a suppressed revenue line is an accounting artifact with a stated exchange rate, and it runs out.

When cloud mix stops rising, reported revenue converges on the book and the two numbers say the same thing. At that point the argument for the discount has to be the control stake, or nothing.

Freshpet spent a decade building refrigerated kitchens before it had the demand to fill them. This year the company budgeted $150m of capital spending and is deliberately spending less, because the production lines it already installed are running better than planned and management wants to optimize a new bagging technology before committing to more. Free cash flow was $14.7m in the June quarter against $0.5m a year earlier, the second consecutive positive quarter after $12.7m in the March quarter.

That matters because Freshpet's unit of business is not a shelf. The company makes refrigerated fresh dog and cat food and sells it out of branded coolers it owns and installs inside other retailers' stores — 30,721 stores as of the June quarter, of which about a quarter now carry two or three fridges, with total distribution points up 13% year over year. Growth is household penetration multiplied by how much each household spends. Margin is a capacity-utilization number, which is why the depreciation on those company-owned kitchens sits inside cost of goods sold.

The gap between two gross margins

The June quarter shows the arithmetic plainly: adjusted gross margin excluding depreciation reached 48.6%, up 1.7 points, against a reported gross margin of 42.1%. The roughly six-and-a-half-point difference is the build-ahead. Net sales rose 15.5% to $305.6m, essentially all of it volume, and guidance went up twice — sales growth to 10–12% and adjusted earnings before interest, taxes, depreciation and amortization to $210–220m.

"We now have 3 lines utilizing our new bag product technology 2 in Bethlehem and 1 in Ennis, and we are encouraged by the improvement in quality, throughput yield and unit economics," chief executive Billy Cyr told investors on the August 5 call. "At fully optimized performance, we expect over 100 basis points of gross margin improvement on the entire business from the lines we have already installed."

Where the model is straining

Two things cut the other way. Household penetration reached 15.9 million households, up 5% year over year — down from 8% growth the prior quarter — while buy rate rose 7% to $117 and the heaviest users, who spend about five times the average, now account for 71% of sales. Growth is migrating from new households to existing ones. And logistics ran at 6.9% of sales against 5.7%, roughly $8m more than originally guided, on fuel and trucking costs.

The category is no help. On September 9, Chewy's management said the broader pet consumables market was roughly flat year over year, with de-premiumization visible in treats and toppers; its shares fell about 9% that session. Meanwhile Mars, General Mills, Nestlé Purina and Colgate-Palmolive's Hill's have all entered fresh formats — refrigerated and frozen dog food grew 17.8% to $1.6bn in the year to August 2025, the only pet food category growing.

The shares have moved in jumps, not drifts. Freshpet fell 38% between March and June, on a Bank of America downgrade to Neutral with a $60 target, the Hill's entry and removal from Russell growth benchmarks, then rose 30% — including 15.1% on the August 5 print. Two years of de-rating remain: the stock is half its September 2024 level while trailing revenue is 21% above the 2024 full year, compressing the sales multiple from roughly 7x to 2.7x. At 13.3x trailing enterprise value to EBITDA the shares are not expensive against their own history; at 33.7x forward earnings they are not cheap in absolute terms.

The mirror

Central Garden & Pet, filed under the same packaged-foods classification, runs pet supplies alongside a weather-driven lawn-and-garden half and demonstrates the opposite engine. Fiscal third-quarter net sales fell 8% to $882m, organic sales rose 2% after exiting pet distribution, gross margin expanded 1.3 points to 35.9% — and earnings per share fell on both reported and adjusted measures, even as full-year adjusted guidance was raised to "$2.85 or better." The diluted share count is down 3.6%, which offset roughly three-quarters of the decline in net income. Revenue has now fallen four years running, and consensus models further declines in each of the next two. At 7.4x trailing EBITDA with a trailing free-cash-flow yield near 14%, it is priced as what it is: a cash machine that shrinks.

The verdict divides cleanly. Freshpet's operating inflection is earned — accelerating volume, a widening margin as the kitchens fill, capital spending falling below budget rather than above it. What nothing in the June quarter explains is how the company keeps compounding once penetration growth halves and the flat category stops handing it new households. Central's improvement is arithmetic performed on a declining base, and the arithmetic has a floor.

Freshpet's next print carries a comparison headwind of more than two points from a large club customer's prior-year expansion. It is the first quarter in which the fridge network has to grow through the calendar rather than with it.

Navitas Semiconductor — a 190-person fabless designer of gallium-nitride and silicon-carbide power chips in Torrance, California — spent the summer telling investors when the architecture it is built for arrives. On its 27 July call, management dated gallium-nitride content in graphics-processor voltage conversion to mid-to-late 2027, accelerating into 2028. At Citi's Global TMT conference on 8 September, the same step — native 800-volt systems, with direct-current conversion moved inside the compute tray — was dated to early 2028.

Six weeks, roughly two quarters of slippage on the one line item the equity is priced against. Navitas's June quarter brought in $10.5m of revenue, down 27.3% year on year, at a reported gross margin of minus 9.5% and an operating loss of $27.2m. Four quarters through June produced roughly minus $0.6m of cumulative gross profit, so there is no price-to-earnings or price-to-gross-profit figure to compute; the shares change hands at 79.7 times trailing sales. What the company does have is $557m in cash and no debt, after raising $373m during the quarter at $21.89 a share — 95% above Thursday's $11.26 close. Diluted share count is up 18.5% in a year. Until the racks ship, this is a claim on runway.

"It would have happened without NVIDIA, it's happening much sooner due to their influence," chief executive Chris Allexandre said of the 800-volt shift at the Citi conference, where he also called the pivot away from mobile and consumer chargers "pretty much done." Nvidia's own published schedule puts 800-volt direct-current architecture into full-scale production alongside its Kyber rack systems in 2027, and names ten silicon providers for it — Analog Devices, Infineon, Innoscience, MPS, Navitas, onsemi, Renesas, ROHM, STMicroelectronics and Texas Instruments. A place on a list is not a socket.

The supplier that already gets paid

Monolithic Power Systems, which designs the voltage-conversion chips and modules that sit under server processors and accelerators, is on that same list and is not waiting for it. June-quarter revenue was $980.6m, up 47.6% — the fourth straight quarter of accelerating growth. Enterprise data revenue reached $380.6m, up 164.3%, and the company lifted its full-year growth floor for that segment from 85% to 130%. Gross margin held at 55.2%, inside the band it has occupied for eight quarters, and operating margin widened to 31.0% from 24.8%. Founder and chief executive Michael Hsing told investors on the 30 July call that the company "relies on its own silicon carbide devices for the 800-volt solution" — its answer to the transition is internal, and sampling rather than shipping.

The shares have fallen 29.5% from their 3 June high. Measured against gross profit, Monolithic Power now trades at 32.3 times trailing and 25.4 times forward, against 49.8 times trailing in May. That forward figure is level with Vicor — the modular converter maker that lost Nvidia's H100 power socket to Monolithic years ago and now collects royalties on the patents — whose revenue grew 1.6% last quarter.

What the selling was actually sorting for

Over thirty days the power and analog complex fell in order of valuation, not end market: Alpha & Omega Semiconductor down 29.1%, Power Integrations 21.4%, Navitas 16.0%, Monolithic Power 15.0%, against Texas Instruments down 7.0% and Analog Devices 6.5%. Memory went the other way — Micron up 14.5%, SanDisk 33.3% — as capital moved toward scarcity that exists today. The backdrop is the same discount-rate repricing the market has been trading all month, with the 10-year Treasury yield near 4.90%. Nvidia itself has now fallen 5.4% in four sessions, which removes the tidy reading that the customer was rising while its suppliers sank.

So: Navitas's de-rating its own numbers explain, and its own calendar just reinforced. Monolithic Power's they do not — a business compounding at 47.6% has been marked to the price of one growing at 1.6%. The threat to Monolithic that would justify it is competitive rather than cyclical, and it is visible: Analog Devices paid $1.5bn for Empower Semiconductor, whose vertical power regulators sit in exactly the position under the accelerator that Monolithic is priced to own.

Nvidia's 800-volt racks reach full production next year on Nvidia's schedule. Navitas now expects its in-tray silicon a year after that. Every quarter that gap widens is a quarter in which the cash pile, not the design wins, does the work.

Teleflex spent 2025 preparing to split itself into two listed companies. It sold three businesses instead, agreeing in December to $2.03bn of cash deals — the OEM unit to Montagu and Kohlberg, Acute Care and Interventional Urology to Britain's Intersurgical. The OEM sale closed on 3 August for $1.5bn; the Intersurgical piece is still pending. There is no stub to value, only a smaller company.

That smaller company told investors three days later that it would grow more slowly than promised and earn more than promised. Full-year constant-currency revenue growth guidance came down to 3.5-4.5% from 4.5-5.5%. Adjusted earnings guidance went up, to $6.90-$7.20 a share from $6.25-$6.55. The raise was sourced from a $700m debt paydown, a $250m accelerated repurchase launched on 7 August, net interest expense cut to roughly $85m, and a tax rate trimmed to about 12.25%. None of it came from demand.

The hole is in the business it bought

Teleflex sells single-use critical-care and surgical devices — the Arrow vascular catheter family, the MANTA closure device, the UroLift system for enlarged prostates. It bought BIOTRONIK's vascular-intervention business to add scale in interventional cardiology, and that is the segment now going backwards: Interventional revenue fell 1% to $211.9m in the June quarter, against Vascular up 8% and Surgical up 9.2%. The integration that was to finish by mid-2026 now runs to the end of it, on order-to-cash system transitions that left customers confused, distributors working down inventory, and a repositioned sales force with roughly six-month onboarding. It is "not a product issue," chief executive Jason Weidman told investors on the August 6 call; the portfolios "fit beautifully together."

Margins moved with it. Gross margin fell 280 basis points to 61.7% on tariffs and the lower-margin acquired business, and adjusted operating margin fell 520 basis points to 19.6%. Full-year operating margin is guided to about 19% against the 23% steady state management says the post-divestiture company should reach. Part of that gap is chosen — research spending was lifted to 7.9% of sales from a historical 6% — and part is the cost of a company that has shed roughly a third of its revenue and not yet shed the overhead behind it.

What the same end market is doing

Demand is not the problem. Merit Medical, which sells disposable vascular-access, angiography and hemostasis devices into the same hospital departments, grew 9% organically in constant currency, its best quarter in three years, expanded gross margin by 314 basis points and raised full-year organic guidance to 6.9-7.5%. It trades at 20.1x forward earnings, the richest of the four, and is the only one down over twelve months, by 2.6%.

Becton Dickinson is the counter-case for restructuring done. Its Biosciences and Diagnostic Solutions business went to Waters in a deal that closed on 9 February, taking China down to 4% of revenue; every one of the four segments left grew in the June quarter, organic growth was 4.4%, and earnings guidance was raised to $12.62-$12.72. At 13.9x forward it is the cheapest of the group, up from roughly 11.7x in June. ICU Medical, which makes infusion pumps, needle-free connectors and IV solutions, grew reported revenue 0.5% but widened gross margin from 37.9% to 42.6% and raised guidance twice over; it sits at 18.0x.

All four are absorbing tariffs — 110 basis points of margin at BD, $0.21 a share at Merit, $30m-$40m a year at ICU Medical — and all four sell the exact products named in the unresolved Section 232 national-security investigation into syringes, catheters and IV bags.

The verdict

Teleflex's Vascular and Surgical numbers are earned; the share price is not yet. Its forward multiple has climbed from roughly 15.5x in March to 18.3x now on an unchanged consensus of $7.20 — a re-rating delivered across the same six months in which the company cut growth guidance and pushed out an integration. Measured against gross profit, the shares cost more on next year's figures than on last year's, because the profit base is shrinking faster than the price. What is being paid for is a consensus 2027 earnings rebound of 52% that management has framed only as a "meaningful step-up," and the standing possibility, pressed publicly by Irenic Capital, that the whole company is sold.

Weidman is two months into the job, still running a comprehensive review, and has promised more direction at the next call. Until it arrives, the most consequential thing Teleflex did for holders this year was write checks with the proceeds of businesses it no longer owns.

SolarEdge gathered analysts on 10 September to describe a business it does not yet have a customer for. The Israeli maker of DC-optimized inverters and power optimizers for rooftop solar told investors it targets $2.4bn of revenue by 2029, at a 35% gross margin and $360m of operating profit, and said the medium-voltage-to-800-volt conversion stage of its AI data-center powertrain is now running under load. It also published a white paper with Nvidia on 800-volt direct-current protection and grounding — and acknowledged it holds no purchase order or agreement with a hyperscaler or neocloud operator. The stock fell during the presentation.

That is the second data-center announcement from this pair in three days. On 8 September Enphase, which sells module-level microinverters and AC batteries into solar distributors and installers, said its solid-state transformer modules had entered production in Texas. Both roadmaps land in the same place: a working system late in 2026, pilots in 2027, volume shipments in 2028. Enphase's own credit treatment for the product is still being finalized. Whatever either company earns from AI power, it earns after two more years of the business it actually has.

The business it actually has

That business has split in two. SolarEdge's June quarter brought revenue of $346.2m, up 19.6% year on year, with gross margin of 27.5% against 11.1% a year earlier and a first non-GAAP operating profit — $10.2m — in nearly three years, plus $3.1m of free cash flow against $601.6m of cash and securities. Europe carried it, rising 36% sequentially to $154.4m; US residential slipped 2% on slow tax-equity funding. The company is still GAAP loss-making, at -$30.8m, and $13.3m of tariff refunds flattered the margin.

Enphase went the other way. Revenue of $291.9m was down 19.6%, the third consecutive drop of roughly a fifth, and US sell-through — systems actually installed — fell 34% year on year. Consensus has revenue falling 19.1% this year and earnings per share 28.8%, and does not restore Enphase to its 2025 revenue of $1.47bn until about 2029. The June quarter included $84.3m of safe-harbour shipments to third-party-ownership partners, "some of whom are supported by strong balance sheets," chief executive Badri Kothandaraman said on the 28 July call, where he also guided to modest under-shipment in the third quarter to work down microinverter channel inventory.

The hole is statutory. The residential 25D tax credit expired at the end of 2025, leaving third-party-owned leases claiming 48E — where the developer picks the hardware — as the federal route for homeowners, and SEIA and Wood Mackenzie see US residential installations falling roughly 18-21% this year. US installations were already down 27% in the first quarter. From 4 December, a Section 232 proclamation sets minimum import prices of $0.38 a watt on modules, raising the cost of the systems inverters attach to.

What the market did with it

Both stocks roughly doubled between 12 May and 2 June on the data-center narrative — Enphase up 93%, SolarEdge 94% — then surrendered it, down 48.6% and 54.3% from those peaks. SolarEdge lost 28.1% in the 5 August session after guiding third-quarter revenue to $310-340m against consensus near $370m. Every distributed-solar name fell over three months: Fluence 54.7%, Sunrun 27.4%, First Solar 16.7%.

One de-rating is earned and one is not. Enphase's price-to-gross-profit multiple has halved from 14.55x in May to 7.81x, and its forward multiple of 8.74x sits above the trailing one — arithmetic that only works if gross profit shrinks, which is exactly what consensus models. SolarEdge fell further, to 7.57x from 16.09x in May, with the forward figure level at 7.64x, meaning no growth is priced into a company whose margin has expanded for six quarters. Against that sits a guided sequential revenue decline, a GAAP loss, and $337m of 2.25% convertible notes due July 2029.

So two companies whose revenue lines point in opposite directions now cost almost exactly the same per dollar of gross profit. The market has stopped distinguishing between them, and both managements have answered by pointing at 2028. SolarEdge's chief executive, Shuki Nir, framed the 800-volt transition as a question of whether such systems can be trusted at scale — a fair engineering answer to an investor question that was about revenue timing.

The next event either company controls is not a lab result. It is a purchase order with a date on it.