HealthEquity enters fiscal 2027 holding the largest independent balance sheet in health savings accounts, with account assets closing in on the 38 billion mark on growth of fourteen percent, and the central question for the equity is no longer growth but how that growth gets monetized. The quarter ended in late July delivered a record adjusted EBITDA margin of 48 percent alongside growth in account assets of fourteen percent, which means the machine is compounding balances while squeezing operating costs at the same time. Interpretation matters more than the print: the gross spread between what depositary partners pay on cash and the yield HealthEquity charges those partners has become the margin engine, and that spread is sensitive to the rate path in ways the record headline does not show.
Mechanically, the quarter's economics split into three streams of often divergent behavior. Service revenue grew modestly while the average service fee per account declined, a sign that pricing negotiations with large employer clients are producing volume-led rather than price-led growth. Interchange revenue expanded on wider debit acceptance, a higher-velocity stream tied to member spending behavior. Custodial revenue rose ten percent in the quarter, driven by both higher average account balances and the number of accounts earning yield, the stream where Federal Reserve policy and the depository-remuneration debate bite hardest.
The tension investors hold in one head is this: management raised full year guidance on record momentum, the Second Quarter margin implies substantial operating leverage from artificial intelligence and security tooling, and yet the float heavy model is entering a phase where the federal funds environment is fading from a tailwind to a neutral influence. If long bond yields stay near their recent plateau while the effective policy rate sits materially lower, the custodial spread narrows at the cash end first, and hedging losses already sitting in accumulated other income accounts quantify exactly how much cushion management deployed to defend that spread.
The timing catalyst is concreter than the macro. Late September through winter brings open enrollment, the concentrated sell-in window where the OBBBA expansion of HSA eligibility to bronze marketplace plans and to direct primary care members gets its first true volume test. The January quarter print, scheduled for release around March 2027, and the fiscal 2028 guidance issued alongside that print are the two decision points on which the margin arc and the funding-cost arc either confirm the compounding story or break it.
HealthEquity earns its keep on a float-like engine, and the strategic context begins with scale advantages that feed each other. As the largest independent custodian in health savings accounts by account volume, the company holds a negotiating position with depository and insurance partners that smaller custodians cannot replicate: cash gets placed at better terms, and the spread between partner remuneration and partner yield has become the quiet margin engine of the model. Scott Cutler, the president and chief executive, frames the strategy with a flywheel of helping members save, spend, and invest for healthcare, and the framework is not slogan fluff. Careful savings behavior seeds balances, balances fund the custodial stream, spending engagement seeds interchange and marketplace economics, and investment adoption converts idle cash into fee bearing assets over time.
Scale also compounds through distribution. HealthEquity cultivates partnerships with employers, benefits advisors, and health and retirement plan providers, including relationships that route new accounts through retirement and payroll platforms rather than through direct sales alone. Dr. Steve Neeleman, vice chair and founder, anchors the clinical side of the pitch, arguing that tax advantaged accounts make healthcare consumers cost conscious because their own balance sheet sits behind each claim. James Lucania, executive vice president and chief financial officer, runs the finance side of the equation, and his telling of the model leans on the virtue of efficiency: fewer service touches per account, artificial intelligence triaging member questions, and security tooling that has pushed fraud cost to roughly 1.1 basis points of volume, a level management places in the top percentile of comparable portfolios on the debit network.
The legislative environment turned from headwind story to growth story with enactment of the One Big Beautiful Bill Act in the summer of 2025. The statute makes permanent the pre-deductible telehealth exception for high deductible plans, allows holders of qualifying direct primary care arrangements to contribute to accounts, and treats bronze and catastrophic marketplace plans as qualifying high deductible coverage starting January 2026, which opens retail accounts to millions of exchange enrollees. Guidance issued by the tax authority in December 2025 cleared the administrative path. For a custodian, the consequences compound over years in balances and in the debit interchange attached to spending, and management quantify the ambition on the third quarter call with the statement that millions of exchange participants counting bronze plans as high deductible coverage become reachable in the enrollment channel.
Open enrollment is therefore not a rumor here, it is the revenue event. The sell-in window runs from late September through winter, when employers and exchange shoppers decide where accounts live for the following calendar year. The fiscal January quarter is where those decisions land in reported new accounts from sales, and the second quarter print showed new accounts from sales of roughly 202 thousand, up 24 percent year over year, evidence that the pipeline strength started before the legislative tailwind even reached its first full cycle. The distribution question is where the outsized years hide, because the accounts arriving through the new channels are the first healthcare savings accounts many younger workers have ever held, which makes cohort stewardship, keeping them funded and invested, worth more over a decade than any single season of opening volume. Satisfying the savvier read requires acknowledging what the model faces in the channel: benefit administration is a competitive business inhabited by payroll heavyweights and retirement platforms that bundle accounts into overall deals, and pricing pressure shows up in the disclosure that average service fees per account have declined even as accounts grow. Retail accounts opened around marketplace bronze plans arrive through web enrollment and direct funding paths rather than employer payroll files, which shifts mix toward smaller initial balances with higher fee capture, and the direct enrollment platform the company built for exchange eligible shoppers exists precisely to catch that flow. The contribution infrastructure matters as much as the account count: payroll deduction built the balance compounding of the last decade, and the new retail channels need equivalent automation before small balances turn into the six figure balances that power the later fiscal years. The statute also lets employers subsidize the accounts of employees who buy their own coverage, extending the pooled eligibility trend that made the category narrative mainstream among mid market employers in the first place.
The product stack divides into an account economy and an engagement economy, and the two feed each other. On the account side sit the core health savings account, flexible spending accounts, health reimbursement arrangements, commuter benefits, and COBRA administration, gathered under the consumer directed benefits label the company uses for the complementary book of about 7.0 million accounts. On the engagement side sits the marketplace platform, which packages clinical programs members can fund from their own balances, with the first named program supporting weight loss through GLP-1 medications and early retention data that management describe as encouraging. The marketplace turns the custodian from a passive savings vault into a healthcare purchasing front door, and each program added deepens the reason members keep balances at HealthEquity rather than sweeping them to a checking account. The platform abstracts the employer negotiation itself, because a benefit that pays for itself in healthcare savings sells through the benefits consultant channel without a bespoke pricing argument. Engagement becomes the product rather than the byproduct of administration, and that distinction separates the growth math of the category from the slower arithmetic of plain recordkeeping.
Technology spending now carries a double role, as infrastructure and as moat. Artificial intelligence tooling now triages member service contacts, which raises margins in the near period and lowers the marginal cost of serving each additional account, the same mechanism that took adjusted EBITDA margins from the 39 percent level toward 48 percent across five quarters. The debit card rails generate interchange revenue from merchant fees, and the 1.1 basis point fraud cost reading places the portfolio in the best percentile of the network, a security outcome that doubles as sales collateral when employers compare administrators on member protection. The company also launched a direct enrollment platform for exchange eligible shoppers, aimed exactly at the bronze plan population named in the 2025 statute.
The durability question reduces to switching friction on both sides of the platform. Employers face reenrollment friction through plan document changes, file feed integrations, and employee reeducation, all of which make the incumbent administrator the default choice unless service fails visibly, and members face transfer paperwork that most never start, so balances stay planted even through job changes. The infrastructure behind that stickiness is not ornamental: separation of member cash across multiple federally insured partners, daily reconciliation of debit settlement, and the compliance machinery around health privacy all function as scaling walls that a subscale entrant cannot fund. The embedded vendor risk question that every client diligence process now asks gives additional weight to the largest, most audited operator in the category, which is how a franchise converts a bad year for the industry into a share gain.
The competitive frame deserves a paragraph of candor. Payroll and data giants such as ADP hold account relationships inside broader human capital suites, legacy banks including Fidelity and HSA Bank operate custodial franchises, and administrators such as Wex compete for the engagement layer. The archive reveals how the largest competitor has moved before: a firm added roughly 22 percent to managed balances in mid 2026 through a transfer deal, and HealthEquity itself later added hundreds of thousands of accounts through a custodial arrangement with a marketplace lender, named SoFi, in a single month for the transfer book.
The first half of fiscal 2027 produced revenue above the 700 million mark, up eight percent year over year, and the composition matters as much as the total. Service revenue grew roughly four percent across the half while custodial revenue grew ten percent, so the balance driven stream is outrunning the fee driven stream, which is exactly what the thesis needs if balances keep compounding. Cost of revenue declined outright, from roughly 200 million to 191 million, meaning the company absorbed a mid single digit billing increase while trimming direct costs, a combination available only when automation removes service touches faster than volume adds them. The mix detail is the structural news: interchange kept climbing on velocity, custodial widened on balances, and the residual service stream carried the pricing softness.
Consolidated profitability set records in the quarter closed at the end of July. Adjusted EBITDA reached a fresh record of 167.0 million in the quarter, a margin of 48 percent of billing. A year earlier the same margin stood at 46 percent, on direct costs near a quarter of billing, so the gain compounds automation savings with scale discipline rather than a single item. The quality of the print needs its own sentence, because the adjusted lens removes items that remain real economics for shareholders: stock based compensation ran above 41 million in the half, and amortization of acquired intangibles ran about 53 million, both absorbed by GAAP results but excluded from the non-GAAP measures investors anchor on. The operating cost discipline matters in the comparison too, because the framework needs to price not just what the model earns but how little it spends earning it, and the separation between corporate cash and the funds held on behalf of members remains the governance fact underneath the whole structure. Net income for the half came in above 135 million, and the gap between that figure and the adjusted number is the gap analysts should price rather than assume away. Guidance rose with the print, and the full year outlook for adjusted EBITDA settled in a range just above the 620 million mark for the fiscal year ending in January 2027.
Liquidity and the capital return story complete the picture. The revolver sits drawn at a third of its size, against a facility of about a billion that matures in 2029. The senior notes of 600 million principal carry a fixed coupon also due in 2029, and the two structures together leave refinancing risk distant rather than imminent, with no maturity wall inside the coming fiscal year. Repurchases in the half ran to a bit under a fifth of a billion before the pace accelerated in earnest. Roughly 108 million landed in the second quarter alone after the board added a fresh 1.0 billion authorization in May, a signal that management sees the equity as undervalued relative to the float compounding inside it. That stance costs real cash flow each quarter, and the depth of the commitment gives the capital return story a credibility that a lighter authorization never carries. The hedging cost of defending the custodial spread already shows in equity accounts: treasury bond forwards produced losses of about 64 million in the half, of which roughly 37 million landed in the second quarter as long yields stayed elevated.
The named thesis variables reduce to three, and each has a number or decision that resolves it. The first variable, the custodial spread path, turns on where federal policy and the long bond settle: the effective policy rate stands near 3.63 percent while the ten year yield holds near 4.66 percent, an inversion shaped curve that pressures the cash leg of partner economics first, and the hedging losses already booked quantify management's bet that the defense was worth it. Watch the treasury bond forward notional and the reclassification guidance, which management pegs near 6.1 million of swing into custodial revenue over the next year, small relative to the stream but directional for the slope of the margin curve.
The second variable, the open enrollment yield from the legislative expansion, resolves between the October and January prints. New accounts from sales grew 24 percent in the July quarter, and the January quarter is where enrollment season lands in the reported funnel. The mechanism to watch is tender composition: retail accounts from exchange shoppers carry smaller initial balances than employer routed accounts, so account growth can beat while balance growth lags, and the rate of change in average balance is the tell that separates volume quality from volume optics. The third variable, the marketplace and GLP-1 monetization arc, is earlier stage: the first clinical program launched in the winter window with encouraging early retention, and each employer sale multiplies the reach of the platform without proportional sales cost, because the distribution channel already exists inside the 7.0 million complementary community.
The counterargument deserves a direct statement rather than a footnote, because a serious case runs the other way on both pillars of this thesis. The bear reads the margin surge as a harvest phase rather than a structural reset: automation yields are front loaded, competitor pricing attacks the service fee base exactly where weakness already shows, interest rate normalization reverses the spread tailwind, and the newly eligible retail accounts arrive with small balances and high service intensity, which is a poor mix for a margin story. That case implies multiple compression toward the low teens on the guided engine even without any operational stumble, and it treats the raised guidance as a peak rather than a midpoint. The rebuttal is that mix deterioration of that kind has a visible signature, in the service margin line and in the balance per cohort, and neither signature appears in the first half data, so the bear case stays a scenario rather than a trend.
Execution risk concentrates in the January quarter operational bill. Enrollment season is the heaviest service period of the administrative calendar, and cost discipline in the remainder of the year shows up most credibly there, where automation either absorbs the seasonal surge or gives way to headcount. The forward question management wants the market focused on is whether that surge gets absorbed at a margin at or above the 48 percent just printed, because fiscal 2028 guidance arrives with the March release and any glide path lower resets the multiple conversation. Fiscal 2027 guidance places non-GAAP earnings per diluted share in the band near 4.70. Revenue guidance sits between 1.411 and 1.421 billion, and the spread between the guide and the Street, which had already drifted above the initial range into the print, is the scoreboard for the back half. The sequencing matters as much as the level, because a quarter that absorbs the surge with flat service costs re-rates the whole arc, and the market has historically paid for that proof with multiple expansion rather than with estimate revisions alone.
The fiscal 2025 breach litigation anchors the governance side of the risk ledger. In the spring of 2024 a business partner account holding personally identifiable information was compromised, the notification eventually reached about 4.3 million members, and the matter consolidated into a putative class action in federal court in Utah that remains unresolved. The mechanism that makes this more than legal noise is deposit kinship: members who experience identity anxiety move balances, and the interchange attached to spending follows the balance. No loss accrual was recorded because the probability and amount remain unestimable, which is procedurally defensible and economically imprecise at once, and an adverse development in the case would surface in a disclosure update rather than in the guidance arithmetic where the market watches.
Rate and spread risk carries the larger financial tail. As of the July close, past hedging losses of about 68 million sat on the balance sheet awaiting reclassification, which means additional marks flow into other income accounts as long yields move. If the easing cycle the market priced at midsummer gives way to the hold scenario hawkish officials described heading into the Jackson Hole symposium, the growth engine keeps compounding but the cash spread narrows, and the custodial growth rate would decelerate from ten percent growth toward high single digits even as accounts keep climbing.
Two further scenarios frame the downside. A directional employment downturn would throttle new account formation, because employer onboarding is procyclical with hiring, and a soft January print after two record cycles would hit the growth premium embedded in the shares. A stop at bank or credit union partners is the custodian-specific tail, an event the filing itself names in the risk language, and while diversification across partners blunts a single counterparty failure, the periodic stress of regional banks keeps the tail in the model. Cybersecurity exposure compounds all of these: the company lives on personal health and financial data, and a second incident would interact with the litigation overhang at the worst possible moment in the demand cycle.
The financial reporting arithmetics carry their own risk to perception. Adjusted EBITDA excludes stock based compensation above 41 million in the half, together with intangible amortization near the 50 million mark. Both remain real claims on the enterprise, and the gap between GAAP and adjusted net income in the half exceeded 70 million. Headline margin records built on the adjusted lens invite the multiple to compress first when the pattern of add-backs peaks, which is a risk that shows up in the shares before it shows in cash flow. Disclosure conventions around adjusted measures rarely tighten on their own, so the burden of testing the gap between the lenses belongs to the analysis every quarter rather than to the company.
The framework starts from enterprise value, the only clean lens on a model where the accounts and the balances belong to members rather than to shareholders. The September 5 close stands at 95.51 per share. Applied across roughly 82.7 million shares outstanding, that price puts equity capitalization near 7.9 billion. Adding the principal of the debt stack while excluding corporate cash near 256 million brings the enterprise measure to about 8.6 billion, and that figure carries the math through the rest of the report.
Set the enterprise measure against the guided engine and the range brackets itself. Against the raised full year outlook for adjusted EBITDA in the low to mid 620s, the mark sits between roughly 13.5x and 14x, and against the approximate 662 million midpoint implied for the coming fiscal year toward mid cycle it lands near 13x. The company screens at about 16.9x on data vendor conventions, and that gap between the vendor figure and the derived figure is exactly the pattern an analyst re-derives rather than quotes. The vendor figure embeds a market snapshot rather than a derived claim, so the divergence is arithmetic rather than opinion, and quoting it without re-deriving invites exactly the drift this framework exists to check. The derived lens is the one this analysis relies on from here.
The earnings lens requires an analyst decision before it yields a number. GAAP trailing earnings near 2.80 per diluted share puts the trailing multiple near the mid thirties, while the adjusted figure in the band near 4.70 flatters the optics by excluding real claims on the enterprise, chiefly stock based compensation and intangible amortization already treated in the risk discussion. A fair way through the fog prices the equity against the adjusted guide with a haircut for the dilution claim, which lands the effective multiple near the mid twenties, and that is where the debate between engine value and claims on the engine happens. Even after the stock comp claim, trailing operating cash generation covered the buyback comfortably, a meaningful annualized offset at the current capitalization.
Company specific history brackets the multiple range without pricing a rate regime shift, and the current regime is the analytical pivot. The float model compounds best under the old rate plateau, and at the present settlement the federal funds environment is fading from a tailwind into a neutral influence, so the custodial tailwind that powered the margin surge from the high thirties toward 48 percent has already spent part of its impulse. Accounts still compound at eight percent and assets at fourteen, so this analysis treats the current measure near the mid teens on the guided engine as a full but not generous price, and the framework passes its sanity check in both directions from that anchor.
The bear case concedes the engine and attacks the claim on it. At roughly 11x the guided engine, against about 75 per share on the same arithmetic this report has carried throughout, the multiple would still sit at something like half the density anyone attached to the shares at the winter peak, and a growth scare would hand back the remainder of the premium before it touched the balance sheet. Nothing in the franchise breaks in that world, the accounts keep arriving, and the shares still cheapen until the claims on the engine, stock comp chief among them, stop being ignored at the moment of stress. The distribution of outcomes matters here, because a scenario that damages the multiple without breaking the machine is a drawdown shape rather than a thesis break, and those two conditions separate in the data within a single pair of prints. The separator between the two arrives through the account flows, where outflows surging after a stumble would make the damage thesis shaped, while balances holding steady through a share slide would make it purely a repricing event.
The bull case argues the market still prices the custodian of a decade ago rather than the engagement platform taking shape, and it treats the spread defense as a solved problem rather than an open one. At roughly 19x the guided engine the shares sit about 35 percent above their September level, which is close to where the stock stood at its winter high, and the case leans on the behavior the archive recorded: rivals bought balances from wound down plans rather than winning them inbound, which is what consolidated scale looks like from the inside. Add the exchange funnel and the marketplace layer on top of that base, and the pattern of consecutive raised guides becomes the assumption the market prices rather than the surprise it rewards, which is the mechanism by which quality of earnings upgrades reprice an entire equity. The stronger version of the case adds the capital question, because a custodian retiring shares while compounding member balances brackets the per share value from both directions. A multiyear repurchase program of the size authorized changes the equity math in a way that sales growth alone cannot, and the float engine supplies the cash flow that funds it.
The base case holds the present multiple territory and maps to a value in the low hundreds per share, with rates settled and the enrollment season landing in the funnel at even the guided pace. This is the probability weighted read: the margin arc glides rather than breaks, the custodial stream keeps growing high single digits while accounts compound at eight percent, and the buyback narrows the per share claim at a low single digit annual clip without straining the balance sheet. The distribution around that base is asymmetric in an interesting way, because the rate risk and the litigation tail both sit on the cost side while the enrollment yield and the marketplace buildout both sit on the revenue side, and the revenue levers have spent the last year proving themselves in the funnel. What remains embedded in the price is neither euphoria nor neglect, and the framework reads that balance as fair value with a compounding bias.
The judgment favors the base case. The engine compounds, the refinancing wall stays distant, and the ongoing retirement of shares keeps narrowing the gap between enterprise value and equity value, yet the market already collects a mid cycle toll for that compounding at the September mark, which makes the present price fair rather than generous. Two settlement points decide the argument: the slope of the custodial spread through the rate transition, and whether the January quarter absorbs the enrollment surge inside the 48 percent margin just printed, and until one of them breaks the pattern the compounding earns the benefit of the doubt. The honest read weighs both sides, and it comes down on the side of the compounding: the engine keeps compounding, and the losses in the hedge book are a bounded price already counted, while the litigation overhang is a real but shallow tail rather than a structural claim on the model. Downside to the bear scenario travels through the multiple rather than the model, since nothing in the float framework breaks without a partner failure or a reporting shock, while upside to the bull scenario requires only that the January quarter prints inside the guided range and the rate path stays steady, which is why the risk reward skews mildly favorable even at a full price. The honest read on the balance of evidence is that the compounding earns the benefit of the doubt until the January quarter or the spread slope says otherwise, and both deliver their verdict within two reporting cycles.