Box, Inc. BOX

The business

Box sells cloud content management: one secure online place where a company keeps its files, with sharing, permissions, e-signature, compliance controls, and now AI that can read a contract or a claims file and answer questions about it. It serves paying organizations weighted toward regulated industries—financial services, healthcare, government, legal—where losing a document or failing an audit is expensive. The money comes in as annual per-seat subscriptions, billed up front. The moat is switching cost: once a firm makes Box its system of record and wires its compliance rules and workflows into it, moving to a rival means re-permissioning millions of files and retraining everyone. The thing a customer actually buys today is the "Enterprise Advanced" tier, a higher per-seat price that bundles the AI features. Box's pitch against Microsoft and Google, who give file storage away inside bundles firms already own, is that it stays neutral and works across both.

The numbers

The sequence shows a business that stopped growing, then started again.

Revenue growth ran to 9.2% year over year last quarter, the fastest print in this window and up from the mid-single digits Box was stuck in through FY2024 and FY2025. That is the whole near-term case: the AI tier is pulling growth back up. The earnings column lies, though. The $68.5M in Q4 FY2026 carried a tax benefit, and the tax noise the record has shown for two years held again this quarter, a 40% effective rate on pretax income made reported EPS a poor read on the business, which is why the cash line, not earnings, anchors the price here.

The lens that matters for a software name is what growth costs and what it converts into.

Free cash flow held near 30% of sales through the slow years and the fast ones, which is the point: this is a cash machine whether it grows 5% or 9%. Trailing free cash flow is about $350M against a $4.9B market value, a 7.8% cash yield, and the diluted share count fell from 151M to 140M over the last year as the company bought back stock. If revenue compounds at high single digits and that margin holds, cash per share grows around 10% to 12% before the yield. What this memo believes that the market does not: the 13-times-cash price treats Box as a no-growth file cabinet, and the reacceleration may be more durable than that if the AI tier keeps lifting the per-seat price. The print that settles it is whether revenue growth stays above 9% with remaining performance obligations rising alongside it.

Management

Aaron Levie, the founder, still runs it, and his pay tells you where his money is: $257,076 last year, almost all of it cash incentive, no new stock grant, about 0.25% of net income, because he already owns a large founder's stake. Co-founder and CFO Dylan Smith sold about $2.2M in September and October 2025; total insider selling was 42 sales for $12.4M against no buys, roughly a quarter of one percent of the company, modest and routine, though the filings here do not split scheduled 10b5-1 sales from discretionary ones. Buybacks are the story: $290M retired in FY2026 at an average price near $26.89, below today's $35.19, and consistent every year for five years. Sitting ahead of common holders is the KKR Series A preferred from 2021, which draws about $14M a year and is why cash to common runs below reported net income.

How it fails or surprises you

Seat pricing meets AI (the fact the read explains least well). Box charges per user. The same AI that justifies the pricier tier also lets customers do more work with fewer people. If the 9% reacceleration is a one-time price and tier bump rather than durable demand, growth rolls back to mid-single digits within two prints and the re-rate case dies. Watch seats per customer and the growth rate together.

Hyperscaler bundling. Microsoft and Google give content management away inside bundles enterprises already pay for. Box's whole pitch is neutrality. If a large customer consolidates onto SharePoint at renewal, it shows up first in remaining performance obligations, the signed-but-unbilled backlog, not in reported revenue, which lags a quarter or two. Watch backlog growth against revenue growth for the early crack.

The AI tier re-rates the whole name (right tail). If Enterprise Advanced pushes revenue growth into low double digits and holds it, the market stops pricing Box as a utility and the 13-times-cash multiple expands. Nobody is paying for this today. The first evidence is two consecutive quarters of double-digit growth with rising backlog, which would convert a 7.8% cash yield into a genuine compounder.

Closing thoughts

A specific print settles this, so name it: revenue growth holding at or above 9% with remaining performance obligations climbing for two more quarters says the AI tier bought a new trajectory. An ambiguous print, growth drifting back to 7% to 8% with flat backlog, drops you back onto a 30%-margin cash utility at 13 times free cash flow, which is a perfectly good thing to own but not a re-rate, and the reader who wanted the compounder should wait rather than pay up near the high. My read, offered as judgment not fact: the downside tail, seat erosion and bundling, is the fatter long-run risk, but the near-term cash yield and steady buyback put a floor under the stock while that question resolves. If the downside breaks, the multiple compresses toward its cash-yield floor; if the right tail lands, the multiple is what moves, not the fundamentals.

The bet is still that big companies keep paying Box a rising yearly fee to keep their files in one safe place, and keep paying more for the AI tools layered on top. What breaks it is seat counts flattening while the AI story fails to lift the price per seat, and the one pair of numbers that tells you first is the revenue growth rate set against backlog growth. If the next two prints show growth slipping below 8% with flat backlog, the reacceleration was a one-time tier bump, not a new road.