Loading today’s stories
Invest1 distinct publisher3 min readPublished
A 17 percent billings quarter against 9 percent revenue is the live public test of AI repackaging on seat pricing. The margin cost was about $640,000; the retention ceiling looks like 106 percent.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Twenty basis points of gross margin on $321.1 million of quarterly revenue works out to roughly $640,000 [1] [1] [15]. That is what a quarter of shipping AI cost Box at the gross margin line. Figma's gross margin fell about five full points year over year on AI credit costs, with beta products burning inference against no revenue, and Canva cut its growth forecast by a third partly because consumption and usage controls could not keep up with demand [16] [17]. Box's AI margin cost is one twenty-fifth of Figma's [5], and the reason is structural: Box was a launch partner for models from Anthropic, Google, OpenAI and Meta, and shipped MCP integrations with agents including Claude for Legal, Harvey and Notion, with the token bill for several of them landing on the other vendor [18]. What Box charges for is permission. Agent guardrails, prompt injection detection, MCP-scoped permissions and agent audit trails all went behind Enterprise Advanced, the tier carrying a 20 to 40 percent premium [19] [5].
The billings figure deserves less awe than it is getting. Box invoiced $309.5 million and recognised $321.1 million, so the leading indicator is still $11.6 million behind the lagging one in absolute dollars [2]. Composition matters too. In constant currency, long-term RPO grew 22 percent against 14 percent for short-term, an eight-point spread identical to the billings-over-revenue spread [3] [8]. Customers signed longer as well as bigger, and duration flatters billings once, in the quarter the invoice goes out, then has to be lapped.
The retention ladder is the sturdier evidence. Sales and marketing was $106.7 million, up 4.4 percent while revenue rose 9.2 percent, which puts prior-year spend near $102.2 million and the incremental quarterly selling cost at about $4.5 million [11] [4]. SaaStr values the three points of NRR at roughly $38 million of annualised revenue on a $1.28 billion base [13]. G&A fell 4.7 percent, stock compensation dollars were flat, and R&D grew faster than sales at 9.3 percent [12]. That is a company buying growth off its own price list. It is also bounded, at 106 percent against Snowflake's 126 percent, where the meter runs on consumption rather than seats [9] [14].
The earnings line is the part to discount. Non-GAAP net income rose 12 percent to $55.7 million while diluted shares fell 7.5 percent to 139.7 million, producing 21 percent EPS growth [20]. Nine of those 21 points came from the share count, which is the 43 percent SaaStr attributes to the buyback [7] [21]. Strip that out and $0.40 against a $0.39 guide [3] reads as operating margin holding at 29.4 percent [2] rather than earnings power compounding. Meanwhile this quarter's revenue reflects deals signed in FY26, and this quarter's invoices get recognised over the next eight to twelve quarters [22]. The verdict on whether repackaging can reaccelerate seat-priced software therefore arrives in FY28 revenue, and Box has $59.7 million of quarterly free cash flow to wait with [4]. Enterprise Advanced revenue has now accelerated in constant currency five quarters running [10], which is the longest run of evidence anyone has, and it has moved the total company from 6 percent to 9 percent [1] [5].
Ranked by verification strength, evidence, and original report placement.
Box reported Q2 FY27 revenue of $321.1 million, up 9% year over year and up 11% in constant currency; growth had bottomed in the low 6 percent range over the prior six quarters.
Box's Q2 FY27 non-GAAP operating margin was 29.4%.
Non-GAAP EPS was $0.40 against a $0.39 guide.
Free cash flow was $59.7 million, up 67%.
Q2 billings were $309.5 million, up 17% year over year and 16% in constant currency, against guidance of low double digits; billings grew 5% the prior quarter.
Remaining performance obligations reached $1.7 billion, up 15% and up 17% in constant currency.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Detailed reported financials, single interpreting publisher
The numeric core — revenue, margins, billings, RPO, NRR, opex lines, EPS, share count — is company-reported and internally consistent, and the article's derived arithmetic checks against those figures. But the cluster contains exactly one publisher and no primary filing, transcript, or second analysis, and the causal step from AI repackaging to the billings and retention gains is asserted rather than demonstrated with product-level disclosure.
Real contract and integration signals, no tier-level disclosure
Adoption evidence is above anecdote: a 17% billings quarter against a low-double-digit guide, $1.7 billion of RPO weighted to long-dated contracts, NRR climbing to 106%, five straight quarters of Enterprise Advanced constant-currency acceleration, and eight named third-party MCP integrations plus four model launch partnerships. What is missing is any Enterprise Advanced customer count, attach rate, or dollar contribution, and any usage data for the shipped agent-governance features, which caps how far the adoption read can go.
Modestly overstated causality, self-limiting narration
The framing — a live public test of AI repackaging working — runs slightly ahead of what the numbers prove, since none of the disclosed lines isolate AI packaging from contract-length, FX, or ordinary enterprise renewal effects, and Enterprise Advanced revenue is described only as 'accelerating' without absolute figures. The gap stays small because the same article volunteers the deflators: absolute growth is still modest, 106% NRR is middling and 'stopped there', 43% of EPS growth came from the buyback, and the buyback cost roughly $4 million a quarter of interest income.
Company-framed metrics relayed by an audience-aligned publisher
Most favorable inputs originate with Box, which has a clear interest in presenting its AI repackaging as working: billings, constant-currency framing, NRR steps and 'fifth consecutive quarter of acceleration' are all issuer-selected disclosures. SaaStr's audience is SaaS founders and executives, which rewards clean pricing-playbook narratives and cross-company comparisons. Offsetting this, the article volunteers unflattering items — the EPS guide cut, buyback dependence, falling interest income and the retention ceiling — that a purely promotional treatment would omit.
Numbers solid, interpretation single-sourced
Confidence is anchored by dense, checkable reported figures and by the article's own disclosure of offsetting facts, but held down by single-publisher coverage, the absence of primary filings or transcripts in the cluster, no Enterprise Advanced-level revenue disclosure, and a truncated final section that leaves the constant-currency argument incomplete.
invest
Canva guided 2026 growth down to 20%. The worse detail is who never got asked.1 distinct publisher
product
Washington's secret AI test is coming for open weights, and release dates go with it2 distinct publishers
product
ChatGPT Work's real ask is your Slack, and somebody has to say yes on everyone's behalf1 distinct publisher
invest
Databricks raises $5B at $190B, and the multiple barely moved2 distinct publishers
Distinct publishers with included, body-backed reporting in this cluster.