Summary

  • Bentley reported US$1.536 billion of ARR at 30 June 2026. For 51% of the total, the company annualized the last three months of recognized revenue from contractually recurring consumption subscriptions with measurement periods shorter than one year.
  • The estimate has substantial support: account retention was 99%, recurring-revenue dollar-based net retention was 109%, the 51% share was unchanged year on year, and last-twelve-month recurring revenue reached US$1.486 billion.
  • E365 usage, US$496.322 million of paid-but-unconsumed CSS deposits and US$280.641 million of remaining performance obligations are separate states. Deposits can roll over or be refundable; RPO is not a denominator against which ARR can be declared “covered.”

One ARR label, two measurements

At 30 June, Bentley’s ARR was US$1.535988 billion, up from US$1.379161 billion a year earlier. The company calculates constant-currency growth at 12%. Those are straightforward reported figures. The more consequential disclosure sits inside the definition.

Bentley first takes the annualized value of the contract portfolio that produces recurring revenue at the reporting date. It then adds the annualized value of the last three months of revenue already recognized from contractually recurring, consumption-based subscriptions whose consumption-measurement periods are shorter than a year. Spot exchange rates are used. The latter method generated 51% of total ARR in both June 2026 and June 2025.

This is not evidence that half the ARR is imaginary, nor that half is uncontracted. The subscriptions are explicitly described as contractually recurring, and the input is recognized revenue rather than a sales forecast. It is, however, a different kind of claim from a fixed annual contract value. A recent quarter is being asked to stand for the next four quarters. Workloads, users and asset activity must keep appearing for the estimate to hold.

The distinction matters because Bentley sells software into infrastructure design, construction and operation. Use can rise when a large engineering programme enters an active phase, and ease when work moves between stages. Annualizing three months catches expansion sooner than waiting for a full trailing year. It can also carry a concentrated burst of activity forward. ARR is therefore a timely operating estimate, not a promise that every reported dollar is already an allocated performance obligation.

E365 turns access into a measured activity

E365 is central to that second clock. The programme represented 46% of total ARR in June 2026, up from 45% a year earlier. It produced US$168.730 million of revenue in the second quarter and US$337.869 million in the first half, against US$150.166 million and US$297.070 million in the comparable periods.

Bentley describes E365 as a global, consumption-based subscription with access to an integrated software portfolio and uniform pricing across countries. Most E365 revenue is attributed to daily application consumption. Some Bentley Infrastructure Cloud offerings are instead charged by the number of users during a calendar quarter or by fixed asset bands. This is a usage system, but not a single meter.

Nor is every charge free to fall to zero. E365 subscriptions typically contain quarterly usage floors or ceilings, and revenue is recognized on actual use. Floors can provide a contractual base; ceilings can limit what Bentley collects during intense use. The filing does not publish the aggregate floor, aggregate ceiling, current headroom or distribution by account. Any analyst who calls the entire 51% either guaranteed or completely variable is inventing a sensitivity that Bentley has not supplied.

That missing sensitivity is the analytical centre of the story. A quarterly floor can make recent usage more durable than a pure on-demand meter, but only the size of the floor reveals the protection. A ceiling can strengthen the customer proposition by containing surprise bills, but it may also prevent Bentley from capturing all upside during peak activity. Without the aggregate terms, the reported retention and realized revenue are better evidence than a hypothetical contract model.

The deposit account is funding, not consumption

E365 participants use the Cloud Services Subscription programme as their funding mechanism. CSS asks an account to estimate annual use and deposit funds in advance. Actual consumption is then monitored and invoiced against the deposit each calendar quarter. Balances that are not used for eligible products or services may roll into future periods or be refunded.

At 30 June, paid and unconsumed CSS deposits stood at US$496.322 million, up from US$463.312 million at the end of 2025. The first-half cash-flow statement recorded a positive US$37.260 million change in CSS deposits, compared with US$27.426 million a year earlier. At the end of 2025, accounts representing approximately 60% of total ARR had adopted commercial models eligible for CSS.

This is valuable evidence. Customers have put substantial cash into Bentley’s procurement system, and Bentley can observe quarterly drawdown against funded balances. The mechanism can reduce purchase-order friction and gives the company working capital before use becomes revenue. But a deposit is not an exercise session already completed. Bentley records paid-but-unconsumed balances as a liability precisely because the service has not yet been consumed, and unused funds retain rollover or refund rights.

The two extreme readings both fail. Adding the deposit balance to ARR would count funding as performance. Dismissing the balance as irrelevant would ignore a substantial signal of procurement intent and liquidity. The disciplined reading is that CSS strengthens the path from budget to use while leaving the final revenue outcome dependent on actual consumption and the customer’s eligible choices.

RPO answers a different question

Bentley reported US$263.154 million of current deferred revenue and US$17.487 million of long-term deferred revenue. Its remaining performance obligations were US$280.641 million, about 94% of which the company expects to recognize within the next 12 months. At this reporting date, RPO equals the two deferred-revenue balances added together.

That arithmetic invites a bad comparison. ARR of US$1.536 billion is much larger than RPO of US$280.641 million, but subtracting one from the other produces no meaningful “uncovered ARR.” ARR is a non-GAAP run-rate metric assembled from recurring contracts and recent recognized consumption. RPO is consideration allocated to customer-contract obligations that will be satisfied later. Variable consideration and performance-obligation rules can place arrangements outside the intuitive backlog picture. Bentley does not disclose enough to turn the difference into risk capital.

Deferred revenue has its own flow. During the first half, Bentley recognized US$191.560 million that had been in the opening deferred-revenue balance and recorded US$182.639 million of new deferrals, mainly from new billings. US$19.473 million of the closing balance related to portfolio-balancing exchange rights. These movements describe payment and performance timing, not the current pace of every recurring workload.

Four labels should therefore remain four labels. ARR estimates an annualized recurring pace. E365 records how software is consumed and charged. CSS deposits hold funding not yet consumed. RPO records allocated obligations still to be performed. They connect commercially, but they are not interchangeable accounting containers.

The estimate has earned credibility, not immunity

Bentley offers meaningful counterevidence to the claim that its run rate is merely quarter-end optics. Account retention was 99% in both June periods, and recurring-revenue dollar-based net retention was 109%. The 51% consumption-annualized share did not jump as ARR grew. Last-twelve-month recurring revenue reached US$1.485992 billion, only about US$50 million below period-end ARR, although the two measures cover different clocks and must not be forced into a formal reconciliation.

The reported revenue line also moved with the story. Second-quarter subscription revenue rose US$45.183 million to US$378.635 million. Bentley says expansion among existing accounts was the principal driver, while growth attributable to new accounts was 3%, most notably among small and medium-sized accounts. Existing-account expansion is consistent with 109% net retention. It also makes customer use, rather than contract labels alone, the continuing growth engine.

Scale diversifies that behavior. Bentley reported nearly 42,000 accounts in 189 countries at the end of 2025. Yet “account” is a contractual and billing relationship; affiliated entities under one parent can have independent accounts. It is not a count of distinct companies, projects, users or people. Geographic breadth reduces dependence on one construction cycle, but the filing does not reveal concentration within the consumption-annualized pool.

Cash provides one final check. First-half operating cash flow was US$264.921 million, down from US$280.500 million, even though the CSS deposit movement was more favorable. The deposit system can assist working capital without determining the entire cash result. That is another reason to resist using a single balance as proof of the business model.

The fair conclusion is neither that ARR is fake nor that ARR is backlog. Bentley has constructed a fast measure for a business in which access is increasingly metered. Its retention, net expansion and realized trailing revenue make the measure credible. Its undisclosed floors, ceilings, usage distribution and refundable funding make quarterly behavior the durability test.

Sources