A customer signs a three-year software agreement. The supplier has won work, the finance team has a contract to account for, and the people using the software still need a successful deployment. Those are three different stages of the same sale.
Remaining performance obligations, usually shortened to RPO, help describe the work still to be recognised as revenue. The figure is useful precisely because it looks beyond this quarter. Trouble begins when a contract balance is read as next year’s sales, money already collected or profit already secured.
For software businesses selling AI alongside established products, the distinction matters. A longer commitment can improve visibility while leaving integration, support and delivery economics to be demonstrated. Here is how to read the contract book without asking it to answer every question.
Fact: the balance has a recognition clock
Adobe (NASDAQ: ADBE) reported $22.16 billion of RPO at 28 August 2026. Its quarterly filing says approximately 67% would be recognised over the next 12 months, excluding specified enterprise customer agreements. Non-cancellable, non-refundable funds associated with those agreements represented about 3% of total RPO. That qualification belongs beside the percentage: multiplying the entire balance by 67% would lose the disclosed distinction.
Salesforce (NYSE: CRM) reported $66.3 billion of RPO at 31 July 2026, including $33.5 billion expected to become revenue within 12 months. Its filing explains that the balance includes both unearned revenue and amounts not yet billed. It also identifies renewal timing, contract length, currency and acquisitions among the influences on RPO.
Microsoft (NASDAQ: MSFT) reported commercial RPO of $678 billion for its June 2026 quarter in its investor metrics. The company defines this as the commercial portion of revenue allocated to obligations still to be performed, including amounts to be invoiced and recognised later. “Commercial” is part of the definition, rather than decorative wording to drop when comparing totals.
These are dated balances covering different businesses and reporting dates. They establish contracted activity under each company’s definitions. They do not establish a league table of product quality, AI adoption or next-year profit.
Interpretation: extend the term, change the headline
Consider a simplified example, rather than an estimate for any named company. A supplier sells an immediately available subscription for £120,000 a year. Assume revenue is recognised evenly as service is provided, with no separate implementation obligation, discount, cancellation right or variable charge.
A one-year agreement begins with £120,000 of revenue still to recognise. A three-year agreement at the same annual price begins with £360,000. Both initially contribute £10,000 a month under these assumptions. The longer contract triples the starting obligation without tripling the monthly revenue rate.
That longer commitment has value: it secures more contracted work and pushes out the next renewal decision. It also changes what the headline measures. To judge momentum, ask whether the business won more customers, sold more to existing customers or simply secured more years at once. A growing RPO balance alone cannot separate those possibilities.
The reverse deserves the same care. As a contract is delivered, its remaining balance falls. In our one-year example it reaches zero just before renewal even if the customer is satisfied and fully intends to stay. A lower balance is a prompt to inspect the renewal cycle, rather than a self-contained verdict on retention.
Cash has its own timetable
Now keep the hypothetical three-year contract but change its billing terms. The customer might pay £360,000 at the start or £120,000 at the beginning of each year. Our assumed service and revenue recognition are identical; the supplier’s immediate financing position is different.
RPO should therefore be read alongside deferred revenue, receivables and operating cash flow, rather than added to them as separate pools of value. Some of the contract balance may already have been billed. Some is awaiting an invoice. Counting all of it as fresh cash on top of the accounts would count overlapping stages of the same commercial arrangement.
There is another limit: a revenue commitment does not disclose the cost of fulfilling it. In this example, additional support hours or more expensive computing could reduce the contribution from a fixed-price contract. The contract would still contain revenue to recognise. Its profitability would require a separate calculation.
Who could benefit?
Interpretation: Adobe and Salesforce could benefit operationally when their contracted subscriptions convert into revenue while customers complete deployments and renew. The mechanism is continuity of customer spending, giving the supplier time to plan staffing, product work and support. The relevant horizon is the disclosed recognition period and subsequent renewal cycle.
For Adobe, the evidence to follow is conversion on the qualified timetable in its filing, together with delivery costs. For Salesforce, compare current RPO with subsequent recognised revenue and renewal disclosures, keeping acquisition and currency effects visible. Neither total isolates the incremental contribution of an AI product. Assigning all contracted growth to AI would go beyond the evidence presented here.
Microsoft’s commercial commitments could similarly support planning across its commercial software and cloud services. Whether that visibility becomes an attractive economic result depends on revenue delivery relative to the resources required. Its metrics also report Microsoft Cloud gross margin of 65% for the June quarter; that is a broader business measure, rather than a margin attached specifically to the $678 billion contract book. It cannot price the profitability of each remaining obligation.
Who’s at risk?
Interpretation: The same three suppliers carry execution risk after signing. A contract does not finish a customer’s integration, resolve support requests or establish the usefulness of a new tool in daily work. If delivery costs rise faster than the revenue attached to commitments, visibility may coexist with pressure on returns. This is a conditional mechanism, not a claim that any particular contract is loss-making.
The materiality cannot be inferred from RPO alone: the disclosures cited here do not supply a margin for every contract or a complete separation of AI-related obligations. Salesforce’s stated currency and acquisition influences are an additional reason to avoid treating every movement as an organic change in demand. For all three, useful evidence would connect comparable contract measures to revenue, costs and cash over several reporting periods.
Speculation: a possible outcome, not a forecast
One possible scenario is that longer agreements give suppliers time to make deployments more efficient, improving the economics of work already signed. Another is that demanding integrations and rising delivery costs absorb the benefit of that visibility. No probability is assigned to either outcome, and the reported balances do not settle which will occur.
The next assessment should start with consistent definitions, the near-term recognition share and any change in contract duration or business mix. Then examine cash conversion and the cost of serving customers. A larger contract book is useful evidence. Following it through the customer’s workflow and the supplier’s accounts makes it more useful still.
Sources checked 20 October 2026. This article explains company disclosures and conditional business mechanisms for information; it does not recommend an investment.
Photo by RDNE Stock project on Pexels. Paperwork in coloured document trays.