SAP Q3 2026: A €24.1B Cloud Backlog, and How Much of It SAP Bought

SAP ended Q3 2026 with €24.1 billion of current cloud backlog, up 26% at constant currencies. That is the same growth rate as Q2, after four acquisitions in six months. On SAP's own disclosures, the deals explain about 1 point of that growth at most. The SAP cloud backlog is growing on its own, and the acquisitions are showing up in profit instead.

The short answer. SAP's Q3 2026 earnings show current cloud backlog of €24.1 billion, growing 26% at constant currencies. Acquisitions of Reltio, Dremio and Prior Labs explain roughly 1 point of that at most. Organic backlog growth sits near 25%, close to last year's pace, while the deals cost SAP over €100 million of 2026 operating profit.

SAP's Q3 2026 earnings in five numbers

SAP reported on 21 October 2026. The headline release is short, and the detail sits in the quarterly statement. Here are the five lines that matter for the backlog question, all from SAP's own filing.

Current cloud backlog was €24,144 million, up from €18,839 million a year earlier. Current cloud backlog is the contracted cloud revenue SAP expects to recognise over the next 12 months. It is not total contract value, so a five-year deal only counts its next year.

Cloud revenue was €6,668 million, up 26% nominally and 25% at constant currencies. Cloud ERP Suite revenue grew 29% nominally and 28% at constant currencies. CFO Dominik Asam said on the call that Cloud ERP Suite now accounts for 89% of total cloud revenue.

Total revenue crossed €10 billion in a quarter for the first time, at €10,270 million, up 13%. Non-IFRS operating profit was €3,030 million, up 18%. Free cash flow was €1,718 million for the quarter and €7,968 million for the first nine months.

Two of those numbers carry asterisks that SAP itself attached. Cloud revenue growth was helped by what the statement calls "a multi-quarter revenue catch-up from customers returning from financial distress". Asam put the quarter-specific effects at approximately 1 point of cloud revenue growth. Profit growth was flattered by two charges that depressed Q3 2025, one for tax litigation and one for a workforce transformation.

My read is that none of these asterisks touches the backlog line. Backlog is a contract measure, so a revenue catch-up from a recovering customer does not inflate it. That is why backlog, not revenue, is the right line to test SAP against.

SAP cloud backlog, quarter by quarter

A single quarter tells you little. The table below lines up five quarters of current cloud backlog from SAP's own releases, with every acquisition or deal-structure effect SAP chose to disclose.

Sources: SAP results releases and quarterly statements for each quarter, dated 22 October 2025, 29 January 2026, 23 April 2026, 23 July 2026 and 21 October 2026. Q3 2025 and Q2 2026 values in millions come from the comparison columns of the following year's statements. SmartRecruiters, consolidated from 11 September 2025, sits inside the year-on-year comparison for the first three quarters of 2026 and was never quantified for backlog.

Why nominal and constant-currency growth diverge

SAP reports in euros but sells heavily in dollars. When the dollar weakened against the euro through 2025, dollar contracts translated into fewer euros, so nominal growth fell to 16% in Q4 2025. As the comparison base reset, nominal growth climbed back above the constant-currency rate.

For a question about competitive position, constant currency is the right lens. Currency tells you about the exchange rate, not about whether customers are signing. Every comparison in the rest of this post uses it unless stated otherwise.

The Q4 2025 dip was contract structure, not demand

The weakest constant-currency print in the run was Q4 2025 at 25%. SAP attributed about 1 point of drag to large transformational deals with cloud revenue ramping in outer years, plus termination-for-convenience clauses required by law. Those deals land in total cloud backlog, which SAP put at €77.29 billion at year end, but they barely touch the next 12 months.

That matters for reading Q3. Some of the 2026 strength may simply be those back-loaded contracts rolling into the 12-month window. It is a timing effect, and it flatters the current quarter without telling you anything new about demand.

What Reltio, Dremio, Prior Labs and TechWolf add to the SAP cloud backlog

SAP has bought four companies in 2026. Every one is pitched as part of its data and AI story. Reltio is a master data management platform. Dremio is a data lakehouse. Prior Labs builds tabular foundation models, which are AI models trained on structured data rather than text. TechWolf builds what SAP calls an AI work intelligence platform for HR.

The question is how much of the backlog growth those deals bought. SAP's disclosure on this is thin but not empty.

Sources: SAP Q2 2026 quarterly statement filed on Form 6-K, 23 July 2026, and SAP Quarterly Statement Q3 2026, 21 October 2026. Dremio and Prior Labs prices were not stated in either document.

Reltio is the only deal with a backlog number

The Q2 statement says Reltio's first-time inclusion "contributed less than 1 percentage point to the constant currencies growth rate" of current cloud backlog. That is the only time in five quarters SAP has quantified an acquisition's effect on the backlog line. It also tells you the scale: a $1.26 billion purchase moved a €23 billion backlog by under a point.

Dremio and Prior Labs show up in profit, not backlog

SAP gave no backlog effect for Dremio or Prior Labs in Q3. What it gave instead was a guidance cut. In July, SAP lowered its 2026 non-IFRS operating profit outlook to €11.8 billion to €12.2 billion, from €11.9 billion to €12.3 billion. The stated reason was dilution from the two deals, "projected to be in excess of €100 million".

TechWolf is not in the numbers yet

The Q3 statement records that SAP "entered into an agreement" to acquire TechWolf, with no price. It had not closed by the end of the quarter, so it contributes nothing to Q3 backlog or revenue. Anyone attributing Q3 growth to TechWolf has the timeline wrong.

Estimating organic SAP cloud backlog growth

SAP does not publish an organic backlog growth rate. You can still bound it. The method below is mine, not SAP's, and every input comes from the Q3 statement.

Start with the ratio of backlog to quarterly cloud revenue. SAP's €24,144 million of current cloud backlog is 3.6 times its €6,668 million of Q3 cloud revenue. Now assume, generously, that all €49 million of acquired revenue was cloud subscription with the same backlog ratio. That implies about €177 million of acquired backlog.

Against the €18,839 million base a year earlier, €177 million is 0.9 points of growth. Currency added roughly 2 points on top, the gap between 28% nominal and 26% at constant currencies. Strip both out and organic constant-currency growth lands near 25%.

Compare that with Q3 2025, when constant-currency backlog growth was 27%. On this estimate, SAP's organic pace has eased about 2 points in a year. That is a slow deceleration, and it is exactly what SAP guided to: "constant currencies current cloud backlog growth to slightly decelerate" in 2026 from 25% in 2025.

My view is that this is a healthy result. It is not an accelerating one. The market narrative after the print leaned on the 28% nominal figure, which is the least informative number in the release. If you want the same method applied to a different incumbent, the breakdown of Salesforce's Agentforce ARR with Informatica stripped out runs the organic test on a CRM vendor.

The profit cost of buying AI data assets

If the deals barely move backlog, why buy them? SAP's answer is strategic: it wants to own the data layer that its AI agents run on. The cost of that strategy is visible in operating profit, and it is larger than the revenue it brings in.

In Q3, the three closed acquisitions contributed €49 million of revenue and a €39 million non-IFRS operating profit drag. On IFRS, which includes acquisition-related charges, the drag was €59 million. So each euro of acquired revenue cost SAP close to 80 cents of adjusted profit in the quarter.

Add the €39 million back and non-IFRS operating profit would have been about €3,069 million. Growth would have been close to 20% instead of 18%, on my arithmetic. That is the price of the strategy in one quarter, and SAP's full-year guidance puts the Dremio and Prior Labs share alone above €100 million.

There is a second, smaller signal in gross margin. Asam said on the call that non-IFRS cloud gross margin was 74.3% in Q3, down 0.7 points year on year. SAP did not attribute that decline to acquisitions or to AI compute. You should not attribute it either, but it is the number to watch if AI usage grows inside fixed-price contracts.

The strategic bet is not obviously wrong. Paying for data plumbing is how an ERP vendor makes its agents useful. The debate about whether that plumbing is defensible is covered in the comparison of what SAP and Oracle customers actually switched on. But a buyer of SAP shares, or SAP software, should treat it as a cost centre today.

What SAP disclosed about Business AI, and what it did not

The target of this post was a Business AI number. There is not one. The Q3 statement contains no AI revenue, no AI backlog, no AI consumption units and no count of paying AI customers. It lists seven customers that "chose SAP's AI and data solutions" and nothing more quantitative.

The earnings call adds claims, not metrics. CEO Christian Klein said "nine out of 10 of our 50 largest deals included AI and Business Data Cloud". He said SAP now has "40 assistants and 300 agents across the autonomous suite". He also said "the majority of our deals are closed with AI".

Why attach claims are weak evidence

An attach claim tells you that AI was in the contract. It does not tell you what share of the contract value was AI, whether the customer paid separately for it, or whether anyone uses it. Bundling AI into large ERP renewals would produce a 9-in-10 attach rate with zero incremental AI revenue.

That is not an accusation. It is a description of what the number can and cannot prove. The disclosure pattern is common across software vendors, and the ranking of AI revenue disclosure metrics scores which versions carry information and which do not.

What SAP said about pricing

The more revealing line came from the CFO. Asam said SAP wants "to sell outcomes to customers as opposed to tokens". He described the credit model loosening over time toward "hybrid models all the way to full consumption based". Klein, asked about agent quality, said "probably on accuracy, we will never be done".

I read those as honest statements of a company still working out how to charge for agents. They are also the reason a Business AI revenue line has not appeared. You cannot report a clean AI number while the pricing unit is still moving.

Is SaaS dead? What the ERP backlog says about agents

The "is SaaS dead" argument says AI agents will hollow out seat-based application software. If that were happening to ERP, you would expect it first in the backlog: fewer new cloud ERP contracts, shorter terms, slower renewals. The Q3 numbers show the opposite at the core.

Cloud ERP Suite backlog grew 28% at constant currencies to €5.92 billion. That is the system-of-record business that agents are supposed to make obsolete, and it is growing faster than SAP overall. The quarter does not support a story about agents eroding ERP demand.

The weak spot is Extension Suite, up only 8%. That is the platform layer customers use to build on top of SAP. It is also the layer most exposed to cheaper agent-built and AI-coded alternatives. I would not call 8% a collapse, but it is the line where I would look first for agent substitution.

The real decline at SAP is in the old models. Asam said software licence revenue fell 9% in Q3, and SAP guides that the software support decline "will accelerate" as customers move to cloud. Those are migrations inside SAP, not losses to agents. For how this compares with other application vendors, the SaaS revenue growth tracker sets SAP's rate against its peers.

My position: on this evidence, ERP incumbents are gaining ground, not losing it. The agent era is so far being sold as an add-on to the system of record. Whether that holds when pricing moves to consumption is a 2027 question, and SAP has told you it is coming.

Where this reading is weakest

Every number above is SAP's, and every interpretation is mine. Here is where the interpretation can break.

Backlog is a contract measure, not a usage measure

Current cloud backlog tells you what customers have committed to pay over 12 months. It does not tell you whether they use what they bought. A customer can sign a RISE renewal with AI bundled in, never switch on an agent, and still add to the backlog. If agents erode ERP, the first sign may be smaller renewals in 2028, not slower backlog in 2026.

The other measurement problem is concentration in Q4. Asam said the fourth quarter "accounts for around 50% of our annual bookings volume". He also warned that conditions "could weigh on Q4 bookings and, as a result, our 2027 trajectory". A Q3 read covers the quiet half of the year.

The organic estimate is ours, and it is crude

SAP's 26% is rounded, so the real figure could sit anywhere from 25.5% to 26.5%. The acquired-backlog bound assumes all €49 million was cloud subscription with SAP's average backlog ratio. A startup with longer prepaid contracts could carry more backlog per euro of revenue. If the acquired ratio were double SAP's, the acquisition effect would be near 2 points, and organic growth closer to 24%. That would still be a slow deceleration, not a break.

There is also the competing view to concede. Some analysts on the call read the quarter as slightly accelerating on an organic basis. Management did not confirm it, and my arithmetic does not support it. But a more favourable currency method could, and I cannot rule it out from public data.

Frequently asked questions

Where to start before SAP's Q4 print

Read the Q4 2026 statement in January for one sentence: whether SAP quantifies any acquisition contribution to current cloud backlog. If it says under 1 point again, organic growth is holding. If it stays silent while Dremio, Prior Labs and TechWolf are all consolidated, treat the silence as information.

If you buy SAP rather than own its shares, ask your account team this week what share of your RISE renewal price is attributable to AI. If they cannot answer, the 9-in-10 attach figure includes you, and you are paying for something nobody has priced.

Related reading