SAP says it is using AI on itself, and it has put a number on the result it wants: an “efficiency run rate by end of 2028” of more than โฌ2 billion, about $2.2 billion. The figure is on a slide titled “Turning SAP into an Autonomous Enterprise” in SAP’s third-quarter 2026 earnings presentation, published on 21 October 2026.
The same slide gives SAP’s early results: an increase of about 70% in code pull requests merged per developer this year, more than 110,000 users of Joule Work Desktop, and productivity gains of more than 20% across central functions.
Business Pill · AI AGENTS
A short Business Engineer explainer on AI agents: software that moves from answering questions to doing tasks. It teaches the general idea only and says nothing about SAP or any company in this piece.
Sap Q3 2026 explained · video
The quarter in about a minute and a half, from our published pieces (21 Oct 2026).
The key insight: As we read it, the slide’s one dated, priced number is the target: more than โฌ2 billion of efficiency run rate by the end of 2028.
SAP’s own AI figures for its work inside SAP; “up to” figures are SAP’s maximums. Source: SAP Q3 2026 earnings presentation, 21 October 2026.
What the Numbers Mean
A pull request is a proposed code change; merging it means it goes into the product. So the 70% figure counts more finished changes per developer. It does not say how large each change is.
A run rate is a figure expressed as a yearly amount. SAP’s slide does not define how the โฌ2 billion will be counted or against what starting point.
Staff Grew While the Tools Spread
SAP’s statement reports 113,076 full-time equivalents at 30 September 2026, against 110,730 a year earlier, about 2% more (our arithmetic). Research and development staff rose to 38,735 from 37,909.
That runs the other way from Meta, whose headcount fell 15% to 66,775 over the same period, as our Meta headcount piece reported. SAP’s slide does not link its AI figures to staffing plans.
SAP’s presentation also shows sales and marketing costs at 21.7% of revenue in the quarter, down 1.6 percentage points, and its non-IFRS operating margin rose to 29.5% from 28.3%. The presentation does not attribute those changes to the AI programme.
How We Read It
As we read it, SAP is presenting itself as its own first customer: the slide that sells “Joule Work” to clients sits next to the one that reports how SAP’s own staff use it.
As we read it, the โฌ2 billion target is the number to track, because it is the only one on the slide with a date and a currency attached.
The Structural Read
As we read it, SAP is presenting itself as the first user of the product it sells, Joule Work.
As we read it, the staff figures cut against the idea that AI gains arrive as job cuts: SAP’s full-time equivalents rose about 2% (our arithmetic).
As we read it, the “up to” figures describe best cases, so the 70% code figure and the โฌ2 billion target carry more weight.
Christian Klein, SAP chief executive, Q3 2026 statement
“They see that Joule Work makes the business case for their transformation more attractive than ever.”
Three Implications
FOR SAP CUSTOMERS As we read it, the internal figures are SAP’s sales case for Joule Work, measured on SAP’s own staff.
FOR SAP STAFF The slide gives no staffing plan. As we read it, how the โฌ2 billion is counted will matter to SAP’s workforce.
FOR INVESTORS As we read it, the โฌ2 billion target gives investors a dated figure to check against SAP’s cost ratios over the next two years.
The framework names two forces splitting the software market: “the democratization of development through AI-assisted coding and the intensification of enterprise integration dependencies.”
As we read it, SAP’s slide reports both from the inside: more code merged per developer, and an assistant built into its own workflows. The framework does not discuss SAP.
What Is Not Established
The slide gives no baseline for the 70% code figure, no method for the productivity figure, and no breakdown of the โฌ2 billion between staff costs and other spending. Up-to figures describe the best cases SAP chose to report.
Connecting the Coverage
Our results piece covers SAP’s โฌ10.27 billion quarter. Our AI and work piece reported a survey in which 58% of managers expect to oversee more AI agents than human employees within the next year, and our Meta headcount piece covers Meta’s 15% cut.
What’s new here: SAP’s own numbers for what AI is doing inside SAP, and a dated โฌ2 billion target.
SAP reports about 70% more merged code per developer and more than 110,000 internal users of Joule Work Desktop, and targets an efficiency run rate above โฌ2 billion by the end of 2028. Its staff numbers rose about 2% over the year (our arithmetic).
A note on sourcing. SAP figures and quotations come from SAP’s third-quarter 2026 quarterly statement and earnings presentation, published 21 October 2026. Dollar figures are our conversions at the European Central Bank reference rate of $1.1234 per euro on 21 October 2026. Headcount figures are full-time equivalents from SAP’s statement. The Meta and survey figures come from our earlier pieces. We did not contact SAP. Nothing here is investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
