After establishing SAP as the world’s fastest-growing applications, agentic AI, and data company, CEO Christian Klein recently had to answer a bizarre question about whether he’ll have to follow longtime rival Oracle into the astonishingly expensive realm of AI infrastructure.
That very strange question came during SAP’s recent Q3 earnings call in which it reported excellent results, highlighted by these three numbers:
- cloud revenue jumped 27% to $6.14 billion;
- Cloud ERP Suite revenue spiked 31% to $5.32 billion; and
- current cloud backlog rose 27% to $21.8 billion.
As I noted in my analysis last week, those results — both in volume and in growth rates — are vastly superior to the most-recent quarterly numbers reported by SAP’s three largest competitors:
- SAP: $6.14 billion, up 27%;
- Workday: $2.17 billion, up 14%;
- Oracle: $3.8 billion, up 11%; and
- Salesforce: $10.2 billion, up 10%.
- For the full story on all that, please see “SAP Remains Hottest Enterprise-Apps Vendor by Far; Workday #2, Oracle #3, Salesforce #4.”
So I found the question to be bizarre because why in the world would a company performing as brilliantly as SAP is right now suddenly decide to veer off into a totally foreign category requiring Klein to invest at least $75 billion every single year to merely have an outside shot at picking up a few scraps of business?
Yes, both SAP and Oracle make applications and are surging into agentic AI and data clouds, but the similarities end there.
Oracle has been deeply engaged in enterprise hardware for the past 15 years via its acquisition of Sun Microsystems, which was the foundation for Oracle chairman Larry Ellison’s long-range strategy of “hardware and software engineered to work together.”
AI Agent & Copilot Summit is an AI-first event to define opportunities, impact, and outcomes with Microsoft Copilot and agents. Building on its 2025 success, the 2026 event takes place March 17-19 in San Diego. Get more details.
And on that foundation, Ellison and now-CEO Clay Magouyrk spent several years creating Oracle Cloud Infrastructure, which is on the verge of becoming one of the fastest-growing businesses the tech industry has ever seen.
But the price of all that success has been head-spinning, requiring Oracle to radically boost CapEx spending to meet what is now an insatiable demand for AI training and inferencing. Don’t forget that while Oracle’s trending toward spending upwards of $50 billion or $60 billion per year on CapEx, the other three hyperscalers more-expensive approaches are requiring them to invest $80 billion to $100 billion per year.
In light of that, why would anyone think SAP should — or would, or could — make an extremely radical pivot in its very successful strategy that would require it to spend insane amounts of money that it does not have on a totally new type of business for which it has relatively slight expertise and in which it would be required to claw business away from not only Oracle but also three of the world’s biggest, wealthiest, and most-innovative companies: Microsoft, Google, and Amazon?
Anyhoo, here’s what Klein had to say about his intentions, and I’ll start off with the verbatim question from the financial analyst: “If I can ask a question on your competitive position versus Oracle, do you see an incremental risk from their pivot toward infrastructure in terms of offering? They mentioned SAP a number of times at their recent [could not comprehend], so I’m hoping to get an update on where you see yourself positioning at this time.”
While Klein’s reply is a bit long, I feel it’s important to see it in its full context to help understand what SAP is doing, why it is doing it, and how that is significantly different from Oracle’s approach.
“I know that some of our competitors are playing the game on scaling the infrastructure and training LLMs. When I listen to our customers, and that is super important, they are super positive around the value creation we are now providing to them and for their end users with SAP Joule and supply chain and HR and finance.
“For us, the infrastructure is, of course, part of the stack. When you look at software and cloud, I mean, we can actually always offer highly competitive offerings in all parts of the world. There is no need in my eyes at all to change our strategy and to suddenly start building data centers everywhere in the world.
“Our strategy is proven and it works. I’m deeply, deeply convinced that when we are now looking at some of the large language modules our goal is not to train them but instead to use them. Many of them, like OpenAI, are now coming to SAP and saying, ‘Hey, for this applied AI, I want to do business in the public sector — can you work with us? Can you give us the BDP? Can you give us Joule Studio to build all of these agents? Because we need a development platform. We need this applied AI thing.’
“Over time, we need to move up and deliver that and we see a huge momentum there also working with these LLM providers in really not only infusing them in our technical stack, but now going to customers together and actually delivering AI agents together.
“That is for me the absolute winning formula. Again, it’s for us super, super important to make our AI foundation world-class so we will stick to the winning formula we had throughout the whole year.
“It’s about the apps, and they give us high-quality data. BDC was a genius move for us. Now with that high-quality data together with the LLMs, we can provide more and more value to our customers.
“And we can see it in the numbers and we see it in the pipeline — it’s working. So again, on software and cloud, we have everything we need including on the infrastructure level without building those data centers.”
Final Thought
There’s an old line that goes something like, “Hey, a billion here and a billion there, and pretty soon you’re talking real money.”
But in the world of cloud and AI infrastructure, it’s tens of billions and hundreds of billions that rapidly cascade into the trillions, and Klein — whose company is the hottest enterprise apps/agents/data company on the planet — is wisely and unconditionally flushing away any suggestion that SAP might go that route simply because Oracle has done so and Oracle also makes apps/agents/data.
Ask Cloud Wars AI Agent about this analysis
