SAP, the 53-year-old German tech giant, builds software for virtually every business function, from supply chain and resource management to finance, sales, and human resources. Its products are used by over 440,000 customers worldwide, including 98 of the world’s 100 largest companies. Taken together, its client base generates over 80% of global commerce, according to the company.
At the helm sits Christian Klein, 45, who has been at SAP since 1999. “You can say it’s a big minus or a big plus when you’re spending your whole career in one company,” he says. “I started here as a student. And I still know people from back then. We have four generations here working for SAP—it’s a company of 110,000 people.”
Under Klein’s leadership, the company has accelerated its transformation into a cloud-first enterprise, with cloud revenue accounting for over half of its total revenue in the first quarter of 2025. Meanwhile, SAP is embedding AI into its core products with the goal of becoming the “#1 enterprise application and business AI company.” SAP is one of the most valuable public companies in Europe and made headlines when it took the top spot in March.
Klein spoke with TIME on June 4 about his success as a leader, how AI is changing enterprises, and the difference between power and influence.
This interview has been condensed and edited for clarity.
What have you changed your mind about since becoming sole CEO in 2020?
Our software helps to build bridges and facilitate global trade. We have multinationals in the U.S. and China, in Asia—everywhere—doing global business. Five years ago, when I came into this job, in my perspective of the world everyone was a winner. I saw democratic values in most parts of the world, and said “ah, that will never change.”
Because I’m still reasonably young and have not lived in times like this, I never thought things could change so fast—at least when it came to global trade. But here we are. I have to deal much more with all of the geopolitics than I had to, say, five years back. That definitely has changed.
Why do you think you’re good at your job?
When you’re a CEO, over the years you learn that when you believe “I have the right strategy on a nice PowerPoint—I have written it all down—the rest is just about execution,” you’re completely wrong.
Especially when you’re a European company with many stakeholders, you need to think about strategy first from the customer-perspective. Everyone says that [laughs], but you really have to make sure that you’re hitting that nail. Otherwise, you could steer the company in a completely wrong direction.
You need to make sure everyone—your employees, shareholders, the workers council, and so on—is excited, committed, and passionate about strategy and where you’re leading the company. You have to be a bridge builder: to make sure everyone is involved and understands the strategy, and everyone is moving in the same direction. Otherwise, things can fall apart very easily.
In your May CEO address, you said we can think of AI agents as “digital coworkers.” If AI agents can robustly function as digital coworkers in the near-future, why hire humans at all?
Here’s an example: we just had financial earnings at SAP. Now, AI gives me certain simulations and predictions on how the year could end, given all of the trade conflicts and the uncertainty out there in the market. Would I fully trust AI to say “this is how you should put out your financial guidance for the rest of the year?” No: I still feel we need a human being at the end of the chain who can make slight adjustments, incorporating their past experience.
Or think about selling software. When you are traveling the world, the cultures are so different. When I walk into a customer meeting in Japan, it’s different from walking into a customer meeting in Germany or the U.S. AI can give me a beautiful sales pitch or a great demo, but at the end it’s human beings who need to understand how to position it, how to emotionally talk about it, particularly across different cultures. And I don’t see an AI yet that is able to do that—at least not better than a human being.
You point to emotional connection and cultural understanding. AI is already highly-persuasive and can understand emotional nuance. A key limit is that current AI systems lose coherence over long time periods. If that changes—and the same system can run for months or years at a time—do you still think AI won’t be able to do these parts of the job?
You’re right, emotional intelligence will get better and better. No doubt about that. But at the end of the day, there needs to be someone in the company you can hold accountable. I don’t want to see SAP in the headlines, with a customer saying “I relied only on SAP AI agents to close my books or to run my supply chain, and they completely screwed it up. Despite AI doing 99% good work, it didn’t play out as expected.”
Ultimately, I’m convinced there must be some human beings still in the mix. Do I expect to need the same amount of developers, salespeople, and consultants in the future? Definitely not with the job profiles that they have today. But do I still need other jobs that are coming up—more data scientists? More people thinking about the future of the industry? Yes, absolutely.
It would be an illusion to believe AI will help and drive more productivity, but the workforce will still look the same. That will be absolutely not the case. But I also can’t imagine a workforce only with digital workers.
Can you imagine a scenario where in five years time, 90% of your workforce is gone, so you have closer to 10,000 employees than 100,000?
Oh, that is tough. In certain job profiles, I can absolutely see they can be 60% to 70% digital. In others, for example, take audit: Of course, you have policies as a company, but with every policy—for example, the E.U. Data Act, which I don’t like so much—there is always a gray zone. You ask five lawyers and five large language modules about interpretation—does this contract adhere to the E.U. Data Act?—and you get different answers. It’s like when you have issues with your back and you ask five doctors, and they come with five different root causes. These things will still exist. So in these jobs, I don’t believe that there will be only digital workers. In other jobs, I definitely see a much higher share. It really depends on the job profile.
Do you think you’ll live to see an AI system do every part of your job?
Part of it. I need to make a lot of decisions every day. They are sometimes pretty logical decisions, where you just look at the facts. But sometimes there are tough decisions you have to make using your emotional intelligence. There are certain market trends which may not be captured by the facts, but you talk to people, to other stakeholders, and you make a different decision. So I don’t believe that a CEO can be purely digital in the future. Sometimes you’re still making decisions based on your gut feeling.
What are the biggest bottlenecks to enterprise adoption of AI?
In the enterprise world, where we are setting up our agents, you need 100% accuracy. So for example Joule, our digital assistant, cannot mess around with compliance checks on travel and on sourcing, or on directing the flow of materials. People are betting their jobs and their companies on our software and on AI. This needs 100% accuracy: if you as a tech company don’t understand the business process—if you don’t have the data or you can’t access all of it—that is a big issue.
This is a big obstacle for many companies: understanding how to apply the technology. The good position of SAP is: we are running these business processes, we know the rules and workflows, we have the data. Others who are more on the infrastructure and hardware layer… they don’t have the business context. They’re missing the data.
Is accuracy the biggest challenge? Or are there others?
The second piece is on data. Every company you walk into has their data siloes: there have been trends with collecting data and creating data lakes, but no one has solved the problem of making all the data match. And when it doesn’t fit, AI can’t do magic immediately to say, “I 100% understand how this data fits together and I can correlate it to produce good results for the company.”
The third piece involves regulation, which often kills innovation before it gets started. Certain parts of the world need to be careful to not only see risk, risk, risk with regard to AI, but also the upside for the economy.
What do you think is the appropriate regulatory framework for AI?
Here’s my pragmatic view. In the European Union—it’s good that we have a union, I’m all in for it—we have AI regulation in many member states, and then the E.U. puts another regulation on top. The result is confusion, different interpretations, and before companies or startups can use the technology to race against others in the world, it’s already game over. That is the problem. I’d say: have one framework for all of Europe, and then give some freedom within this framework, especially when you are early in the development and testing cycle—you cannot do harm in this early phase.
Of course, the moment when you bring it to market and to scale, there must be regulation. But don’t regulate the technology! Regulate the outcome, so AI is unfolding in the right way in the chemical industry, the automotive industry, the defense industry. But don’t regulate the technology, because then you regulate technological innovation, which is never good.
You need to see you are not living on an island here in Europe. All of these tech players—we are the only large tech player in Europe, but there are many startups—there is competition everywhere, and we cannot give these companies and startups a disadvantage when it comes to speed of innovation just by over-regulating.
If you were 22 today, fresh out of university, what would you do?
At 22, I still wanted to become a professional skier. I would try it again. It’s my passion. I love to be in the mountains, I love to ski, and I’d try to turn that passion into a profession.
So you wouldn’t set out to become a CEO?
I wasn’t planning to become the CEO when I was 22 years old. That goal developed over time. I don’t like it when you’re too early on, saying “I need to be the CEO.” I’m more on the “first deliver” side: prove yourself, prove that you can work in and deliver great results as a team, and the rest will follow.
It was only when I became the chief operating officer of SAP, and I considered our transformation into a cloud company, that I developed the goal of becoming CEO. We had a strategy, and the software was instrumental for that. But I saw it was not only a piece of technology which would make transformation work. It’s also an understanding of the culture, and the tone from the top. You need to understand: where do you want to go with your company? Do you want to be a cloud-pure SaaS company, or do you want to still be a legacy? What does it take? Then you can connect software and technology and AI to it.
So it was only around 2017 that I thought, “oh, I could be the CEO of SAP. I have a vision for this company on how to move it into the next century. It probably sounds a little bit odd, but it’s not the power and the responsibility that drew me to the role. It was about: “you can influence a lot of things to create a great future for SAP.” I saw how we worked and what was needed.
What do you think distinguishes power and influence?
Becoming a CEO and believing that now you’re making a decision, and you have the power, so everyone will just follow, is probably the biggest mistake you can make. You can put a lot of policies in place, you can put more pressure, but people will not just automatically follow. You need to over-communicate in times of change to convince people. When we did this drastic change and our share price collapsed five years back, I couldn’t just say, “Oh, now we did it. The strategy is clear. I have the power now to tell you exactly what to do.” You need to influence people. You need to convince them.
(To receive weekly emails of conversations with the world’s top CEOs and decisionmakers, click here.)