We Are Heavily Investing In AI

Do you expect your AI offerings to impact your topline and bottom line in the coming quarters? Can we expect SAP to include AI in its P&L statements, akin to cloud and software licenses?

We are infusing AI and generative AI into all our solutions. At SAP, we prioritise understanding our customers' needs and optimising business processes, such as finance, supply chain, warehouses, HR (human resource), procurement and customer experience. Our underlying business technology platform also incorporates generative AI and AI capabilities. This integration is fundamental to delivering enhanced application processes and is integral to our value proposition for customers and will impact both, our business and the customer side. We are heavily investing in generative AI, which will affect both our topline and bottom line. By offering competitive products, we expect to attract more customers and enhance business efficiency. I would be suspicious if someone told me that AI will not impact our P&L statement at all while we do all these amazing things.

You have mentioned SAP's efforts in developing its own foundation model. What distinguishes its value proposition in a landscape primarily led by giants like OpenAI, Meta and Google?

Various models excel in different aspects. For instance, GPT-4 outperforms others in numerous dimensions, but its larger size can result in slower response times. So, if immediate responses are essential for your use case, GPT-4 may not be the ideal choice.

For our own LLM, what we want to do is build the best one in the world for business AI. For SAP, it comes back to data as we concentrate on leveraging proprietary SAP data and our expertise across diverse business sectors. SAP has established best practices for 26 industries, offering solutions to companies like Mahindra, HCLTech, Deloitte and many more. Our unique data includes qualified structures and metadata, which are key elements in building our large language models. Moreover, we have obtained consent from numerous customers to anonymously train machine learning and AI models using their data. It's crucial to note that our focus is on business AI, it may not be great at passing a Law or Medical exam because that is not its purpose.

You have a lot of launches coming up in the first quarter of next fiscal year which will embed AI into many solutions. This would translate into lesser involvement of Salesforce and ServiceNow in a lot of processes. Should they be concerned?

I hope so.

SAP is set to host solutions in an India-based data centre by mid-2024 for reasons of regulatory compliance. This aligns with data sovereignty debates as well but adds costs. What's your perspective on this from business and technology angles?

As a chief technology officer and technology person, I would look to optimise a little differently on where to put workloads. India, at a certain point in time, would be one node in that network. But it's not just a technology world, right? There is community, political, governmental and academic discourse to consider. SAP is efficient at spinning up local data centres and it’s good to be globally available. But from an argumentation perspective, as the CTO of SAP, if I take this up just purely on technology terms, we would not need it. We protect customer data, no matter what. Our customers trust us and this trust is independent of where the data is stored. But we comply with every regulation.

There are three prevailing narratives on AI: one from the scientific community warning about AI dangers and advocating for regulation, and the other involving governments, with advanced economies leaning toward regulation while developing economies prioritising technology adoption without extensive regulations. On the business front, there's a pro-business stance favouring limited regulation. How do you view this complex landscape? What's your comment on possible job losses due to AI?

Right now, for example, you are recording this interview. Probably you don't need to type it. But five years ago, probably you had to type it. Are you now very sad that this part of your job is not there anymore? The absence of typing could be seen in different ways; it might lead to increased productivity, allowing more time for drafting and working on additional stories in the same timeframe. This trend of evolving job roles will persist as we move forward.

Similarly, developers' roles are evolving beyond just coding. It is a misconception that developers only code and develop. In reality, coding is only a small part of their work. They spend the majority of their time in discussions with stakeholders about various topics, including data protection and security, as well as architectural decisions. AI will handle mundane tasks, but assuming developers only code, and hence thinking about job losses is wrong. In fact, the dynamic is shifting developers towards more of a reviewer role, as AI takes on more tasks. However, critical aspects will still rely on human expertise for some time.

Predicting the future is challenging. But it's important to note that accountability can't be delegated to a large language model. Humans have distinct names, roles and accountability, reflecting their significance in the process.