SAP CEO: Europe Should Focus on AI Race, Not Data Centres

AI is transforming both how individual businesses and entire industry sectors operate.

Underlying this profound shift is a comprehensive change in the complexity and scale of the vast data centres needed to meet rising AI demands. 

From surging energy demands and increased cooling requirements, to the environmental implications of site selections, this new breed of data centres requires a considered, strategic approach.

According to McKinsey & Company, the demand for data centres in Europe is expected to grow to approximately 35 gigawatts (GW) by 2030, up from 10 GW today. To meet this new IT load demand, more than US$250-300bn of investment will be needed in data centre infrastructure, excluding power generation capacity.

However, Christian Klein, CEO of SAP, has pushed back on the notion that Europe needs more data centres and should, instead, focus on winning the AI software race.

Europe has to pick its battles

Speaking to Bloomberg at SAP's headquarters in Walldorf, Germany, Christian argues for a continental focus on winning the AI software race, rather than investing resources in new data centres.

“On hardware and infrastructure, I would not now try to compete with companies who do a pretty good job, predominantly in the US and China,” says Christian. 

“Where the race is not decided is who now develops the best AI use cases for our life science industry here in Europe, our logistics or our manufacturing,” he adds.

“Now, finally, we can actually develop new innovation on the AI software layer – this is what matters most for Europe.”

Christian’s sentiment about there being “not so much demand” for data centres in Europe contradicts the recent statements from NVIDIA CEO Jensen Huang, who on his recent European tour made the case for Europe needing infrastructure, to build data centres and to fill them with chips if the continent wants to compete in the AI race.

When questioned on the Nvidia CEO’s stance by Bloomberg’s Tom Mackenzie, Christian explains that the need for chips and data centres only increases when companies demand them. 

But, he argues: “There is not the same demand for these chips in Europe, like in the United States, where you have a lot of companies who need them for training large language models, etc.

“Of course, when you now apply AI, develop software, build new use cases – there will be, at a certain point, also a higher demand for chips and for data centres.”

According to Christian, European companies should only build new infrastructure when the need justifies it, rather than “just building data centres for the sake of building data centres”.

European solutions for European problems

The conversation surrounding AI and whether new data centres are needed as demand skyrockets cannot unfold without recognising a related discussion: the question of regulation.

But Christian argues for reframing the approach to the regulation debate, by first discussing possibilities rather than limitations. 

“Can we maybe first talk about competitiveness and opportunities, and how we use AI to make our industries more competitive, and when we are leading there, then ask the question of how to regulate it? 

“I’m not against regulation – as long as we’re regulating business outcomes and ensuring there is fair competition.

“The problem here in Europe is regulators have all the right strategic intent, but then they are putting these rules in place and member states keep the measures they already have – adding another layer.”

Christian believes this added layer of complexity hinders innovation and growth.  

“Imagine you are a startup and you want to scale your technology, and in every country you are going to in Europe, you hear a different story and how to adhere to any kind of AI act. 

“Europe needs one framework and not 20 different policies with 100 different interpretations of them. Scaling is not going to happen in that environment.”