SAP CEO: Europe Doesn't Need More Data Centers

Key Concepts

AI use cases, infrastructure layer (chips, data centers), software/AI layer, European competitiveness, AI Act, regulation, sovereignty (data location, access, encryption), hyperscalers, demand-driven development.

AI Development Focus for Europe

The speaker argues that Europe's priority in AI should be on developing AI use cases for industries like life science, logistics, and manufacturing, rather than focusing solely on building infrastructure (data centers and chips). The innovation on the infrastructure and LLM layers has already happened, and now is the time to develop new innovation on the software/AI layer.

• Main Point: The AI race is not decided on the software layer, and Europe should focus on developing AI use cases.
• Counterargument: Jensen Huang (Nvidia CEO) advocates for building data centers and filling them with chips in Europe.
• Speaker's Rebuttal: While infrastructure is necessary, demand for chips in Europe isn't as high as in the US (where companies are training large language models). Building data centers without sufficient demand is not the optimal approach.
• Demand-Driven Development: AI application development will drive demand for chips and data centers. The focus should be on the software/AI layer first.

The EU AI Act and Regulation

The speaker expresses concern that the EU AI Act and regulatory environment may hinder AI development in Europe.

• Problem: The speaker notes that discussions in Brussels often start with regulation instead of competitiveness and opportunities.
• Ideal Scenario: Regulation should follow leadership in AI development.
• Regulation Concerns: While not against regulation, the speaker criticizes the fragmented implementation of the AI Act across member states.
• Fragmented Implementation: Instead of a unified framework, member states maintain their existing policies, creating overlapping and inconsistent regulations.
• Impact on Startups: This fragmented approach makes it difficult for startups to scale their technology across Europe due to varying interpretations of the AI Act in each country.
• Call for a Unified Framework: Europe needs a single, clear framework for AI regulation, not multiple interpretations.

Sovereignty and Dependence on US Technology

The speaker addresses the issue of European technological sovereignty and dependence on US technology.

• Data Center Example: SAP's data center in Germany is presented as an example of European cloud sovereignty.
• Hardware Reality: Acknowledges that hardware components (servers, etc.) may come from the US or other countries.
• Redefining Sovereignty: The speaker argues that sovereignty is not about the origin of hardware but about data location, access control, and encryption.
• SAP's Approach: SAP offers data encryption and control over data location to ensure sovereignty for its customers.
• Hyperscalers and Choice: Customers have the choice to use US hyperscalers, and SAP partners with them.
• Hyperscaler Benefits: US hyperscalers provide infrastructure, automation, and system reliability (SLAs).
• Workload Flexibility: Customers can shift workloads to SAP's own data centers if desired, reducing dependence on US hyperscalers.

Conclusion

Europe should prioritize developing AI use cases and applications to drive demand for infrastructure. The EU AI Act needs a unified framework to avoid hindering innovation and scalability. Sovereignty should be defined by data control and security, not necessarily the origin of hardware components.