AI Weekly | September 6, 2025

All Insights

Apple and Google deepened their strategic AI collaboration this week. A U.S. court upheld Google's multibillion-dollar payments to Apple to remain the default search engine, clearing the way for Google's Gemini AI to power Siri's next upgrade. Meanwhile, NVIDIA and Cisco made major infrastructure announcements, Snowflake shares jumped 19% on AI-driven demand, and the EU's AI Act entered its first enforcement phase.

Apple and Google Team Up on AI

A federal court ruling allows Google to continue paying Apple around $20 billion a year to remain the default search engine in Safari. With this decision behind them, Apple confirmed plans to integrate Google's Gemini AI into Siri. The new feature, called "World Knowledge Answers," will deliver richer responses in text, images, and video starting in spring 2026. This marks Apple's biggest step yet in bringing advanced AI directly into its devices.

Apple and Google's alliance shows how even the biggest platforms are choosing cooperation over going it alone to stay competitive in AI. For smaller players this raises the bar, as closed ecosystems and default settings can decide who reaches billions of users.

NVIDIA and Cisco Push AI Infrastructure Forward

Germany inaugurated Jupiter, Europe's first exascale supercomputer built on NVIDIA GPUs. The system will be used for climate research, materials science, and advanced AI development. Cisco, working with NVIDIA and VAST Data, also introduced its Secure AI Factory—a framework designed to help companies deploy AI systems that are both powerful and secure.

These moves show that scaling AI is no longer just about having enough chips. Enterprises want systems they can trust with sensitive data, and they want those systems to be reliable, compliant, and cost-efficient. The focus is shifting from experiments to secure, production-ready deployments where resilience matters as much as speed.

Snowflake Rides AI Demand

Snowflake shares rose 19% after the company reported quarterly results that beat expectations. CEO Sridhar Ramaswamy said enterprise customers are running more AI workloads on its platform, particularly in regulated industries where secure data management is critical.

The results confirm that demand for AI is translating into demand for strong data platforms. Enterprises cannot train or deploy models without reliable access to governed, high-quality data. The real bottleneck for scaling AI is not always the model—it is the infrastructure that supports it.

EU AI Act Takes Effect

The EU’s AI Act entered its first phase of enforcement on 2 August 2025. General Purpose AI (GPAI Model) providers must now meet transparency and documentation requirements. The new European AI Office has the authority to audit and fine companies up to €35 million or 7% of global turnover.

Any new GPAI models placed on the market after 2 August 2025 must comply immediately. Providers of models already in use before that date have until 2 August 2027 to bring their systems into compliance.

Regulation is now real and enforceable. Global AI providers must treat compliance as a core part of strategy if they want access to Europe. Governance and legal alignment are becoming as important as model performance.

Conclusion

Start with three practical moves: review where your organization depends on external platforms and infrastructure providers for AI capabilities; take stock of your regulatory exposure under the EU AI Act, especially if you are using or planning to use General Purpose AI models; and stress-test the data layer that feeds your models for cost, governance, and latency.