The news: Google and Anthropic inked a cloud partnership that could put Amazon Web Services (AWS) on notice and raise the stakes for the stability and services Anthropic will offer enterprise and consumer customers in the future. The deal gives Anthropic access to up to 1 million of Google’s custom Tensor Processing Units (TPUs), per CNBC.

Amazon will remain Anthropic’s most deeply integrated partner and provide support from its custom Trainium hardware.

Zooming out: Amazon suffered one of its worst-ever outages October 20, degrading and disrupting performance for the nearly one-third of the web that uses AWS for cloud storage. This spotlights the importance of using multiple providers to prevent downtime and raises doubts about major platforms relying on single vendors.

Moving markets: Anthropic’s move toward Google Cloud—even while AWS remains a key partner—marks a broader unraveling of cloud loyalty among leading AI firms. As custom chips and compute capacity become essential for scaling models, access to reliable and high-performance infrastructure is taking precedence over long-standing partnerships.

Why it matters: The success of future AI tools could depend on diversified infrastructure providers to remain resilient against downtime and leverage the benefits of various platforms. Marketers should pay attention because:

• Faster AI could translate into faster marketing: Anthropic’s access to Google’s TPUs could result in smarter AI models that can provide better personalization, analytics, and creative optimization.
• Marketing organizations need to think beyond genAI content creation and build long-term strategies for complex applications. A million chips aren’t just for experimentation—they’re for large-scale execution.

Next steps for brands: To remain resilient against outages and capitalize on the fast-moving development of AI models, companies should:

• Diversify AI bets. Avoid reliance on tools that are tied to a single cloud or model provider by building flexibility into your tech stack.
• Shift from output to intelligence. Use AI to power decision-making—like market insights, campaign planning, and spending optimization—not just as an asset creator.
• Ask vendors to explain infrastructure choices. Clarify with AI partners how infrastructure decisions affect speed, reliability, and scale, especially if genAI tools will be consumer-facing.