streugut.ch
Archive

The direction

Where the AI race is heading

AI competition is moving away from frontier models alone toward control of the stack that turns models into governed execution. That shift is being carried by agent environments tied to enterprise identity, data, workflow, and security, and by model access that is increasingly routed through managed clouds, compatibility layers, and owned product surfaces. The infrastructure race underneath is widening into financing, datacenter capacity, and energy planning. What is particular now is that this operating layer is no longer just being assembled; it is landing in institution-scale deployments while regulation starts to shape where and how AI can ship.

The move

What actors actually did

Google secures massive NVIDIA GPU deal, Anthropic receives $35 billion chip financing, Microsoft expands enterprise agentic AI

The most striking move this week came from Google, which agreed to pay SpaceX $920 million per month starting this October for access to about 110,000 NVIDIA GPUs hosted at the Colossus 1 data center. This arrangement gives Google unmatched compute resources and could significantly boost its cloud and AI capabilities, affecting both developers and enterprise customers looking for scalable AI platforms.

Anthropic also gained ground: Apollo Global Management and Blackstone finalized a $35 billion financing deal to enable Anthropic’s purchase of custom Google AI chips and further expansion of its computing capacity. This financial backing positions Anthropic to accelerate both model development and deployment, opening the door for more rapid competition with leading cloud and model providers.

Microsoft advanced its position in enterprise AI, rolling out Microsoft Agent 365 and Copilot across over 276,000 KPMG professionals globally and deepening its collaboration with Atos Group for secure, large-scale AI agent deployment. Meanwhile, AWS made Anthropic’s Claude Fable 5 model available to all Amazon Bedrock customers, and Google Cloud rolled it out for enterprise clients, ensuring competitive access to advanced generative models across platforms.

Sources