Anthropic Cranks Up the Pressure on NVIDIA with New Google TPU Partnership — Unveiling the ASICs Jensen Dreads

Anthropic is making waves in the tech industry with its recent decision to bolster its computing power significantly. This move, spotlighted by the company’s partnership with Google Cloud, marks a strategic shift away from relying solely on NVIDIA and AMD technologies. By tapping into Google’s advanced AI infrastructure, Anthropic is setting the stage for a major leap forward in its capabilities.

Anthropic’s Groundbreaking Partnership with Google Cloud

Anthropic has unveiled plans to dramatically expand its computing power through a partnership with Google Cloud. This agreement includes the use of up to one million TPU chips, making it the largest deal Google has landed for its custom AI chips. Anthropic will soon have over a gigawatt of capacity, positioning it as a significant player in the industry without utilizing NVIDIA or AMD tech stacks. This collaboration highlights Anthropic’s commitment to utilizing AI-optimized infrastructure to enhance its research and development operations.

This represents the largest expansion of Anthropic’s TPU usage to date. Anthropic will have access to up to one million TPU chips, as well as additional Google Cloud services, which will empower its research and development teams with leading AI-optimized infrastructure for years to come. Anthropic chose TPUs due to their price-performance and efficiency, and the company’s existing experience in training and serving its models with TPUs.

Computer servers with interconnected cables in a data center.
Image Credits: Google

Rivalries and Industry Implications

Anthropic’s decision to align with Google Cloud over NVIDIA is not the company’s first interaction with the tech giant. They’ve been leveraging Google’s services since 2023, making this expansion a logical next step. However, the move is also fueled by Anthropic’s ongoing tension with NVIDIA, evidenced by public disagreements over AI innovations and practices. Google’s TPUs are viewed as formidable challengers to NVIDIA’s dominance. Meanwhile, NVIDIA continues to lead with significant deals from major players like OpenAI and Oracle, underscoring the competitive landscape in AI infrastructure.

To that point … one of the biggest key debates … is this question of GPUs versus ASICs, Google’s TPUs, Amazon’s Trainium. Google… They started TPU1 before everything started. … The challenge for people who are building ASICs.

In the midst of this rivalry, the true victor is the AI industry itself, which stands to benefit from the technological advancements and innovations arising from such competitive dynamics.