Anthropic Challenges NVIDIA with New Google TPU Partnership — Integrating ASICs That Jensen Likely Dreads
Anthropic’s recent partnership with Google Cloud signals a strategic move in the competitive landscape of AI computing, particularly given the ongoing tensions between Anthropic and NVIDIA. This collaboration comes as Jensen Huang, CEO of NVIDIA, acknowledges Google’s Tensor Processing Units (TPUs) as formidable competitors to traditional GPU architectures.
Anthropic’s Milestone Agreement with Google Cloud: A Major Adoption of TPUs
In a significant announcement, Anthropic has revealed plans to enhance its computational capabilities through a new agreement with Google Cloud. This deal, which involves up to one million TPU chips, represents a major step forward for both companies. With access to “over a gigawatt of capacity”expected by next year, Anthropic’s focus on using Google’s custom AI chips is substantial, particularly given its divergence from NVIDIA and AMD technologies.
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.
Image Credits: Google
This partnership isn’t entirely new; Anthropic has been utilizing Google Cloud’s services since 2023. Expanding on this established relationship appears to be a logical step. However, there are intriguing dynamics at play, especially since Anthropic has frequently challenged NVIDIA and its CEO regarding various issues, including alleged AI chip exports to China. Conversely, Jensen Huang has criticized Anthropic’s models, like Claude, for their ‘closed-source’ development approach.
This situation adds another layer to the rivalry, as the advent of Application-Specific Integrated Circuits (ASICs) like Google’s TPUs presents an alternative to NVIDIA’s offerings. With Anthropic emerging as the largest external customer of TPUs, this move indicates a burgeoning competition in the AI technology sphere. Although this perspective remains speculative, Jensen Huang has expressed his views on the competitive landscape regarding GPUs and ASICs during a recent podcast appearance, highlighting the evolution of AI hardware.
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.
Despite NVIDIA’s stronghold in the compute market—engineering multi-gigawatt agreements with notable clients like OpenAI and Oracle—Anthropic’s choice to partner with Google Cloud suggests a strategic pivot, either driven by competitive tensions or a sincere belief that Google offers superior solutions. Regardless of the underlying motivations, this partnership represents a pivotal moment for innovation within the AI sector.
Leave a Reply