Betting Big: Why is Amazon investing $4 billion in Anthropic, a fledgling AI startup

Amazon is making a substantial move into the world of artificial intelligence by investing up to $4 billion in Anthropic, an AI startup. This strategic investment, announced on Monday, includes an initial injection of $1.25 billion for a minority stake in Anthropic, with the potential for further investment up to $4 billion down the line.

Amazon aims to establish a strong partnership with a prominent AI startup, similar to Microsoft’s alliance with OpenAI, the organization behind ChatGPT. As part of this partnership, Anthropic will leverage Amazon’s cloud computing platform and dedicated AI chips to develop its AI models.

In today’s AI landscape, startups are in fierce competition to secure the necessary resources, including costly chips and data centre infrastructure, to build cutting-edge AI systems, such as large language models.

Anthropic was recently valued at nearly $5 billion in a funding round earlier this year, making it a key player in the surge of AI startups raising substantial funds. Notable examples include Inflection AI, which secured $1.3 billion from Microsoft and NVIDIA, as well as Toronto-based Cohere, which received $270 million in investment from NVIDIA and others.

Amazon’s partnership with Anthropic appears to represent a shift away from Google, which invested $300 million in Anthropic last year. This transition comes just seven months after Anthropic announced its plans to use Google’s chips and cloud infrastructure for its AI model training.

For Amazon, this deal represents its latest move to capitalize on the growing excitement surrounding generative AI, a technology capable of producing human-like text and realistic images. Amazon is positioning its Trainium and Inferentia chips as viable alternatives to NVIDIA’s processors for training and running generative AI models. Anthropic’s Claude chatbot competes with ChatGPT and is already among the AI products available on AWS’s Bedrock service, enabling customers to create generative AI applications in the cloud.

Adam Selipsky, head of AWS, describes this partnership as a “significant expansion” in collaboration with Anthropic. The startup will gain access to “significant quantities” of Trainium chips to train future iterations of its foundational models.

While Microsoft’s partnership with OpenAI is exclusive and involves a substantial investment, Amazon’s approach differs. Selipsky states that Amazon’s strategy is centred on providing customers with choice and security, with exclusivity not being the goal.

According to the companies, Amazon will be the primary cloud provider for “mission-critical workloads” for Anthropic. Co-founder of Anthropic, Dario Amodei, has emphasised that this deal doesn’t change the startup’s arrangement with Google, as Google was previously described as Anthropic’s “preferred cloud provider.”

NVIDIA, a prominent chipmaker, has been a significant beneficiary of the excitement surrounding generative AI. This new deal allows AWS to showcase that NVIDIA’s chips are not the only option for AI developers. Jim Hare, a technology analyst at Gartner, suggests that Amazon’s partnership with Anthropic could help dispel the perception that Amazon has lagged behind in AI.

Anthropic was founded by a group of former OpenAI employees, led by Dario Amodei, after differences in the organization’s direction following Microsoft’s $1 billion investment in 2019. Amodei expressed confidence in the safety provisions of the Amazon deal.

The specific valuation and Amazon’s ownership stake will be determined in a subsequent funding round. Insiders familiar with the deal suggest that Amazon’s stake will be significantly smaller than the 49% ownership Microsoft sought with its investment in OpenAI.