Amazon has made a significant investment in Anthropic, contributing over $13 billion to establish a strong presence in the AI sector. Recently, the company transitioned its integration of Anthropic’s Claude model on AWS to a token-based billing system, effective from June 2026. This change has raised concerns about the cost-effectiveness of deploying advanced AI models at a large scale.
What challenges arise from pricing models? The switch to token-based billing has revealed unexpected costs for enterprises. The pricing is structured on per-million-token rates or via Claude Consumption Units, which are priced at $0.01 each. Reports indicate that some companies have seen their monthly expenses exceed $500 million, prompting CFOs to critically assess their return on investment. This shift in payment model is not isolated to Amazon and Anthropic; other major players like OpenAI have similarly adopted token-based pricing.
How does funding complicate Amazon's relationship with Anthropic? As the largest stakeholder in Anthropic, Amazon has already invested an additional $5 billion in April 2026, bringing its total commitment to $13 billion, with discussions around future investments potentially reaching $20 billion. However, complications have arisen due to recently imposed export controls on Anthropic’s advanced models by the U.S. Commerce Department. These restrictions stemmed from security concerns raised by Amazon’s leadership, adding pressure to the partnership.
What is Amazon developing independently? Aside from its substantial financial backing of Anthropic, Amazon is also making strides in creating its own foundational AI models through its AWS platform. The company plans to allocate up to 5 gigawatts of computing power by 2026 to support these efforts. Amazon's intention to compete with Anthropic and OpenAI indicates that its internal AI projects are advancing and that the company is ready to establish itself as a formidable competitor in the field.
What are the implications for investors? The challenges facing Anthropic are considerable. Export restrictions could hinder its international revenue, particularly at a time when it must validate Amazon’s large investment. If Amazon succeeds in developing its own competitive models, Anthropic risks losing its most significant distributor alongside its largest client. The geopolitical factors, including the possibility of expanded export controls affecting additional AI models, further complicate this scenario. Companies that have standardized on a single provider may find themselves unable to operate in critical global markets.