Amazon Weighs Ditching Anthropic for OpenAI and Its Own AI After Price Hike

Amazon is weighing whether to replace Anthropic's AI models with OpenAI's GPT models and its own Nova family after Anthropic raised prices for using its models in Amazon products, according to sources cited by The Information. The move would mark a significant shift in the relationship between Amazon and one of its most important early AI backers.

Anthropic, which has grown into a leading enterprise AI provider, increased the fees it charges Amazon for embedding its Claude models into services such as Alexa and AWS. The price hike prompted Amazon to evaluate alternatives, including OpenAI's offerings and its internally developed Nova models, which Amazon has been building to reduce reliance on external vendors.

Meta's parallel AI supplier shuffle

Meta has been secretly relying on Google's AI models for customer service chatbots, ad tools, and content moderation, according to the Financial Times. Google warned Meta around March that it could not provide all the capacity the company wanted, disrupting and delaying internal AI projects. The cutoff forced Meta to scramble for alternative AI capacity.

Separately, Meta is restricting its engineers' use of Anthropic's Claude Code and OpenAI's Codex to prevent output from these rival AI tools from being incorporated into its own training data, as reported by The Decoder. The policy aims to keep Meta's models from learning from competitors' AI-generated code and responses.

• Amazon's potential switch affects its Alexa voice assistant and AWS services that currently use Anthropic's Claude.
• Meta's secret use of Google AI spanned customer service, advertising optimization, and content moderation systems.
• Google's capacity warning to Meta came around March 2026, according to the Financial Times.
• Meta's restriction on Claude Code and Codex applies to internal engineering teams working on AI training pipelines.

Implications for enterprise AI strategy

The moves by Amazon and Meta highlight a growing trend: large tech companies are treating AI models as interchangeable commodities and are willing to switch suppliers to control costs and protect proprietary data. Anthropic's price increase may have been a bet on its market power, but Amazon's willingness to walk away shows that even deep investors will shop around.

Meta's experience with Google's capacity limits underscores the risk of depending on a single AI provider. The company is now diversifying its AI infrastructure, including building its own large language models and investing in custom silicon. The restriction on Claude Code and Codex reflects a broader concern among AI developers that using rival models for training could leak competitive advantages.

What comes next: Enterprises will likely adopt multi-model strategies, mixing open-source and proprietary models to avoid vendor lock-in and manage costs. Cloud providers such as AWS, Google Cloud, and Azure will compete to offer the widest array of models, while model makers like Anthropic and OpenAI will face pressure to justify premium pricing with superior performance.