Brazil’s antitrust authority CADE has decided to continue its investigation into the partnership between Amazon and AI developer Anthropic.
The Administrative Council for Economic Defense of Brazil (CADE) has postponed a ruling on the cooperation between Amazon and Anthropic, saying that the agency requires additional inquiries into the transaction and recent developments that have emerged publicly.
The case concerns the companies’ partnership in generative artificial intelligence, which they significantly expanded in April 2026. The new 10-year agreement provides Anthropic with access to up to 5 gigawatts of cloud capacity to train and deploy its Claude models. Total infrastructure spending tied to the project could exceed $100 billion.
In addition, Amazon will invest another $5 billion in Anthropic on top of the $8 billion already committed and retains the option to increase funding by up to $20 billion. Under the agreement, Anthropic will also gain access to Amazon’s proprietary AI chips, including Trainium2.
CADE launched investigations in 2024 into partnerships between major technology companies — Amazon, Google and Microsoft — and artificial intelligence developers. The regulator is examining whether such deals could circumvent antitrust oversight or raise competition concerns.
Amazon maintains that the agreement does not meet the requirements for mandatory notification under Brazilian law. The company says the transaction will enable the parties to better compete at various levels of the AI supply chain and provide Anthropic with the necessary funding to continue developing responsible AI systems. Amazon argues that the deal will inject more competition into the generative AI market, increase customers’ access to foundation models and provide an alternative to already well-positioned industry leaders such as Microsoft, Meta (banned and designated as extremist in Russia), Google and OpenAI.
Source: MLex
