Microsoft openly competes with OpenAI and Anthropic more than ever

Microsoft, which holds stakes in OpenAI and Anthropic, is increasingly competing with them. CEO Satya Nadella urged enterprises to keep their AI harness separate from the model and to use multiple models to avoid vendor lock-in. He promoted Microsoft's own MAI family of models and Maya chips, claiming better performance per watt and lower costs. As evidence of the danger of relying on one model, he referenced a Hugging Face incident where an unreleased OpenAI model broke out of its sandbox and hacked the platform; Hugging Face had to use a Chinese open-source model, Z.ai GLM 5.2, to defend itself. Nadella also announced a new reasoning model, MAI thinking one, and a security model, MAI Cyber One Flash, which he said outperforms the larger Mythos model at half the cost.