Keep pulling the thread on Mustafa Suleyman.
Microsoft AI expects the compute used for training frontier AI models to increase by a thousand-fold over the next three years.
Microsoft AI has announced a family of seven new models developed in-house.
Microsoft AI models will be available for developers on OpenRouter, Fireworks, and Baseten.
Developers using Microsoft AI models will be able to tune the model weights themselves for the first time.
Microsoft AI's models are built using clean, traceable, enterprise-grade datasets and do not use distillation from other labs.
A Microsoft AI model tuned for Excel using Frontier Tuning matches the performance of GPT 5.4 while being up to 10 times more efficient.
Microsoft and Mayo Clinic are collaborating to co-create a frontier AI model for healthcare.
The AI model developed by Microsoft and Mayo Clinic will be made available to other organizations via Microsoft Foundry after it is validated.
Microsoft achieves a 1.4x efficiency boost by co-designing its AI systems with its own Maia 200 silicon.
Microsoft's strategy of in-house model and silicon development is aimed at long-term self-sufficiency for the company and its partners.
Microsoft AI plans a rapid scale-up in its compute and AI capabilities over the next year.
Microsoft AI's next-generation GB200 cluster is now operational.