Microsoft AI Updates: April 3, 2026

1. Microsoft Commits $10 Billion to Japan AI Infrastructure Over Four Years

Microsoft. Microsoft announced on April 3, 2026 that it will invest $10 billion in Japan between 2026 and 2029 to build AI data centers and related infrastructure. The company is partnering with Sakura Internet and SoftBank to host GPU capacity domestically, addressing data sovereignty requirements from Japanese enterprise and government customers. Microsoft also announced agreements with NTT Data, NEC, Fujitsu, and Hitachi to train 1 million AI professionals in Japan by 2030, and plans to support development of domestic large language models. The investment follows Microsoft’s $1 billion Thailand commitment announced in late March and reflects ongoing hyperscaler competition to lock up AI infrastructure positions across Asia-Pacific. Source

2. Microsoft Reveals Plan to Develop In-House Frontier AI Models by 2027

Microsoft. Microsoft’s AI unit announced on April 2, 2026 a push to build large, cutting-edge AI models in-house, aiming to reach state-of-the-art performance across text, image, and audio modalities by 2027. The initiative represents a strategic hedge against Microsoft’s dependence on OpenAI, whose models underpin Copilot and Azure AI services. As part of the effort, Microsoft released a new speech transcription model that outperforms competing products on benchmark testing across 11 of the 25 most widely spoken languages — a first-party capability that the company says is already more accurate than rivals at launch. Source