[JPMorgan: Microsoft Maia 300 inference performance improves by 40%, cost reduction with deployment of one million chips] DeepTech TechFlow reports, according to Trend Research, JPMorgan's October 12, 2026 research note states that Microsoft Maia 300 delivers 40% higher tokens per megawatt compared to leading GPU systems. With a full configuration of 96 accelerators, rack-level token throughput is seven times that of Maia 200. Maia 200 delivers 1.7 times more tokens per watt compared to GPU-based infrastructure. Reuters reported that Microsoft plans to deploy over 1 million Maia 300 chips by 2027. Microsoft's capital expenditure for fiscal year 2027 is estimated at approximately $210 billion, and around $175 billion for the 2026 calendar year. JPMorgan predicts that Maia becoming the primary inference system will enhance Copilot performance and reduce the cost per compute unit, potentially improving Microsoft's cloud gross margin. The firm maintains an overweight rating with a target price of $625.