By Annette Ikponmwonba | October 12, 2026
Keypoints
• Microsoft plans to deploy its in-house Maia artificial intelligence chips on a large scale from 2027.
• The Maia 300 chip is expected to deliver 40 per cent more AI computing work per unit of electricity than leading systems.
• Microsoft says its existing Maia 200 chip is already operational and supports services including Microsoft 365 Copilot.
• The company aims to reduce its dependence on Nvidia as competition in AI infrastructure intensifies.
• The chips are designed to improve computing performance and energy efficiency as demand for AI services grows.
Main Story
Microsoft has announced plans to deploy its in-house Maia artificial intelligence chips on a massive scale from 2027, in a move that could reduce its reliance on Nvidia for the computing infrastructure behind its AI services, The company said Monday that its Maia chips would become a major part of the hardware used to run artificial intelligence across its products. The planned deployment is expected to consume several gigawatts of electricity, comparable to the output of several large nuclear reactors.
AI chips are essential components of the data centres that power artificial intelligence systems. While Nvidia has maintained a dominant position in chips used to train AI models, technology companies are increasingly competing in inference, the process through which trained models respond to questions and perform tasks, The market is expanding as businesses introduce AI assistants and autonomous agents capable of completing multi-step jobs. Nvidia, AMD, Amazon and Google are among the companies developing products to compete in the sector.
“Today marks a new phase for Microsoft’s AI infrastructure,” Microsoft executive vice president Scott Guthrie wrote in a blog post announcing the plans, Microsoft said its Maia 200 chip is already operational, including in support of the Microsoft 365 Copilot assistant. Its more powerful successor, Maia 300, is scheduled to launch in 2027.
The Maia 200 supports models from OpenAI, Microsoft’s own MAI models and leading open-source models, According to Microsoft, the chip has been deployed in the United States, Europe and Australia. The company says it delivers up to 1.7 times more output per watt than leading GPU-based systems for AI features integrated into Teams and Copilot.
Microsoft also expects the Maia 300 to perform 40 per cent more AI work using the same amount of electricity as leading systems, and more than twice as much as its previous chip, Energy efficiency has become a major consideration for technology companies expanding their AI infrastructure. The growing electricity demands of data centres are placing pressure on power grids and increasing operating costs.
Microsoft said each server rack would accommodate 96 Maia chips, twice the number used previously, and handle up to seven times the workload, The chip’s computing components will be manufactured by Taiwan Semiconductor Manufacturing Company, commonly known as TSMC.
In a separate post, Microsoft engineers said they had focused on reducing the electricity wasted when data moves between components inside the machines, an increasingly important challenge as AI systems handle more complex tasks, The company added that its systems would primarily use liquid cooling but could also operate in older data centres that rely on air cooling.
The Issues
Microsoft’s investment in its own AI chips reflects the growing competition to control the infrastructure required to operate artificial intelligence services, Reducing dependence on external chip suppliers could give the company greater control over computing capacity, costs and hardware development. However, large-scale deployment also requires substantial investment in data centres, electricity supply and cooling systems, Energy efficiency remains a critical issue as demand for AI computing continues to increase, putting pressure on technology companies to deliver more processing power without a proportional rise in energy consumption.
What’s Being Said
Microsoft executive vice president Scott Guthrie described the announcement as a new phase in the company’s AI infrastructure strategy, The company says its Maia 200 chip is already supporting AI services and that the Maia 300 will offer further improvements in computing efficiency, Microsoft engineers have also highlighted the importance of reducing power consumption associated with moving data within AI systems.
What’s Next
Microsoft plans to begin large-scale deployment of its Maia chips from 2027, with the Maia 300 expected to launch that year, The company will need to expand its computing infrastructure and demonstrate that its chips can deliver the expected performance and energy-efficiency improvements at scale.
Bottom Line
Microsoft is stepping up efforts to develop its own AI computing hardware as competition in the sector intensifies. Its Maia chips are designed to improve efficiency and strengthen the company’s control over the infrastructure powering its AI services, potentially reducing its dependence on Nvidia.
