Microsoft’s Latest Chips Aim to Slash High Tech Expenses

Microsoft’s Chips to Revolutionize AI Services and Cut Costs

Microsoft has unveiled two custom-designed computing chips, Maia and Cobalt. Microsoft wants to bring core technologies in-house and offset the high costs of delivering artificial intelligence (AI) services and this decision was a part of their strategy.

Maia was introduced at the Ignite developer conference. It is designed to accelerate AI computing tasks, particularly for large language models powering Microsoft's Azure OpenAI service. The chip will be utilized for the company's subscription software offerings and its Azure cloud computing service. This is expected to reinforce performance and reduce costs.

Microsoft is not planning to sell Maia chips for now. They are aiming to leverage them for its internal services. The expenses associated with AI services can be ten times higher than traditional services and the company has set goals to address this problem. By using foundational AI models and optimizing the Maia chip for their AI efforts, Microsoft aims to provide faster, lower-cost, and higher-quality solutions to its customers.

The second chip, Cobalt, is a central processing unit (CPU) designed to be both a cost-saving measure and a response to Amazon Web Services (AWS). Microsoft intends to use Cobalt internally for its Teams business messaging tool but also plans to offer direct access to compete with Amazon's "Graviton" series of in-house chips.

Microsoft executives emphasize that Cobalt is designed for competitiveness in terms of performance and price-to-performance compared to Amazon's offerings. Information regarding the Maia and Cobalt chips that will allow us for a detailed comparison with other existing chips, has not been provided yet but, both Maia and Cobalt are built with the help of 5-nanometer technology from Taiwan Semiconductor Manufacturing Co.

Microsoft emphasizes on using standard Ethernet network cabling for Maia, to enhance cost-effectiveness compared to more expensive custom technologies. The company aims to provide efficient and competitive AI services until personal computers and phones are powerful enough to handle them independently.