VENDORiQ: Microsoft Launches First Custom AI Semiconductors

Hurray for dedicated AI chips!

The Latest

14 December 2023: Microsoft has introduced its first custom AI semiconductors, Azure Maia 100 and Cobalt 100 to decrease dependence on Nvidia and enhance its Cloud infrastructure. It will use these chips exclusively in its subscription software offerings and Azure Cloud service. The Maia 100 chip is currently tested with Bing, Office AI products, and OpenAI. The Cobalt 100 CPU, built on Arm architecture, focuses on general Cloud services on Azure. The company announced that it has no plans to sell the chips, which will debut in 2024 in some Microsoft data centres.

Why It’s Important

Dedicated AI chips are typically more energy-efficient than general-purpose processors for AI tasks. For enterprises relying on AI for data analysis, customer service automation, and other advanced applications, optimised layers of infrastructure stack are critical. The inclusion of a general-purpose chip based on Arm designs also indicates a significant upgrade in Cloud computing capabilities. 

Dedicated AI chips can also improve the performance and efficiency of Cloud services, which many organisations now rely on. Improved efficiency in data centres can also lead to reduced costs for Cloud services, a benefit that companies such as Microsoft may pass on to its customers in the future. Enterprises can also expect more reliable and advanced AI and machine learning services for faster processing speeds and reduced latency.

Who’s Impacted

  • CTO
  • IT teams

What’s Next?

Consider ways how to trim down costs and optimise your Microsoft Azure Desktop Services as outlined in the advisory paper, Microsoft Azure Desktop Services – Getting The Best Bang For Your Buck.

Related IBRS Advisory

  1. Pros and Cons of Going All-In With Microsoft
  2. VENDORiQ: Microsoft Adds ChatGPT To Azure OpenAI Service

Trouble viewing this article?

Search