Microsoft is taking steps to increase battery life and potential performance gains in AI hardware by developing a new network card specifically designed for its Maia AI server chip, The Information reports.
The move will help improve the performance of Maia AI server chips and reduce the company's reliance on chip designer NVIDIA, helping to improve the Azure cloud experience.
The company's CEO Satya Nadella has appointed Pradeep Sindhu, co-founder of networking equipment developer Juniper Networks, to lead the network card effort, the report said.
Microsoft acquired Sindhu's data processing startup, Fungible, last year.
The acquisition is expected to strengthen the software giant's data center infrastructure, and the expertise and technology gained from Fungible can be used in the development of new network cards.
The report adds that the new network card is similar to NVIDIA's ConnectX-7 card, which the chip developer sells with a graphics processor.
The device, which could take more than a year to develop, could shorten the time it takes OpenAI to train its models through Microsoft's servers. Additionally, process costs can be reduced and Azure capabilities improved, especially through Azure Boost performance. .
Microsoft has invested billions of dollars in OpenAI, the developer of ChatGPT, an AI-powered chatbot.
The software giant has integrated OpenAI technology into a wide range of products, giving it a head start in the race to sell AI software.
In November, the company launched Maia, a chip designed to run large language models and support AI calculations.
This is in line with the software giant's overall strategy to integrate OpenAI technology into a variety of products.
Microsoft noted that the Sora model is coming to Copilot, strengthening its position in the highly competitive AI software market while providing powerful AI services via Azure.
The development of the network card shows Microsoft's ambition to become self-sufficient in AI hardware.
The project's success could reshape the competitive landscape between Microsoft and NVIDIA and have implications for the AI industry as a whole.