Microsoft reportedly plans to launch its own AI chip next month

The move is aimed at reducing its reliance on chips from other manufacturers, particularly Nvidia

Microsoft's AI chip is reportedly designed to run data centre servers that train and run large language models, which is the underlying technology of generative AI. Reuters

Microsoft is reportedly planning to launch its own artificial intelligence chip to reduce its reliance on Nvidia as it pushes forward with its AI ambitions, according to a media report.

The chip will be similar to Nvidia's graphics processing units and are designed to run data centre servers that train and run large language models, which is the underlying technology of generative AI, The Information reported, quoting a source.

Redmond, Washington-based Microsoft, the maker of the Windows PC operating system, is planning to unveil the chip, a “culmination of years of work”, at its Ignite conference in November, it said.

The emergence of Microsoft's plan comes shortly after Reuters reported on Thursday that OpenAI, the company behind ChatGPT, was considering making its own AI chips.

The San Francisco-based company is evaluating an acquisition target, the report said, quoting sources.

Discussions within OpenAI on how to address their reliance on expensive chips began at least last year, with options also including working more closely with other chipmakers, including Nvidia, and diversifying its suppliers, they said.

With the advent of generative AI, demand for AI chips has soared. Companies that are moving into manufacturing their own chips would be able to rein in on costs and address any shortages.

Meta Platforms, is building is developing its own silicon chip MTIA – Meta Training and Inference Accelerator – especially for its AI ambitions. Amazon, the world's biggest e-commerce marketplace, quietly began working on its own silicon chip in 2013, but has stepped up its plans in the generative AI race.

The share price of Nvidia, one of the world's leading chipmakers, soared nearly four-fold in the past 12 months as it tapped into the generative AI bonanza with powerful new semiconductor offerings – helping it join the elite, trillion-dollar market capitalisation club.

US technology company Oracle had said it was spending “billions” of dollars on chips from Nvidia, as it seeks to strengthen its position in generative AI and cloud computing. The company is also preparing to release its first generative AI services through the first quarter of 2024.

SoftBank Group's British semiconductor unit Arm, meanwhile, created a buzz with its blockbuster initial public offering in September. Its shares surged 25 per cent after the company raised $4.87 billion in its IPO, marking the largest listing of the year.

Chips specially designed for AI work are poised to generate 21 per cent more in revenue this year to reach $53.4 billion, as enterprises continue to adopt AI capabilities, Gartner said in a recent study.

The growth from last year will be largely underpinned by developments in the highly-popular generative AI segment, the US-based research firm said.

Growth in revenue is set to accelerate rising by more than a quarter to $67.2 billion next year and more than double to nearly $120 billion by 2027, it said.

In the consumer electronics market alone, the value of AI-enabled application processors used in devices will more than double annually to $1.2 billion by the end of this year, Gartner analysts estimate.