News Hub

NVIDIA set to develop new business unit for custom AI chips

Written by Mon 12 Feb 2024

Image Credit: Reuters

NVIDIA is building a new business unit to design custom chips for cloud computing firms, including advanced artificial intelligence (AI) processors.

Reuters reported NVIDIA aims to capture a portion of the market for custom AI chips, citing nine people familiar with the plans. Sources added the chipmaker intends to protect itself from the rising number of companies seeking alternatives to its products.

NVIDIA currently controls 80% of the high-end AI chip market. This position has increased its stock market value by 40% this year to £1.37 trillion ($1.73 trillion) after it more than tripled in 2023.

NVIDIA’s users include ChatGPT creator, OpenAI, Microsoft, Alphabet, and Meta who have rushed to secure the limited supply of chips to compete in the growing Generative AI sector.

NVIDIA’s H100 and A100 chips act as versatile AI processors for users. However, these tech firms are now creating internal chips tailored to their specific requirements. This strategy aims to cut energy usage, and possibly minimise design costs and time.

A general partner at Eclipse Ventures, Greg Reichow, highlighted the importance of matching compute resources to specific application needs.

“If you are really trying to optimise on things like power, or optimise on cost for your application, you can’t afford to go drop an H100 or A100 in there,” said Reichow.

He added that companies who are trying to optimise power or cost for an application cannot afford to ‘go drop an H100 or A100 in there’.

Reuters’ sources said NVIDIA is trying to assist companies in creating custom AI chips which have also been adopted by competitors like Broadcom and Marvell technology.

NVIDIA has not revealed the cost of the H100 chips, which surpassed that of the A100. Reuters reported each chip can sell for £12,678 ($16,000) to £79,240 ($100,000) depending on the volume of chips among other factors that were not disclosed. Meta plans to increase its total stock of NVIDIA’s H100 to 350,000 this year.

NVIDIA officials have met with Amazon, Meta, Microsoft, Google, and OpenAI to discuss creating custom chips for them, according to sources. 

Tech Firms Ramp up AI Chip Commitments

Earlier this month, Meta said it plans to deploy a new version of a custom chip in its data centres this year. This move is intended to strengthen its AI initiatives, according to an internal document seen by Reuters.

In January, OpenAI CEO, Sam Altman, said the company intended to utilise funds raised for a chip venture to establish a network of semiconductor manufacturing factories. A month before, OpenAI said it was exploring making its own AI chips and has explored a potential acquisition target.

In November, Microsoft unveiled two custom-designed silicon AI chips, created to address the growing demand for sustainable computing power. 

The first chip called the Maia 100 AI Accelerator will power some of the largest internal AI workloads running on Microsoft Azure. It was designed specifically for the Azure hardware stack. The second chip titled the Microsoft Azure Cobalt CPU is an Arm-based processor chip tailored to run general-purpose compute workloads on the Microsoft Cloud.  

Join Big Data & AI World

6-7 March 2024, ExCeL London

Be at the forefront of change with thousands of technologists, data specialists, and AI pioneers.

Don’t miss the biggest opportunities to advance your business into the future.

Written by Mon 12 Feb 2024

Send us a correction Send us a news tip