Qualcomm takes wrap off AI-powered processors for edge computing

Chipmaker Qualcomm has unveiled a new processor that introduces artificial intelligence expertise to the cloud, as it looks to diversify beyond the mobile-phone market.

The Qualcomm Cloud AI 100 seeks to meet increased demand for AI inference processing in the cloud, the company said in a statement.

The new chip utilises the company's technology in advanced signal processing and power efficiency while facilitating distributed intelligence from the cloud to the customers' end as well as all other points in between, the company said.

Qualcomm will begin testing the new chip with partners such as Microsoft later this year and is likely to roll out the Cloud AI 100 to customers in the second half of this year.

Qualcomm, which has a large share in the market for mobile-phone chips with its Snapdragon portfolio, said the AI software will work better than any competitors' combination of central processing units and graphics processing units of data centres. The AI will infer the data on cloud and process on its own without external (human) directions.

“Today, we bring AI acceleration to over a billion client devices. Qualcomm Cloud AI 100 accelerator will significantly raise the bar for the AI inference processing. We are well positioned to support complete cloud-to-edge AI solutions all connected with high-speed and low-latency 5G connectivity,” said Keith Kressin, senior vice president, product management at Qualcomm.

According to the company, it was supporting developers with a full stack of tools and frameworks for its cloud-to-edge AI solutions. It added that the distributed AI model will help enhance the experience for the end-users, including personal assistants for natural language processing and translations, advanced image search, and personalised content and recommendations.

Edge computing refers to processing data near the edge of a device or network, where the data is being generated, instead of sending the data to cloud and then waiting for some other machine to infer the data.

Qualcomm said the Cloud AI 100 works on a highly efficient chip specifically designed for the purpose apart from supporting machine learning frameworks such as Facebook’s PyTorch and Google’s TensorFlow.

“Microsoft’s vision of cloud-to-edge AI emphasises the benefits of distributed intelligence and our collaboration with Qualcomm continues in this effort,” said Venky Veeraraghavan, partner group programme manager at Microsoft Azure.