Good News For Alibaba - ACCEL Chip Outshines Nvidia's AI Chips, China Claims

Chinese scientists, in collaboration with China’s Semiconductor Manufacturing International Corporation, have developed a new chip that outperforms current high-performance artificial intelligence chips, such as those produced by Nvidia Corp NVDA, in tasks like image recognition and autonomous driving. 

This innovative chip, known as the All-Analogue Chip Combining Electronics and Light (ACCEL), is light-based and utilizes photons for computing and data transmission, offering faster speeds and higher energy efficiency.

Also Read: Alibaba Elevates AI War with Advanced Open-Source Models

While the ACCEL chip isn’t ready to replace the chips in today’s computers or smartphones, it holds immense potential for integration into wearable devices, electric cars, and smart factories, SCMP reports. 

This innovation positions China to boost its competitiveness in the mass application of AI, a crucial development as the country seeks to navigate the technological restrictions imposed by the U.S. The sanctions will likely cost Nvidia orders worth $5 billion from Alibaba Group Holding Limited BABA. 

The Chinese e-commerce juggernaut recently showcased the latest version of its AI large language model, designed to train AI applications like chatbots.

The concept of a light-based chip is not novel, but the ACCEL chip stands out as it relies solely on photons for calculations, a feat that has been challenging due to the complexities of controlling light particles. 

In laboratory tests, the ACCEL chip achieved a computing speed of 4.6 PFLOPS, making it 3,000 times faster than one of the most widely used commercial AI chips, Nvidia’s A100. 

Furthermore, the ACCEL chip’s energy consumption is 4 million times lower than that of the A100.

SMIC manufactured the ACCEL chip using a cost-effective and outdated 20-year-old transistor fabrication process. 

Price Action: BABA shares traded higher by 1.27% at $83.53 on the last check Thursday.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.