Baidu and Samsung Move New AI Chip Into Mass Production

Samsung and Baidu have officially completed the development phase of their new AI accelerator and are now preparing to move the chip into mass production. Dubbed the Baidu KUNLUN, the chip blends Baidu’s proprietary cloud, edge and AI XPU with 14-nanometer Interposer-Cube (I-Cube) technology from Samsung.

The result is a chip with a memory speed of 512 gigabytes per second that can support up to 260 tera operations in the same window. The Baidu KUNLUN is a 150-watt component, and can be deployed for a wide range of AI applications.

According to the two companies, the chip is particularly well-suited to natural language processing, and is much faster than more conventional technologies. However, the tech can just as easily be used for speech recognition, image processing, or self-driving vehicles.  

“We’re expanding our business beyond mobile to datacenter applications by developing and mass-producing AI chips,” said Samsung Foundry Marketing VP Ryan Lee. “Samsung will provide comprehensive foundry solutions from design support to cutting-edge manufacturing technologies.”

“Thanks to Samsung’s process technologies and competent foundry services, we were able to meet and surpass our goal to offer superior AI user experience,” added Baidu Distinguished Architect OuYang Jian.

The KUNLUN extends Baidu’s interest in conversational AI technology. Baidu is especially strong in the realms of speech recognition and comprehension, and just released the Xiaodu Smart Display X8, a new smart speaker that was built specifically for the Chinese market. The company’s DuerOS voice AI platform had already appeared in more than 100 million devices as of August of 2018.

Samsung, meanwhile, recently took home an astounding 46 CES Innovation Awards. The prizes were handed out in a slew of different categories, though the most notable is probably the Best of Innovation award given to the company’s Galaxy Note10+ 5G smartphone.

The tech giant also began supporting Google’s biometric API on its Galaxy S9 and Note9 iris scanners this month.