News Categories

Baidu and NVIDIA partner to bring AI to cloud computing, self-driving cars and smart home assistants

By Michael Low & Wong Chung Wee - on 8 Jul 2017, 4:00pm

Baidu and NVIDIA partner to bring AI to cloud computing, self-driving cars and smart home assistants

Image source: Baidu Research.

In a keynote address at this year’s Baidu AI developer conference, the company’s president and COO Qi Lu announced plans to work closely with NVIDIA. Both companies aim to bring AI to cloud computing, self-driving vehicles, and smart home assistants.

The NVIDIA Tesla V100 GPU.

For a start, Baidu Cloud has plans to use next-gen NVIDIA Volta GPUs to power its deep learning platform. The Baidu Cloud has already incorporated NVIDIA Tesla P40 GPUs and its accompanying NVIDIA deep learning framework. This new move will allow Baidu Cloud Computing enterprise customers to leverage on the upcoming Tesla Volta V100 GPUs, as well as the Tesla P4 GPU accelerators. The upgraded GPU compute platform will be deployed on NVIDIA HGX architecture. On a side note, the next-gen NVIDIA Volta GPU is slated for a Q3 launch this year.

Besides making use of NVIDIA’s deep learning framework, Baidu also aims to optimize its own open source deep learning framework, PaddlePaddle, for NVIDIA Volta GPUs, and avail its deep learning solution to academics and researchers.

The Chinese web titan has plans for end-consumers as well,  where it will adopt NVIDIA’s Drive PX platform its own open source self-driving car initiative known as Apollo. Baidu wants to collaborate with major Chinese carmakers to develop self-driving cars, and the manufacturers include Changan, Chery Automobile, FAW, and Greatwall Motor.

For the Chinese market, Baidu wants to add its Baidu DuerOS, the company’s conversational AI system, to NVIDIA Shield TV system. This move provides NVIDIA Shield TV with voice command capabilities that are tailored for the Chinese market. The localized version of Shield TV is expected to hit the China market this year.

Source: NVIDIA.