Tech heavyweights Fujitsu and Huawei Technologies reportedly plan to throw their hats into the burgeoning AI processor market, taking on the established leaders Nvidia, Intel, and Advanced Micro Devices.
Fujitsu plans to roll out a microchip dubbed the deep learning unit (DLU), according to the Top500 site. The Japanese firm expects that the chip will offer 10 times the performance per watt compared to competitors’ offerings. Fujitsu said it plans to initially release the DLU next year in the form of a coprocessor working alongside with a CPU. Later, the company plans to integrate the DLU directly into a central processing unit (CPU).
Huawei CEO Yu Chengdong announced his company is developing an artificial-intelligence processor to be introduced later this year, according to an article appearing in eWeek. The Chinese firm’s chip is expected to combine a CPU, a graphics processing unit (GPU) and AI processing capabilities on a single piece of silicon.
The introduction of these devices would give the two companies an opportunity to cash in on a market for AI-oriented chips that is expected to expand at a compound annual growth rate of more than 60 percent from 2016 to 2022, according to a forecast from the market intelligence firm Research and Markets. Global revenue for AI processor chips is expected to amount to more than $16 billion five years from now.
U.S. semiconductor makers have been the dominant players in the early phases of the AI processor business.
Nvidia of Santa Clara, California established a strong position in the market largely because of serendipity. The company’s GPU architecture was originally designed to allow devices like computers and video-game consoles to produce higher-quality graphics. However, the highly parallel microarchitecture of the company’s GPUs proved to be well-suited to the fast and efficient processing of artificial-intelligence tasks.
Fellow Santa Clara-based chip supplier Intel has been making headway in the AI chip segment with its efforts to integrate the technology into its processor line. The company in the second quarter announced its Xeon Scalable processors, which deliver improved performance when performing AI workloads compared to previous chips in its product line. The company partly credited AI for its strong second-quarter performance.
Advanced Micro Devices — based in the neighboring city of Sunnyvale, California — is offering GPUs that can accelerate deep-learning tasks.
These companies represent some of the most prominent names in the semiconductor world. Because of this, Fujitsu and Huawei will face major challenges in creating differentiated products and carving out market share.
However, with products of all descriptions increasingly integrating AI capabilities, these firms may have little choice but to join the artificial intelligence processor party.
Stephan Cunningham is vice president, product management at Veritone. Working in concert with core internal teams including industry-specific general managers and engineering as well as directly with clients and prospects, he leads the disciplines and business processes which govern the Veritone Platform.