Cerebras Systems, provider of the fastest artificial intelligence (AI) accelerator, has won the 2023 SEMI Award for North America for process and technology integration at SEMICON West. SEMI honored Cerebras for developing the world’s largest integrated circuit chip for complex artificial intelligence (AI) computation applications training with very large AI databases as a single data core, enabling game-changing performance.
To achieve this milestone, Cerebras Systems, based in Sunnyvale, Calif., implemented wafer-scale integration with advanced packaging and system design for AI and other deep-learning applications as standalone units and in clusters for large-scale data centers. The chip’s integrated logic and memory cores contain 2.6 trillion transistors and 40 gigabytes of on-chip memory with an active area of 46,225 mm2 fabricated on 300mm silicon wafers using TSMC’s 7nm complementary metal-oxide semiconductor (CMOS) technology. The Cerebras Wafer-Scale Engines (WSE-2) are deployed in stand-alone units of the company’s CS-2 Artificial Intelligence system and incorporated as accelerators in large-scale AI computing centers.
“We are honored to be recognized by SEMI for our work in revolutionizing AI compute for generative AI,” said Dhiraj Mallick, SVP, Hardware Engineering and Operations, Cerebras Systems. “Our Wafer-Scale Engine, the industry’s largest AI processor which powers our CS-2 systems, enables the AI community to train truly large-scale models efficiently and easily.”
The Cerebras WSE-2 is the largest processor ever built. It is 56 times larger than the largest GPU, has 123 times more cores, 1,000 times more on-chip memory, 12,000 times more memory bandwidth, and 45,000 times more fabric bandwidth. The WSE-2 is the size of a dinner plate, while the largest graphic processing unit is the size of postage stamp.
“For more than 40 years, the SEMI Award for North America has recognized transformative semiconductor innovations, many of them enabled by miniaturizing or shrinking integrated circuits,” said Joe Stockunas, President of SEMI Americas. “This year’s award shifts direction to recognize Cerebras Systems for developing and commercializing the world’s largest integrated circuits to enhance AI applications, one of the fastest growing markets enabled by semiconductors this decade.”
“AI’s ever-growing hunger for computing power demands uncanny innovation in chip technology,” said Witek Maszara, a member a SEMI Award for North America Committee. “Cerebras stepped up to deliver.”
Microprocessors have made enormous advancements over their 50-year history: transistors shrank by 1,500x, the clock frequency rose by 200x and operations-per-second by an astounding 1013. Yet the physical size of the largest chips has not kept pace, rising a mere 50x due to limitations with lithography and manufacturing yield. In 2015, Cerebras Systems embarked on an effort to increase die size, in five years matching the progress of the entire industry over the previous 50 years. The Cerebras WSE, introduced in 2019, overcame both the lithography and yield limits and delivered a computing system using wafer-scale integration to dozens of commercial customers.
The highest honor conferred by the SEMI Americas region, the SEMI Award for North America, was established in 1979 to recognize outstanding technical achievement and meritorious contributions by individuals and teams in the areas of Semiconductor Materials, Wafer Fabrication, Assembly and Packaging, Process Control, Test and Inspection, Robotics and Automation, Quality Enhancement, and Process Integration.
Nominations are open to individuals or teams from industry and academia whose accomplishments have a broad commercial impact and technical significance for the semiconductor industry. SEMI invites 2024 SEMI Award for North America nominations from individuals or teams with North American-based SEMI member companies. See list of past award recipients.
 are deployed in stand-alone units of the company’s CS-2 Artificial Intelligence system and incorporated as accelerators in large-scale AI computing centers.){kind=link}