AI Industry’s Next Dataflow Processor to be Based on TSMC 7nm and Power Future Wave Deep Learning Systems
CAMPBELL, Calif., August 1, 2018 – Wave Computing®, the Silicon Valley start-up that is revolutionizing artificial intelligence (AI) and deep learning from the cloud to the edge with its dataflow-based systems and licensable IP solutions, today announced a strategic collaboration with Broadcom Inc. (NASDAQ:AVGO). The companies will collaborate to bring Wave’s next-generation dataflow processing unit (DPU) to market at the leading-edge 7nm process node. The next generation DPU, which will continue to be the foundation of the company’s revolutionary AI system solutions, will be manufactured on TSMC’s 7nm process technology and leverage Broadcom’s industry-leading design platform, productization skills, and proven 7nm 56Gbps and 112Gbps SerDes that are ideally suited for high-performance deep learning applications.
Leveraging Broadcom’s best-in-class semiconductor design and implementation skills, experience in packaging and testing, and established 7nm physical Intellectual Property (IP) platform, Wave will extend its industry-leading offering of faster, more intelligent dataflow-based deep learning systems with minimal risk to schedule and cost.
Wave Computing’s next generation DPUs will combine the company’s revolutionary dataflow architecture with its MIPS® 64-bit multi-threaded core technology and TSMC’s 7nm manufacturing process to accelerate the performance and power efficiency of its AI solutions for edge applications. In addition, the processors will incorporate Broadcom’s advanced physical IP to further enhance Wave’s AI systems performance and capacity, which are critical for large, machine learning datasets.
“We are really excited about the ASIC collaboration with Wave Computing. The exploding complexity of AI workloads and neural networks requires state-of-the-art engineering at the most advanced process node,” said Frank Ostojic, senior vice president and general manager of Broadcom’s ASIC Products Division. “By combining Wave Computing’s dataflow technology on our silicon-proven 7nm design capabilities, companies across the AI industry will be able to accelerate innovation from the Cloud to IoT applications.”
“The combination of dataflow technology and 7nm process geometries should be an attractive platform for the next generation of demanding AI applications,” said Karl Freund, senior analyst at Moor Insights & Strategy. “By tapping into Broadcom’s expertise in design for advanced node manufacturing, Wave is well positioned for future AI technologies.”
“This strategic collaboration with Broadcom strengthens both companies’ competitive leadership position in AI and enhances our compelling roadmap that will drive AI acceleration for commercial systems in 2020 and beyond,” said Chris Nicol, CTO of Wave Computing. “As the industry’s first AI system provider to announce plans to develop technology on the 7nm manufacturing process node, Wave will be able to leverage Broadcom’s proven success at 7nm to provide our customers with silicon and system solutions that enable them to realize the full potential of AI.”
About Wave Computing
Wave Computing, Inc. is the Silicon Valley company that is revolutionizing AI with its dataflow-based architecture, systems and solutions. The company’s vision is to “follow the data” and bring deep learning to customers’ data wherever it may be—from the Datacenter to the Edge of the Cloud. Wave is powering AI at the edge by combining its dataflow architecture with its MIPS embedded RISC multi-threaded CPU cores and IP. Offering its solutions to customers globally, Wave Computing has been named Frost & Sullivan’s 2018 “Machine Learning Industry Technology Innovation Leader,” and has been recognized by CIO Application Magazine’s as one of the “Top 25 Artificial Intelligence Providers.” Combined with MIPS, Wave now has over 425 granted and pending patents.