Chinese telecommunications and technology enterprise Huawei Technologies is preparing to launch two new artificial intelligence semiconductors designed to strengthen its enterprise computing ecosystem and directly challenge US hardware maker Nvidia. Huawei rotating chairman David Wang outlined the product roadmap during an industry event held in Shanghai. The company intends to release the 960DT semiconductor during the first quarter of 2027, followed by the rollout of the Ascend 960PR during the third quarter of 2027. Both processors are positioned as central pillars in expanding the firm's domestic computing infrastructure.
Interconnection Architecture and Unified Bus Infrastructure
Modern machine learning models and high-intensity workloads frequently require processing capabilities far beyond what an individual semiconductor can deliver. To address this structural constraint, Huawei is focusing heavily on interconnect technologies that link vast numbers of processors into unified high-performance systems. Wang explained that unified bus technology will serve as the core backbone for next-generation computing deployments, allowing individual silicon units to exchange data rapidly and collaborate with maximum efficiency.
Huawei has already engineered 11 distinct semiconductors built upon this interconnected standard. The enterprise's largest networked deployments can interconnect up to one million separate processors into integrated installations designated as superclusters, creating massive computational scale for industrial-grade artificial intelligence training.
Supernode Deployments and Enterprise Customer Scale
The company has also reached notable milestones in enterprise hardware delivery. Huawei has delivered more than 1,000 unified supernode systems across a base of over 370 enterprise clients. A supernode functions as a clustered computing apparatus combining numerous interconnected accelerators into a cohesive operational unit. Wang did not disclose granular specifics regarding individual chip counts within each supernode configuration, customer identities, or commercial shipment financial values.
Export Controls and China Domestic Hardware Strategy
The intensifying rivalry between Washington and Beijing over artificial intelligence data centers, advanced silicon, and specialized software continues to reshape enterprise supply chains. Strict export curbs imposed by Washington have restricted Chinese technological access to top-tier accelerators and wafer fabrication equipment from international vendors. This regulatory landscape has driven Chinese companies to prioritize domestic alternatives, making multi-chip scaling and clustered computing vital strategies to offset limitations in standalone processing power.
While trade restrictions have positioned Huawei as the leading domestic AI chip supplier within China, overcoming the extensive developer ecosystem established by Nvidia remains a persistent competitive hurdle. Huawei currently counts 5,270 monthly active developers across its proprietary accelerator ecosystem, a critical technical community tasked with refining software accessibility and integrating applications with Huawei silicon architectures.
