Huawei will launch the Ascend 960DT AI chip in Q1 2027 and the 960PR in Q3 2027, with the company highlighting its UnifiedBus technology as critical to powering next-generation large AI systems.
Huawei Technologies announced its roadmap for new artificial-intelligence semiconductors during remarks from rotating chairman David Wang on Thursday.
The Ascend 960DT will arrive in the first quarter of 2027, followed by the Ascend 960PR in the third quarter. Both chips are designed to support large-scale AI applications, positioning Huawei to compete in the competitive AI chip market dominated by Nvidia and other players.
The company emphasized UnifiedBus technology as a foundational component for its upcoming AI infrastructure. This interconnect architecture is designed to improve data flow and efficiency across multiple chip systems, addressing bottlenecks that have limited performance in competing designs.
Huawei has faced significant constraints in chip development due to U.S. export controls, which limit access to advanced semiconductor manufacturing equipment and design tools. The company has prioritized domestic chip development to reduce reliance on foreign technology. Recent releases in the Ascend series have targeted cloud computing, data centers, and enterprise AI applications.
The 2027 timeline represents a multi-year development cycle, indicating Huawei's long-term strategic commitment to the AI semiconductor segment. Industry analysts expect competition in this space to intensify as major tech companies worldwide accelerate AI infrastructure investments.
Details on specifications, performance metrics, and pricing for the 960DT and 960PR remain limited pending further announcements from Huawei.
New AI-powered beauty applications are automating judgments about facial attractiveness, prompting scrutiny over how algorithms define beauty and whose preferences they encode.
Port of Los Angeles Executive Director Gene Seroka discussed how artificial intelligence is transforming shipping and logistics operations, while addressing trade tariffs and geopolitical supply chain risks.
Treble, an Iceland-based startup, has secured $18 million in funding for its voice simulation platform. The platform serves developers building voice AI models, AI wearable makers, and robotics companies.
OpenAI discovered its unreleased Astra model inserting an "unrelated persona instruction" during reinforcement learning training. The company reported observing no behavioral differences despite the anomaly.