Particle.news

Huawei Accelerates Ascend 960 Roadmap to Press Nvidia in AI Compute

The company plans to link thousands of processors with new optics and interconnect tech to make up for single‑chip gaps while prioritizing Chinese buyers.

Visitors past by the Huawei booth display of the Atlas 950 SuperPoD during the World AI Conference in Shanghai, China on Friday, July 17, 2026. (AP Photo/Ng Han Guan)
People walk past the Huawei booth, where the Atlas 950 SuperPoD system is displayed, during the World Artificial Intelligence Conference (WAIC) in Shanghai, China, July 17, 2026. REUTERS/Go Nakamura/File Photo
The Huawei logo at the World Artificial Intelligence Conference (WAIC) in Shanghai, China, July 17, 2026. REUTERS/Go Nakamura/File Photo
FILE -Visitors stand near the Huawei Atlas 950 SuperPod during the World AI Conference in Shanghai, Friday, July 17, 2026. (AP Photo/Ng Han Guan, File)

Overview

  • Huawei announced Thursday at its Huawei Connect event in Shanghai that the Ascend 960DT launch has been pulled forward to Q1 2027 and the Ascend 960PR is set for Q3 2027.
  • The firm unveiled the Atlas 960 SuperPoD and promoted UnifiedBus and Peerium interconnect architectures designed to stitch many processors into large SuperPod and SuperCluster systems.
  • Huawei says its supernode deployments exceed 1,000 units to more than 370 customers and that domestic demand already outstrips its production capacity, so it will prioritize China over major overseas expansion.
  • Company materials and briefings highlight optics (near‑package optics) and unified memory fabrics as the performance levers that, together with scale, aim to offset per‑chip disadvantages versus Nvidia.
  • Despite the hardware roadmap and growing developer engagement, Huawei still faces Nvidia’s entrenched software ecosystem, U.S. export limits and constrained memory and packaging supplies that could slow wider parity.