Huawei Speeds Up Rollout of Ascend 960DT AI Processor to Take on Nvidia by 2027

Huawei disclosed that its upcoming Ascend 960DT AI processor will hit the market in the first quarter of 2027, a step intended to close the performance divide between Chinese and American AI hardware makers.

Having worked on its own AI chips since the early 2020s, the firm explained that the faster schedule mirrors rising rivalry in the field and a deliberate effort to present a rival to Nvidia’s leading GPUs. The Ascend 960DT is slated for data‑center tasks such as massive model training and inference, arenas where Nvidia’s A100 and H100 currently define the benchmark.

The move comes after years of U.S. export restrictions that have curtailed the company’s ability to obtain cutting‑edge semiconductor fabrication gear and design software. To counteract this, Huawei has poured substantial resources into home‑grown R&D, collaborating with Chinese foundries to fabricate ever more advanced chips. The new processor joins a wider lineup that already features Ascend variants deployed in phones, edge equipment, and cloud platforms.

Experts point out that the schedule coincides with China’s wider drive toward AI‑computing self‑reliance. Recent state policies have allocated billions toward home‑grown chip research, and the Ascend 960DT may become a showcase of that effort. Although specific performance figures remain undisclosed, Huawei asserts the chip will provide "significant improvements" in both throughput and power efficiency compared with earlier models.

Market watchers warn that wresting market share from Nvidia will demand more than one new device. Nvidia enjoys a mature software stack, featuring CUDA libraries and a broad developer base. Huawei is pushing its own AI framework, MindSpore, urging developers to migrate models to the Ascend ecosystem, though adoption is likely to be incremental.

Going forward, Huawei intends to embed the Ascend 960DT within its cloud services and possibly place the silicon in partner data‑center products. While pricing and post‑Q1‑2027 availability remain undisclosed, the statement underscores a firm resolve to contend globally. How the chip is received will probably affect the speed of China’s wider AI‑hardware goals and may alter the competitive landscape alongside incumbents like Nvidia, AMD and Intel.