Huawei moves up Ascend 960DT AI chip launch to Q1 2027, but shrinks its supercomputer cluster size

Huawei is accelerating the launch of its next Ascend AI chip by two quarters, even as its own numbers show the giant computing cluster meant to hold thousands of those chips has shrunk from its original plan.

What's new

Speaking at the Huawei Connect conference on Thursday, September 17, Huawei said it is moving up the launch of its next-generation Ascend 960DT AI chip to the first quarter of 2027. The chip had previously been planned for Q3 2027, a Huawei spokesperson told TechCrunch. "The Ascend 960DT is expected to be ready in Q1 2027. Ascend 960 chips are launching ahead of schedule, doubling performance and advancing year by year," the spokesperson said. David Wang, Huawei's rotating and acting chairman, announced the updated timeline at the conference.

Alongside the chip acceleration, Huawei detailed what it calls its Peerium Computing Architecture, an approach meant to link hundreds of thousands — and eventually millions — of AI chips into one giant computer, built on UnifiedBus, Huawei's interconnect technology for linking processors with memory, storage, and networking hardware. Eric Xu, Huawei's rotating chairman, said the company is using Peerium to build larger AI computing systems for both training and inference. Huawei's Atlas 950 SuperPoD and SuperCluster are the first systems built on this architecture; the company says an Atlas 950 SuperCluster can connect up to 256,000 accelerator cards.

Context

The accelerated chip timeline comes with a wrinkle in Huawei's broader systems claims. China tech analyst Rui Ma noted that Huawei had previously said its Atlas 960 SuperPoD would scale to 15,488 Ascend 960 chips, while this week's announcement described a system with only 4,096 chips. "The chip itself is coming WAY earlier, but the SuperPoD they announced is much smaller than what they originally laid out," Ma wrote. Huawei has continued advancing its chip ambitions despite U.S. restrictions on China's access to advanced semiconductor technology; Ma said those restrictions are unlikely to stop China from pursuing its own semiconductor technology regardless.

Why it matters

A faster chip cadence is the headline Huawei wants, and it matters: every Ascend generation that ships on schedule strengthens the case that Chinese AI labs have a viable, improving alternative to Nvidia hardware, which is exactly what U.S. export controls are trying to prevent. But the shrunken SuperPoD figure is a reminder that a faster individual chip and a working system built from thousands of those chips are different engineering problems, and Huawei's own numbers suggest the systems side is lagging the chip side. For anyone tracking whether China can close the AI compute gap on its own timeline, both halves of that story matter, and this week's announcement only confirms one of them.