4 hours ago
31
Huawei just gave itself less time to catch up with Nvidia — and that’s exactly the point. According to the company’s announcement, the launch of its next-generation Ascend 960DT AI chip has been moved up to the first quarter of 2027, two full quarters ahead of the original schedule. The Huawei Ascend 960DT launch, unveiled at the company’s Huawei Connect conference on Thursday, signals just how aggressively the Chinese tech giant intends to close the gap with Nvidia in the race for AI computing power.
Key takeaways
- Huawei moved the Ascend 960DT chip launch up to Q1 2027, from a previously planned Q3 2027 date.
- The company says Ascend 960 chips are arriving ahead of schedule while doubling performance year over year.
- Huawei’s Peerium Computing Architecture links hundreds of thousands of AI chips into single systems, with an Atlas 950 SuperCluster reportedly connecting up to 256,000 accelerator cards.
- A discrepancy has emerged between Huawei’s earlier claim of a 15,488-chip Atlas 960 SuperPoD and this week’s reference to a 4,096-chip system.
- The announcement lands just a week before U.S. President Trump and Chinese President Xi Jinping are due to meet in Washington, DC.
Huawei Accelerates Ascend 960DT AI Chip Launch
Huawei is shortening its own product runway to get next-generation AI silicon into the market faster. A Huawei spokesperson told TechCrunch that the chip had previously been slated for a Q3 2027 release, meaning the new timeline shaves roughly two quarters off the original schedule.
New launch timeline and performance improvements
“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 Huawei spokesperson said. That framing matters: Huawei isn’t just moving up a release date, it’s also claiming that each generation of the chip roughly doubles in performance, a pace the company is using to argue it can keep narrowing the distance with Nvidia’s roadmap.
Announcement details at Huawei Connect
David Wang, one of Huawei’s rotating and acting chairmen, announced the updated timeline during Thursday’s Huawei Connect event. The venue itself is notable — Huawei Connect has become the company’s main stage for signaling how fast its chip division intends to move, and this year’s announcement was designed to show momentum rather than caution.
Peerium Computing Architecture and AI System Scale
Behind the Huawei Ascend 960DT launch sits a bigger ambition: turning hundreds of thousands, and eventually millions, of individual AI chips into one unified computing system. Huawei calls this its Peerium Computing Architecture, and it’s the engineering backbone the company is betting on to compete with Nvidia’s large-scale AI clusters.
Technology enabling large AI computing systems
The architecture relies on UnifiedBus, Huawei’s proprietary technology for linking processors with memory, storage, and networking hardware. In the company’s statement, Eric Xu, one of Huawei’s rotating chairmen, explained that the firm is leveraging its new Peerium Computing Architecture to construct larger AI computing systems capable of handling both training and inference — the two workloads at the heart of competition in the AI chip sector.
Atlas 950 SuperCluster capacity and scale discrepancies
Huawei’s Atlas 950 SuperPoD and SuperCluster are described as the first systems built on this architecture, with the company saying an Atlas 950 SuperCluster can connect up to 256,000 accelerator cards. That figure alone would put Huawei in the same conversation as the largest AI clusters being built anywhere.
But the scale claims don’t fully line up. On X, China tech analyst Rui Ma noted that Huawei had earlier claimed its Atlas 960 SuperPoD would scale up to 15,488 Ascend 960 chips, yet the announcement this week described a system built around 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. In practice, this suggests Huawei may be trading system scale for speed — accelerating the chip’s arrival while dialing back, at least for now, how large a single interconnected cluster will be at launch.
Geopolitical Context and Strategic Implications
Timing is doing a lot of work here. The Huawei Ascend 960DT launch news arrives just a week before Trump and Xi Jinping are scheduled to meet on September 24 in Washington, DC — a summit where semiconductor policy is almost certain to be part of the backdrop.
US-China summit and semiconductor restrictions
Huawei has kept advancing its chip ambitions despite U.S. restrictions on China’s access to advanced semiconductor technology. Ma argued those restrictions are unlikely to actually stop China’s progress: “I think it’s futile to stop China’s development in semiconductors because the stakes for self-sufficiency are just too high at this point,” she wrote. That’s a notable read from a China tech analyst, and it frames the accelerated timeline less as a surprise and more as a predictable response to sustained export controls.
Huawei’s drive for AI chip self-sufficiency
Why this matters beyond one product date: if Huawei can keep compressing its release schedule while U.S. controls remain in place, it undercuts the core assumption behind those restrictions — that limiting access to advanced tools would meaningfully slow Chinese chipmakers. Whether the doubling performance claims hold up at commercial scale is a separate question, but the strategic signal is aimed squarely at showing self-sufficiency is achievable on a faster clock than Washington may have expected.
Statements from Chinese leadership on AI development pace
According to the Financial Times, Eric Xu contended that China must speed up its AI development to close the gap with the U.S., asserting that Chinese firms need to progress further before they can properly grasp and mitigate the dangers posed by increasingly powerful AI systems. That view runs in parallel with Trump’s own stance: he has pushed back against calls from AI industry leaders to slow AI development over safety concerns, arguing the U.S. needs to protect its lead over China. Both governments, in other words, are leaning toward speed over caution — just from opposite starting points in the same race.
FAQ
When will Huawei launch its Ascend 960DT AI chip?
Huawei plans to launch the Ascend 960DT AI chip in the first quarter of 2027, advanced from a previous Q3 2027 target.
What is the Peerium Computing Architecture?
Peerium Computing Architecture is Huawei’s technology to interconnect processors with memory, storage, and networking hardware for building large AI computing systems.
How large is Huawei’s Atlas 950 SuperCluster?
Huawei’s Atlas 950 SuperCluster can connect up to 256,000 accelerator cards.
How do US semiconductor restrictions affect Huawei’s AI chip development?
Despite US restrictions on China’s access to advanced semiconductors, Huawei continues advancing its chip ambitions, with China tech analyst Rui Ma saying the restrictions are unlikely to stop China’s semiconductor development.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
English (US) ·