- Huawei promises annual AI chip leaps while rivals still follow slower development cycles
- Nvidia now faces a rival accelerating infrastructure expansion
- Huawei already operates large computing clusters supporting millions of connected vehicles
On June 5, Huawei Vice President Chen Lin spoke at the 2026 Huawei Cloud INSPIRE Innovators Conference, where every eye was turned toward one announcement — the Ascend 950DT chip, arriving on Huawei Cloud later this year.
The 950DT carries upgraded vector computing power, wider memory bandwidth, and native support for low-precision formats like FP8.
According to Chen, the chip is simpler to program and better suited to intelligent driving than anything before it, but what Chen said next deserves far more scrutiny than the chip itself – especially for rivals such as Nvidia
One generation per year, computing power doubled every single time
“The Ascend chip is evolving at a rate of ‘one generation per year, doubling the computing power,” Chen stated, without qualification or hedging.
That is a public commitment to a release cadence aggressive enough to challenge how AI chip progress gets measured.
Nvidia has long controlled that pace, with each new architecture raising the bar for every competitor chasing it – and a rival locking in annual generational jumps — publicly, on a stage — is not behaving like a company still catching up.
Whether Huawei can maintain that pace without advanced Western lithography tools remains a fair and open question.
The announced cadence only carries weight because there is a genuine infrastructure sitting behind it.
Huawei Cloud has deployed large-scale computing clusters across Gui’an, Wuhu, and Inner Mongolia, with a global network covering 34 regions and 102 availability zones.
Over 100,000 Ascend computing units currently support continuous algorithm iteration for paying customers through Huawei Cloud.
Every day, more than two million intelligent driving vehicles and 60 million connected vehicles run stably on that same infrastructure. Those are operational numbers, not projections from a roadmap slide.
More than 30 automotive OEMs and suppliers have built deep partnerships with Huawei Cloud across intelligent driving and intelligent manufacturing.
That growing customer base absorbs each new chip generation as it lands, giving Huawei a live proving ground that sharpens every subsequent release.
The shift from usable to genuinely easy to use
Huawei’s comment moves beyond hardware specifications and into something harder to counter quickly.
Chen emphasized that systems engineering capabilities are equally decisive as raw computing power in helping automakers improve intelligent driving training efficiency.
Through its Lingqu architecture, Huawei Cloud achieves high-speed interconnection within supernodes, meaningfully improving training efficiency at scale.
Its AI DataLake platform supports the production of hundreds of thousands of data clips every single day.
Huawei Cloud has also worked directly with leading intelligent driving manufacturers across the full model iteration cycle — from computing power integration to algorithm adaptation and optimization.
That level of deep involvement transforms Huawei from a chip supplier into an embedded infrastructure partner.
The stated ambition, in Chen’s own words, is moving from chips that are merely “usable” to a full stack that is genuinely “easy to use.”
Via Guancha (Originally in Chinese)
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
https://cdn.mos.cms.futurecdn.net/RrdrrDsT7TJfFtEZCdUmti-1920-80.png
Source link
