Huawei’s executive says that chip processes aren’t a barrier to the company’s progress in the AI market, as the company has scaled up performance without relying solely on node upgrades.
Huawei’s Ascend Cloud Platform Is Now Shaping Up To be a Global Alternative to NVIDIA’s AI Infrastructure
The Chinese AI giant has evolved significantly in recent years in terms of increasing compute capabilities through a domestic tech stack. Not only has the firm managed to disrupt competition with NVIDIA in Chinese markets, but Huawei is actively investing in both hardware and software-level capabilities to ensure that China isn’t reliant on Western options for its computing needs. More importantly, Huawei’s Executive Director and CEO of Huawei Cloud Computing, Zhang Ping’an, has refuted the claims that the firm is held back by a lack of chip technologies onboard, as the focus has been on architectural advancements.
The chip process (such as 5nm and 7nm) is not the core. What customers really need is high-quality computing results.
The executive also claims that the company has achieved three times higher operating efficiency compared to NVIDIA’s H20 AI chip, a result of continuous advancements in areas such as architecture, system optimization, and software capabilities. Interestingly, Zhang Ping’an has revealed that Huawei’s Ascend Cloud service has managed to reach a generation or decoding of 2,400 text tokens every second, on a single card such as the Ascend 910B, with a Time-Per-Output-Token (TPOT) of less than 50ms, and the firm labels this as “high computing capabilities”.
Not only this, but Huawei also claims to be expanding its support for LLMs, such as those from DeepSeek and Kiwi AI, with the primary focus on making Ascend Cloud a mature platform. And the ambitions of the Chinese AI giant aren’t just limited to the domestic level, as the executive states that Huawei has ramped up spreading its tech stack across global customers, showing that the firm is evolving into an entity that could pose a threat to NVIDIA on the ‘sovereign AI’ front as well.
NVIDIA’s CEO, Jensen Huang, has repeatedly called out Huawei as a formidable threat to America’s tech stack and has stated that the current administration should ensure that every firm around the world has access to US AI technology. More importantly, Huawei has recently unveiled an extensive AI chip roadmap, which includes plans to develop self-built HBM and rivaling architectures such as NVIDIA’s Rubin within the upcoming years.
News Source: MyDrivers
Follow Wccftech on Google or add us as a preferred source, to get our news coverage and reviews in your feeds.
Source link
