Huawei's new AI chip wins over ByteDance and Alibaba with CUDA compatibility breakthrough

Exclusive: Huawei's new AI chip find favour with ByteDance, Alibaba which plan to place orders, sources say

March 27 (Reuters) - Customer testing of Huawei's new AI chip, designed to challenge Nvidia (NVDA.O), opens new tab in the China market, has gone well and big tech giants including ByteDance and Alibaba plan to place orders, two people familiar with the matter said. The development marks a milestone for Huawei (HWT.UL). Despite a government campaign to encourage the use of domestic semiconductors, the Shenzhen-based firm struggled to persuade big tech firms in the private sector to adopt its current flagship chip, the Ascend 910C, in large quantities, industry sources have previously said. This time around, tech firms intend to use the new 950PR more extensively, much happier now that the ⁠chip is more compatible with Nvidia's CUDA software system and has better response speeds, said the two people and a third person with knowledge of those plans. Huawei plans to ship around 750,000 950PRs this year, according to two of the people. They said samples were sent to customers in January, and that mass production should begin next month, setting the stage for fully fledged shipments to start in the second half of the year. The sources were not authorised to speak to media and declined to be identified. Huawei, ByteDance, Alibaba (9988.HK), opens new tab did not reply to Reuters requests for comments. RESTRICTIONS ON NVIDIA CHIPS A launch of the 950PR comes at a difficult time for Nvidia in China. Many of its artificial intelligence chips have been banned from sale in China by Washington on worries that the technology could boost the capabilities ⁠of the Chinese military. The Trump administration last year greenlighted the sale of Nvidia's H200 chips - more powerful than currently restricted products - albeit with a number of conditions that could limit amounts sold. The H200 has also recently received approval from Chinese authorities, but it remains unclear when they will be allowed into the country. Huawei mentioned its new chip last September when it outlined its long-term semiconductor plans for the first time and said it would be launching ⁠some of the world's most powerful computing systems. The 950PR, which uses traditional DDR memory, will be priced at around 50,000 yuan ($6,900) per card, while a premium version with faster HBM memory will sell for around 70,000 yuan, the sources said. Where previously Huawei had stuck to its proprietary CANN software system, the ⁠new chips will allow developers at Chinese tech firms, which have generally used Nvidia's software system thus far, to migrate those models more easily. The sources said that compared to the 910C, the chip only offers a small improvement in raw computing power, but it is ⁠designed to excel in handling inference workloads - the process of running trained AI models to answer queries or execute tasks. Demand for AI inference computing in China is surging as the country's tech sector shifts its focus from model development to real-world deployment, a trend turbocharged by the rapid adoption of open-source AI agent OpenClaw. Reporting by Reuters staff; Editing by Miyoung Kim and Edwina Gibbs Our Standards: The Thomson Reuters Trust Principles., opens new tab

Huawei's new AI chip finds favor with ByteDance, Alibaba which plan to place orders, Reuters reports

Customer testing of Huawei's new AI chip, designed to challenge Nvidia in the China market, has gone well and big tech giants, including ByteDance and Alibaba plan to place orders, two people familiar with the matter said. The development marks a milestone for Huawei. Despite a government campaign to encourage the use of domestic semiconductors, the Shenzhen-based firm struggled to persuade big tech firms in the private sector to adopt its current flagship chip, the Ascend 910C, in large quantities, industry sources have previously said. This time around, tech firms intend to use the new 950PR more extensively, much happier now that the chip is more compatible with Nvidia's CUDA software system and has better response speeds, said the two people and a third person with knowledge of those plans. Huawei plans to ship around 750,000 950PRs this year, according to two of the people. They said samples were sent to customers in January, and that mass production should begin next month, setting the stage for fully fledged shipments to start in the second half of the year. The sources were not authorized to speak to media and declined to be identified. Huawei, ByteDance, Alibaba did not reply to Reuters requests for comments.

Huawei's new AI chip finds favour with ByteDance, Alibaba which plan to place orders - The Economic Times

Despite a government campaign to encourage ⁠the use ⁠of domestic semiconductors, the Shenzhen-based firm struggled to persuade big tech firms in the private sector to adopt its current flagship chip, the Ascend 910C, in large quantities, industry sources have previously told Reuters.Customer testing of Huawei's new AI chip, designed to challenge Nvidia in the China market, has gone well and big tech giants including ByteDance and Alibaba plan to place orders, two people familiar with the matter said. The development marks a milestone for Huawei. Despite a government campaign to encourage ⁠the use ⁠of domestic semiconductors, the Shenzhen-based firm struggled to persuade big tech firms in the private sector to adopt its current flagship chip, the Ascend 910C, in large quantities, industry sources have previously said. This time around, tech firms intend to use the new 950PR more extensively, much happier now that the chip is more compatible with Nvidia's CUDA software system and has better response speeds, said the two people and a third person with knowledge of those plans. Huawei plans to ship around 750,000 950PRs this year, according to two of the people. They said samples were sent to customers in January, and that ⁠mass production should begin next month, setting the stage for fully fledged shipments to start in the second half of the year. The sources were not authorised to speak to media and declined to be identified. Huawei, ByteDance, ⁠Alibaba did not reply to Reuters requests for comments. Restrictions on Nvidia chips A launch of the 950PR comes at a difficult time for Nvidia in China. Many of its artificial intelligence chips have been banned from sale in China by Washington on worries that the technology could boost the capabilities of the Chinese military. The Trump administration last year greenlighted the sale of Nvidia's H200 chips - more powerful than currently restricted products - albeit with a number of conditions that could limit amounts sold. The H200 has also recently received approval from Chinese authorities, but it remains unclear when they will be allowed into the country. Huawei mentioned its new chip last September when it outlined its long-term semiconductor plans for the first time and said it would be launching some of the world's most powerful computing systems. The 950PR, which uses traditional DDR memory, will be priced at around 50,000 yuan ($6,900) per card, while a ⁠premium version with faster HBM memory will sell for around 70,000 yuan, the sources said. Where previously Huawei had stuck to its proprietary CANN software system, the new chips will allow developers at Chinese tech firms, which have generally used Nvidia's software system thus far, to migrate those models more easily. The sources said that compared to the 910C, the chip only offers a small improvement in raw computing power, but it is designed to excel in handling inference workloads - the process of running trained AI models to answer queries or execute tasks. Demand for AI inference computing in China is surging as the country's tech sector shifts its focus from model development to real-world deployment, a trend turbocharged by the rapid adoption of open-source AI agent OpenClaw.

Huawei's Ascend 950PR AI Chip Just Won Over Chinese Customers By Mimicking CUDA Through CANN Next, Threatening NVIDIA's Moat

Huawei's newest AI chip, the Ascend 950PR, might not deliver strong compute performance relative to NVIDIA for domestic hyperscalers, but it offers a major upgrade with CUDA compatibility. The Chinese computing industry has been trying to challenge NVIDIA's market dominance, and while the focus has been on upgrading offerings in terms of architecture and onboard features, it hasn't worked out to a large extent. Reports suggest that Chinese hyperscalers remain strongly inclined toward NVIDIA's hardware, and a key reason isn't just the compute gap; CUDA also plays a significant role. Huawei has tried to 'crack' CUDA with its native CANN offering, but that hasn't worked out yet, which is why, with the Ascend 950PR, the idea is to be a direct replacement for NVIDIA in training/inference workloads. This time around, tech firms intend to use the new 950PR more extensively, much happier now that ⁠the chip is more compatible with Nvidia's CUDA software system and has better response speeds, said the two people and a third person with knowledge of those plans. - Reuters We'll dive into what the Ascend 950PR chip brings to the table in a bit, but let's talk about CUDA compatibility and Huawei's major achievement with this launch. Huawei's CANN Next software stack has undergone a major upgrade, adding a SIMT programming model with features such as thread blocks, warps, and kernel launches, similar to CUDA. The idea with CANN Next isn't to provide developers with a translation layer; it's to bring in near-drop-in replacements for CUDA equivalents, treating CUDA as a language standard while leveraging the strengths of the Ascend ecosystem. CANN Next is optimized for compute on Ascend at scale, meaning parameters such as thread counts and block sizes are tuned for Huawei's own chips, enabling co-design scalability. For a layman to understand what Huawei is actually doing, it isn't to replace CUDA at all; rather, it's to make developers feel like they are writing in CUDA, but in reality, the performance achieved with GPU programming is Ascend-optimized and scalable. CANN Next is one of the reasons the Ascend 950PR is seen as a much more attractive solution than previous offerings. Now, with the Ascend 950PR chip in particular, it is reported that hyperscalers like ByteDance and Alibaba plan to place orders soon, and that the firm is set to produce 750,000 chips this year. In terms of technicals, you are looking at support for low-precision data formats, up to FP8, with 1 PFLOPS of FP8 compute and 2 PFLOPS of FP4. The chip will be equipped with an interconnect bandwidth of 2 TB/s, with the firm's first "self-built HBM," called HiBL 1.0, featuring a capacity of 128GB and a bandwidth of 1.6 TB/s. The HBM technology ensures that Huawei won't face constraints in ramping up production either. China has been in need of alternatives to NVIDIA's compute offerings, particularly for hyperscalers. Getting involved in the regulatory overhead of sourcing chips like the H200 has been a 'pain', which is why they have resorted to options like renting compute offshore or looking towards domestic options. Huawei, with CANN Next and Ascend 950PR, is looking to step up its influence within the Chinese AI industry, yet the only constraints holding it back are chip volume and whether customers are ready for mass deployment. About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.