Key Highlights
- Huawei introduced the Atlas 350 AI accelerator, featuring the Ascend 950PR processor
- The company asserts the Atlas 350 achieves 1.56 petaflops in FP4 computing — 2.8x superior to Nvidia’s H20 chip
- Jensen Huang, Nvidia’s CEO, announced renewed H200 chip production for the Chinese market last week
- KeyBanc projects potential $30 billion revenue for Nvidia from approximately 1.5 million H200 unit sales in China this year
- Nvidia shares gained 1.6% in premarket Monday but closed the session down 3.28%, continuing range-bound trading
Huawei has introduced a cutting-edge AI accelerator card that reportedly surpasses Nvidia’s top-performing chip authorized for Chinese sales. The announcement arrives with strategic timing — just days after Nvidia announced it would resume H200 chip production targeting the Chinese marketplace.
Huawei Unveils 950PR AI Chip…delivering performance 2.87x that of the H20
At the Ascend AI Partner Summit held during the "Huawei China Partner Conference 2026" on March 20, Ma Haixu, Huawei Vice President and President of ICT Portfolio Management & Solutions, officially… pic.twitter.com/kUOl5F1uFN
— Jukan (@jukan05) March 21, 2026
The Atlas 350 represents an accelerator solution centered on Huawei’s proprietary Ascend 950PR processor. Its primary function targets AI inference operations — the computational process that generates outputs from pre-trained artificial intelligence models. This encompasses applications including search engine recommendations, responses from large language models, and generation of multimodal content.
At Huawei’s China Partner Conference held Friday, Vice President Ma Haixu presented the new accelerator card. Zhang Dixuan, who leads Huawei’s Ascend computing division, highlighted the standout specification: 1.56 petaflops of FP4 computational capability.
This represents a 2.8-fold performance advantage compared to Nvidia’s H20 chip — currently the highest-performance processor Nvidia can legally distribute in China under existing US export control regulations.
Nvidia’s Chinese Market Strategy Faces Fresh Obstacles
Last week, Nvidia CEO Jensen Huang revealed the company had resumed production of H200 processors intended for Chinese customers and confirmed initial orders had been received. The H200 occupies a middle position in Nvidia’s lineup — more powerful than the H20 but less advanced than the company’s latest Blackwell architecture.
Huang has previously characterized China as a $50 billion annual opportunity for AI infrastructure investments, with 50% year-over-year growth projections. According to KeyBanc analyst John Vinh’s estimates, Chinese enterprises could potentially purchase approximately 1.5 million H200 chips throughout this year if large-scale sales receive approval — translating to roughly $30 billion in prospective revenue.
That substantial revenue opportunity is critical for Nvidia, which needs momentum. The stock has struggled to break out of its current trading range and requires a positive catalyst.
Nvidia shares climbed 1.6% during Monday’s premarket session, lifted by broader market optimism following President Trump’s announcement that the US and Iran were engaged in discussions regarding a potential agreement. However, by the close of regular trading, NVDA had declined 3.28%.
Evaluating the Competitive Threat from Huawei
Direct performance comparisons between chips present inherent challenges. Various configurations and testing methodologies can yield dramatically different benchmark results, and Huawei’s earlier AI server implementations have faced scrutiny regarding elevated power consumption relative to Nvidia’s comparable solutions.
Nevertheless, having a viable domestic alternative carries significant weight in the Chinese market. Enterprises that might otherwise purchase Nvidia processors now have access to a locally-produced option — one unaffected by US export policy uncertainties and restrictions.
Huawei has been systematically working toward this milestone. The Ascend 950PR processor was initially revealed in September 2025 as a component of the company’s three-year strategic roadmap. Following US sanctions, Huawei has developed its complete AI chip technology stack independently of American technology.
Regarding storage infrastructure, Huawei announced plans for significant enhancements to its OceanStor Dorado and Pacific 9926 platforms throughout this year, alongside the upcoming launch of the FusionCube A1000 cabinet designed for small and medium enterprise AI deployment scenarios.
“While the first half of the AI era focused on computing power, the second half will be defined by data,” said Yuan Yuan, president of Huawei’s data storage product line.
Nvidia’s decision to restart H200 manufacturing for China combined with KeyBanc’s $30 billion revenue projection represent the most current concrete indicators of the stakes involved in this competitive landscape.
