Huawei Accelerates AI Chip Production to Challenge Nvidia’s Market Position in China

Key Highlights

• Domestic demand for Huawei’s AI processors significantly exceeds current production capabilities in China
• Testing phase underway for Ascend 950DT chip, with Chinese AI developers set to begin model training in 2026
• Advanced Ascend 960DT processor timeline accelerated by nine months, now slated for Q1 2027 release
• Peerium computing architecture introduced, designed to interconnect up to 1 million processors in unified systems
• Company asserts its Ascend processor line has captured dominant China market position versus Nvidia

Huawei has revealed that production capacity for its artificial intelligence chips falls short of domestic Chinese demand, effectively limiting any significant international expansion plans. The company’s rotating chairman, Eric Xu, shared these insights during Thursday’s annual Huawei Connect conference held in Shanghai.

While acknowledging limited exports to select international markets with strong demand, Xu emphasized that shipment volumes remain minimal. The company’s immediate priority centers on satisfying the substantial domestic market requirements.

Next-Generation Ascend 960DT Timeline Moved Forward

In a significant development, Huawei has accelerated its Ascend 960DT chip launch to Q1 2027, representing a nine-month advancement from the initial roadmap. The subsequent Ascend 960PR model will debut in Q3 2027, also arriving one quarter earlier than originally planned.

The tech giant has committed to an annual release cycle for new Ascend processor generations. Looking ahead, the Ascend 970 and 980 series are scheduled for market introduction in 2028 and 2029 respectively.

The conference also featured the debut of Huawei’s Atlas 960 SuperPoD computing cluster. This enhanced system delivers superior performance compared to the Atlas 950 SuperPod unveiled just months earlier, offering upgraded capabilities for both AI model training and inference operations.

According to Xu, the Ascend 950DT chip has demonstrated impressive results during testing phases. He anticipates Chinese artificial intelligence model developers will commence training operations on systems powered by this chip throughout the coming year. To date, more than 40 AI models have been successfully trained using Huawei’s computing infrastructure, with over 5,200 developers actively engaging with Ascend chip software on a monthly basis.

Drive Toward Complete Semiconductor Independence

Xu spoke candidly about the strategic imperatives driving Huawei’s aggressive chip development program. American export restrictions have constrained China’s access to Nvidia’s cutting-edge processors since 2019. Huawei itself has operated under US trade limitations beginning in May of that same year.

“We refuse to accept a future where others control our destiny by deciding whether to sell chips to China,” Xu stated.

Beyond individual chip advancement, Huawei is developing strategies to overcome performance constraints by networking massive quantities of processors together. Current industry standards for training sophisticated AI models typically involve clusters containing approximately 100,000 chips.

The company introduced Peerium, a revolutionary architecture engineered to ultimately interconnect up to 1 million processors functioning as a cohesive system. Complementing this is UnifiedBus technology, which integrates processors, memory, storage, and networking components within these expansive systems.

According to Huawei, its Ascend 960 supernode configuration supports connections of up to 4,096 processors. These supernodes can subsequently be merged to form even larger computational clusters.

Deployment of over 1,000 systems utilizing the previous-generation Ascend 910C processors has already occurred. Meanwhile, Ascend 950-based systems have transitioned into commercial operation.

Despite Huawei’s progress, Nvidia maintains a substantial competitive edge through its CUDA platform, which developers worldwide depend upon for building and deploying AI applications. Nvidia declined to provide commentary when contacted.

These announcements from Huawei arrived less than two weeks prior to a scheduled meeting between US President Donald Trump and Chinese President Xi Jinping in Washington, set for September 24.

The post Huawei Accelerates AI Chip Production to Challenge Nvidia’s Market Position in China appeared first on Blockonomi.