Huawei unveils Ascend 960 supernode supporting 10-trillion-parameter models

On September 17, Huawei kicked off its HUAWEI CONNECT 2024 conference in Shanghai, unveiling the next-generation AI computing infrastructure — the Ascend 960 Supernode.

Key Specifications and Upgrades

• The new computing facility supports up to 4,096 computing cards, up from 1,024 in the previous generation.
• It is the first to adopt NPO (Near-Package Optics) technology, a critical innovation.
• Designed to meet the training and inference demands of 10-trillion-parameter models.

Strategic Focus on Computing Power

"If data is the fuel of AI, then computing power is its engine," noted Huawei's leadership. As computing clusters scale from tens of thousands to hundreds of thousands of cards, inter-card communication has become a key challenge.

Wang Tao, Huawei Deputy Chairman and Rotating Chairman, stated:

"We are focusing on the computing power domain and will continue to innovate through system architecture to meet the explosive growth in AI training and inference demands in China."

He added that the development of the Ascend 960 chip has exceeded expectations. The Ascend 960 air-cooled supernode and liquid-cooled supernode will be launched in the second and third quarters of next year, respectively.

Huawei's AI Strategy

Wang Tao emphasized that the core of Huawei's AI strategy is computing power, with a focus on:

• Hardware monetization
• System architecture innovation around the "Supernode + Cluster" model
• Building an open-source and open computing ecosystem
• Supporting mainstream large models for native training on Ascend
• Actively embracing the "hundreds of models, thousands of scenarios" vision

To date, over 1,000 sets of the Ascend 910C Supernode have been deployed, and the Ascend 950 Supernode has also entered commercial use.

Breakthrough NPO Technology

NPO is a key technology in the Ascend 960 Supernode. Traditional large-scale computing systems rely heavily on copper interconnects, which face physical limits in transmission distance and bandwidth. NPO technology helps reduce interconnect power consumption and latency while solving cabling and heat dissipation challenges in high-density computing systems.

Huawei has developed a next-generation NPO optical interconnect product — Hi-ONE, which is:

• The first mass-produced NPO product in the industry
• The largest transmission capacity NPO product, reaching 7.2T
• The only NPO product with a built-in light source

In the Ascend 960 Supernode, 5,500 Hi-ONE modules replace the 48,000 800G optical modules that would otherwise be required. This results in:

• Over 550 kW reduction in power consumption
• Doubled system uptime without failure
• System availability reaching 99.8%

The product integrates high bandwidth and high reliability with low latency and low power consumption, and works with the Lingqu high-speed interconnect protocol to build a scalable supernode interconnection system.

Industry Perspectives

Zhu Keli, Founding Dean of the National Research Institute for New Economy and Chief Expert of the Intelligent Economy, commented:

"Competition in computing power is not just about chips — it is about system architecture, software ecosystems, and overall industrial coordination."

He noted that traditional large-scale computing systems face challenges such as high communication latency, high costs, and insufficient reliability.

Zhu added:

"Huawei's new AI computing infrastructure provides a domestically producible foundation for large model training in China, driving upgrades in upstream and downstream industries such as optical communications and liquid cooling. By opening up interconnect standards, it also helps reduce fragmentation in China's computing ecosystem and promotes the engineering deployment of domestic computing clusters."

Miao Fuyou, Chief Technology Officer of the Global Computing Consortium Secretariat, said:

"China's biggest advantage in AI infrastructure is its full industry chain capability — covering chips, components, complete machines, power supply, and cooling systems. Each area is developing rapidly, thanks to China's manufacturing strengths and strong engineering capabilities. From a technology perspective, connectivity is a particularly strong area for China, playing a key role in the scalability and overall efficiency of supernodes and computing clusters."

Source: Economic Information Daily