Huawei Challenges Nvidia's AI Chip Dominance in China, Focusing on Inference Tasks

Huawei's Strategic Shift in AI Chip Market

Huawei, the Chinese tech giant, is making a bold move to challenge Nvidia's dominance in the artificial intelligence (AI) chip market within China. The company is strategically positioning its latest Ascend AI processors as the hardware of choice for Chinese companies running "inference" tasks, rather than competing directly with Nvidia in the model training segment 1

1

2

2

.

Focus on Inference Tasks

Huawei is betting on inference becoming a larger source of future demand as AI applications like chatbots become more widespread. Georgios Zacharopoulos, a senior AI researcher at Huawei's Zurich lab, explained, "Training is important, but it only occurs a few times. Huawei is mostly focused on inference, which ultimately will serve more customers" 1

1

.

Government Support and Market Dynamics

The Chinese government is actively supporting Huawei's push into the AI chip market. Officials have urged local tech giants to purchase more of Huawei's AI chips and reduce reliance on Nvidia 1

1

. This move comes in the context of U.S. export controls aimed at curbing Beijing's AI development, which has limited Chinese companies' access to Nvidia's most advanced GPUs 1

1

2

2

.

Technical Challenges and Competition

Despite Huawei's ambitions, the company faces several technical challenges:

1. Inter-chip connectivity issues when training large models 1

   1

   .
2. The need to convince developers to switch from Nvidia's widely-used Cuda software 1

   1

   .
3. Supply constraints due to U.S. export controls affecting chip manufacturing equipment 1

   1

   2

   2

   .

Huawei is not alone in this competitive landscape. Chinese companies like Baidu and Cambricon, as well as U.S. tech giants Amazon and Microsoft, are also vying for market share in AI chips 1

1

2

2

.

Market Performance and Future Outlook

According to SemiAnalysis, Nvidia still holds a significant lead in the Chinese market, having sold twice as many AI chips as Huawei in 2024. However, Huawei is rapidly increasing its manufacturing capacity to narrow this gap 1

1

.

Chinese companies are adopting a different strategy compared to their U.S. counterparts, focusing more on inference efficiency to drive down costs and achieve faster commercialization. This approach is exemplified by start-ups like DeepSeek, which recently released a model with low training and inference costs 1

1

.

As the AI chip race continues to evolve, Huawei's focus on inference tasks and its efforts to help companies adapt Nvidia-trained models to run on Ascend chips could potentially reshape the AI landscape in China, despite the ongoing technical and regulatory challenges 1

1

2

2

.