DeepSeek, a rising star in China’s AI scene, faced a major setback while trying to train its new R2 model using Huawei’s Ascend chips. The company aimed to launch the model in May but had to delay due to ongoing technical problems with Huawei’s hardware. This situation underscores how ambitious tech plans can sometimes run into unexpected hurdles, especially when relying on domestically produced components that are still catching up to international standards.

Challenges with Huawei’s AI Hardware

Huawei’s AI chips have been a point of pride for China’s push for self-sufficiency, but they still lag behind leading US-made chips like Nvidia’s. Huawei’s CEO, Ren Zhengfei, openly acknowledged that their chips aren’t yet on par with American rivals. While Beijing encourages local hardware use, this strategy can sometimes lead companies to choose less mature technology out of national pride or policy pressures.

DeepSeek discovered that Huawei’s chips work well for inference tasks—like answering questions or running simple AI operations. However, they struggle with the more demanding process of training AI models, which requires immense processing power and system stability. As a result, DeepSeek had to revert to using Nvidia’s systems, which are better suited for heavy-duty AI training. This switch caused delays that could affect their market plans and competitiveness.

Impact on DeepSeek and the Broader AI Industry

The delay has put DeepSeek on the back foot at a critical moment when speed and innovation matter most. The company’s founder, Liang Wenfeng, is reportedly unhappy with the pace of progress. He’s urging his team to push harder to develop a competitive AI system, but the hardware limitations remain a major challenge. This situation illustrates the difficulties Chinese tech companies face without access to top-tier US hardware and expertise.

While Huawei continues to improve its chips, the incident shows how reliance on domestically produced hardware can slow down breakthrough projects. Beijing’s strategy to become a global AI leader might need to be more balanced, investing more in refining local technology rather than solely promoting self-sufficiency. Without better hardware, Chinese AI firms risk falling behind in the race for advanced AI models.

This setback at DeepSeek is more than just a technical hiccup. It serves as a reminder that ambitious goals require not only good ideas but also reliable infrastructure. The incident has sent ripples through the AI community, prompting questions about Huawei’s ability to support high-end AI development and how China can bridge the gap with US rivals. For now, the path to self-reliance remains challenging, and companies like DeepSeek will need to navigate these hurdles carefully to stay competitive in a fast-moving industry.

Inspired by

• https://www.artificialintelligence-news.com/news/deepseek-reverts-nvidia-r2-model-huawei-ai-chip-fails/

Sources

• ft.com
• ai-expo.net
• intelligentautomation-conference.com
• blockchain-expo.com
• digitaltransformation-week.com