OpenAI CEO Sam Altman said the AI competition between the US and China is more complicated than often assumed.

Altman warned that the US may be underestimating China’s progress in AI, adding that US technology restrictions have not stopped China’s advances.

He noted that China could build AI inference capacity faster, but emphasized that the landscape includes research and product development, making it difficult to determine which country is ahead.

Chinese AI firms, including startups like DeepSeek and MoonshotAI and large tech companies such as Alibaba, have gained traction with open-source AI models that are widely adopted in the industry.

Altman’s comments align with Nvidia CEO Jensen Huang, who recently said that Huawei is able to grow its semiconductor business despite US export controls.

🔗 Source: South China Morning Post

🧠 Food for thought

1️⃣ China’s efficiency-first approach challenges the high-spending AI development model

Altman’s warning about underestimating China reflects a fundamental shift in how AI progress is being achieved.

While US private AI investment reached $109.1 billion in 2024 compared to China’s $9.3 billion, Chinese companies like DeepSeek are developing competitive models at significantly lower costs 1.

DeepSeek’s R1 model, launched in January 2025, demonstrates reasoning and mathematical skills comparable to leading US competitors despite substantially lower development costs 2.

This efficiency advantage allows Chinese companies to rapidly iterate and deploy models without requiring massive capital investments, potentially accelerating their development cycles.

The cost disparity raises questions about whether the traditional high-investment approach used by US companies is necessary for AI breakthroughs, suggesting alternative pathways to competitive AI capabilities.

2️⃣ The model performance gap is closing faster than funding ratios suggest

Despite the 10-to-1 funding advantage favoring US companies, the actual performance gap in AI models tells a different story.

Chinese AI models are now lagging behind US competitors by only three to six months, and are expected to match US capabilities within the year 34.