America 'Sinking' as Global Users Flock to China's AI

The United States is facing intensifying competition from China in the technology sector, particularly in artificial intelligence (AI). Since the emergence of DeepSeek last year, which shook Silicon Valley, global consumers have been presented with a difficult choice. On one hand, American AI technology remains the most advanced, especially as the Trump administration continues to enforce protectionist policies designed to hinder China’s technological progress by blocking access to cutting-edge chips and chip-making tools. On the other hand, China’s relatively cheaper AI technology is becoming increasingly attractive to consumers, even if its sophistication previously lagged behind. However, this dilemma is gradually shifting. China is steadily catching up and is now closer to matching the sophistication of American AI. A new AI model from China, named GLM-5.2, has just been launched by the Beijing-based startup Z.ai. The model is touted as significantly closing the gap with Western-made AI technology. GLM-5.2 has caused a stir in Silicon Valley due to its coding and agentic capabilities—the ability to execute complex tasks with minimal instruction. These two crucial abilities are said to nearly rival top US products, but at a much lower cost. Some experts have labelled this phenomenon a ‘mini DeepSeek moment’. The GLM-5.2 model has rapidly climbed the usage rankings on third-party AI developer platforms like OpenRouter, currently placing above models from the US-based company Anthropic. Several data cloud executives, including Snowflake CEO Sridhar Ramaswamy and veteran venture capitalist Mark Andreessen, have acknowledged the model’s impressive capabilities. ‘We are now seeing Chinese AI models on the market that are just as good as models from OpenAI and Anthropic,’ said David Sacks, former White House AI czar, speaking before Washington lifted a ban on Anthropic’s most advanced Fable and Mythos models. This performance places Z.ai’s GLM-5.2 at the centre of a growing debate over whether China is finally catching up with the US in the AI race, amid warnings from tech industry leaders that unpredictable regulation from Washington risks hampering America’s edge in this cutting-edge technology. Sacks noted that the GLM-5.2’s performance is ‘just slightly below Anthropic’s Opus 4.8 and on par with OpenAI’s GPT 5.5,’ stressing that the US cannot continue to allow factors that slow down its leading tech companies. According to some experts, restrictions imposed by Anthropic and delays in the public release of OpenAI’s latest GPT-5.6 model have fuelled global demand for Chinese models. ‘The international developer community is increasingly realising that sole reliance on closed-source API models based in the US carries significant risks,’ said Brian Tse, founder and CEO of Concordia AI, a Beijing-based consultancy focusing on AI safety. The positive international reception for GLM-5.2 also signals a growing interest in more cost-effective open-source solutions. The Chinese tech industry’s offerings address a real-world problem, as businesses are becoming burdened by the rising and often unpredictable costs of using AI. This is because closed-source agentic AI tools tend to consume more tokens, the units used to measure AI usage. Z.ai, also known as Zhipu AI, declined to comment. Anthropic and OpenAI did not immediately respond to requests for comment. GLM-5.2 currently ranks fifth on the Artificial Analysis LLM intelligence leaderboard, which assesses performance based on various benchmarks measuring overall capability, including reasoning and programming skills. The model also holds second place in the Code Arena coding rankings, which measure the ability to build websites and applications. Notably, GLM-5.2 operates at roughly one-sixth the cost of leading closed-source US models like Claude and the GPT series. Z.ai has not disclosed the detailed costs of developing GLM-5.2. In a response to Elon Musk on X last month, Z.ai founder Tang Jie stated that the Chinese startup could produce a model equivalent to Anthropic’s Fable before the first quarter of next year. ‘The change that GLM-5.2 brings is that open-source models are now plug-and-play products,’ said Tiezhen Wang, former head of the APAC region at Hugging Face, a startup that serves as a hub for developers experimenting with open-source models. ‘You simply deploy the model and, without needing complex fine-tuning systems, it is already in a highly functional and ready-to-use state. This dramatically lowers the barrier to entry for open-source model adoption,’ he explained.