China's Zhipu Closes In on Top US AI Models as Anthropic and OpenAI Face Export Curbs

China's Zhipu AI is closing in on the performance of top American AI models, CNBC reported, at precisely the moment when US export controls and government-mandated release restrictions are holding back Anthropic and OpenAI from their full international reach. The timing is significant: as Washington constrains its own AI champions, Chinese rivals are accelerating.

Zhipu, one of China's best-funded and most prominent AI research companies, has been steadily improving its models across benchmarks and real-world tasks. While US labs have benefited from access to the most advanced semiconductor hardware and enormous capital, Chinese firms like Zhipu and DeepSeek have increasingly demonstrated that the gap in capability is narrowing — and in some specialised domains, may be closing faster than the industry expected.

The broader context is a deliberate US strategy to use export controls, including restrictions on advanced chips and now on AI model access itself, to slow China's AI development. But critics of that approach argue it may be having a counterproductive effect — spurring Chinese investment and ingenuity rather than halting progress, while simultaneously handicapping US firms in global markets where Chinese alternatives can now compete credibly.

For India, which is positioned as a major battleground for AI adoption, the implications are direct. Indian enterprises, startups and government entities choosing between US and Chinese AI platforms will increasingly find the choice less clear-cut on performance grounds, with geopolitical, data-sovereignty and pricing considerations likely to become decisive factors in the months ahead.