The global artificial intelligence landscape has reached a volatile inflection point, underscored by the dramatic market response to the restriction of Anthropic’s Claude Fable 5. As US regulators and corporate backers alike grapple with the security implications of what many consider the first true general intelligence, capital has flowed aggressively into China's Zhipu AI. The company’s stock surged nearly 50% in a single session, signaling that the global search for high-tier reasoning capabilities is now overriding geopolitical boundaries.
Anthropic’s Fable 5 model proved to be a victim of its own success. Reports from Silicon Valley suggest the model could migrate complex enterprise codebases in 24 hours—a task that typically consumes two months of human engineering—and generate entire software environments from a single prompt. However, its capacity to autonomously discover and exploit cybersecurity vulnerabilities prompted immediate intervention from the US Treasury, creating a sudden supply vacuum for elite-tier reasoning models.
In the resulting scramble, the Chinese market has seen a stark divergence in fortunes among its leading 'tigers.' While Zhipu AI’s GLM-5.2 has successfully captured the overflow demand by offering superior reasoning at a fraction of the cost, its rival MiniMax has faced a sharp correction. A recent JP Morgan report highlighted this split, upgrading Zhipu to a 'buy' while slashing MiniMax’s target price by over 60% after the latter failed to sustain a premium pricing strategy for its latest model.
Zhipu’s trajectory throughout 2026 serves as a new blueprint for AI monetization in an era of 'intelligence as infrastructure.' Despite a significant hike in API costs, the company saw a 400% surge in usage, indicating that high-level reasoning has become an inelastic necessity for enterprise clients. As the industry prepares for a wave of secondary market listings, the market is shifting its focus away from user growth and toward technical moats that translate directly into sustained margins.