June 22, 2026, (Inside AI) — Zhipu AI shares surged 42% on Monday, pushing its market capitalization past HK$1 trillion for the first time. The rally followed the release of the company's new GLM-5.2 model, which has intensified competition in China's large language model sector.
The Surge: Numbers and Context
Knowledge Atlas Technology, Zhipu's listed entity, saw its stock peak at HK$2,980. The jump added billions in value, cementing Zhipu's status among China's AI elite. The GLM-5.2 launch triggered a buying frenzy, with trading volumes spiking to 3.2 times the daily average.
Analysts had anticipated a positive reaction, but the scale surprised many. The model's performance benchmarks suggest it rivals leading Western systems, narrowing a perceived gap. Zhipu's rise reflects broader investor confidence in China's AI self-sufficiency drive.
A War of Words Fuels the Fire
Adding drama, an exchange on social media platform X drew global attention. Tech billionaire Elon Musk predicted a Chinese AI model matching Claude Fable 5 would arrive in the first quarter of next year.
Zhipu founder Tang Jie shot back swiftly.
"It won't take that long,"
Tang's retort, now viral, underscored Zhipu's confidence and accelerated the stock's momentum. The exchange highlighted the intensifying AI arms race between China and the West, with timelines shrinking rapidly.
What GLM-5.2 Brings to the Table
The GLM-5.2 model boasts a 1.8 trillion parameter architecture, a leap from its predecessor. Early benchmarks show it outperforms GPT-4o in multilingual reasoning and code generation tasks. Its training data includes 12 trillion tokens, with a focus on Chinese-language nuance.
Zhipu claims the model achieves 92.3% accuracy on the MMLU benchmark, a key measure of knowledge. Efficiency gains also stand out; inference costs are 40% lower than comparable models, a critical factor for enterprise adoption.
Rival Reactions and Market Implications
Competitors like Baidu and SenseTime saw their shares dip as investors rotated into Zhipu. Analysts at CITIC Securities called the release a "Sputnik moment" for China's AI sector.
Yet skepticism persists. Some researchers question whether benchmark scores translate to real-world reliability. A Tsinghua University study noted that GLM-5.2's performance on adversarial prompts remains untested publicly.
The Bigger Picture: China's AI Ascent
Zhipu's milestone comes amid tightening U.S. chip export controls. The company reportedly uses a mix of Huawei Ascend and older Nvidia A800 chips, showcasing domestic innovation. Government subsidies have also played a role, with Beijing designating AI as a strategic priority.
Industry observers note that Tang's bold claim echoes a pattern of Chinese firms accelerating timelines. If GLM-5.2 lives up to its promise, it could reshape global AI supply chains and enterprise software markets.
Zhipu plans to open-source a smaller variant next month, potentially fueling a new wave of developer adoption. For now, the market is betting that China's AI moment has arrived ahead of schedule.
