China’s latest AI offering, the GLM-5.2 model developed by Z.ai, is reshaping the enterprise AI landscape by delivering near-parity with top Western models at a fraction of the cost, posing a significant challenge to US-based leaders like Anthropic and OpenAI.
- GLM-5.2 offers performance near Anthropic’s models at one-quarter cost per token.
- Chinese AI token usage soared to 21.37 trillion in late June versus 5.76 trillion for US models.
- Lower costs could prompt enterprises to favor on-premises AI deployments over cloud-based solutions.
What happened
Jefferies’ latest GREED & FEAR report spotlights China’s GLM-5.2 large language model developed by Hong Kong-listed Z.ai. This model delivers enterprise-grade AI capabilities comparable to leading Western models such as Anthropic’s but at significantly reduced costs, approximately one-quarter of the price per token. This cost advantage positions GLM-5.2 as a disruptive entrant in the enterprise AI market, where budget constraints and security concerns heavily influence purchasing decisions.
The report also highlights data from AI platform OpenRouter indicating a rapid increase in adoption of Chinese AI models. Between late April and late June 2026, Chinese models saw token processing rise from 4.37 trillion to 21.37 trillion tokens, far outpacing the 5.76 trillion tokens processed by leading US AI models over the same period. This surge underscores a growing preference for more affordable alternatives among businesses globally.
Why it matters
The emergence of GLM-5.2 and its competitive pricing model challenges the premium positioning of Western AI leaders like Anthropic and OpenAI. These companies have built substantial enterprise revenue streams by offering high-priced, high-performance AI solutions. However, as Chinese models provide similar capabilities at a fraction of the cost, enterprise customers may reconsider their spending, fundamentally altering the economics of AI deployment.
This shift could accelerate the commoditisation of large language models, where pricing, deployment flexibility, and data privacy gain precedence over incremental performance gains. Enterprises might increasingly opt for on-premises AI deployments to safeguard sensitive data and reduce dependence on public cloud infrastructure. This trend could impact not only AI providers but also cloud service vendors, changing competitive dynamics across the tech ecosystem.
What to watch next
Going forward, adoption trends will clarify whether GLM-5.2 and similar Chinese models can sustain momentum and convert their cost advantage into long-term enterprise market share. Monitoring token usage growth, enterprise client wins, and any shifts in AI deployment from cloud to on-premises infrastructure will provide key signals of evolving customer preferences.
Additionally, semiconductor industry stakeholders could benefit from this trend despite concerns about cheaper models. Increased AI deployments driven by lower costs may boost demand for computing power, AI servers, and memory chips, supporting hardware growth. However, investors should also watch if established AI companies can maintain profitable growth amid intensifying price competition and evolving market expectations.