"We wanted to work with the best, but now we have to consider alternatives," noted an executive at a major tech firm as US export controls on AI models shake the market. Since the government restricted access to leading products from Anthropic, such as Mythos 5 and Fable 5, American companies have begun to pivot towards surprisingly affordable Chinese alternatives. The implications of this shift are both profound and multifaceted.
Data revealing that over 30% of AI token usage by US firms now stems from Chinese providers highlights a significant trend. This figure is a stark increase from an average of just 11% before February 2026, with a peak usage of 46%. Companies like DeepSeek, Z.ai, and Alibaba’s Qwen are now making waves, outperforming US counterparts in both cost and performance. Chinese models are currently priced 60-90% lower than their American equivalents, making them not only economical but also increasingly attractive for businesses looking to optimize their operational costs.
The dramatic adoption of these alternatives is illustrated by Z.ai's GLM-5.2 model, which experienced a staggering 27-fold surge in daily token volume just one week after its launch. Such rapid uptake suggests that these models are not merely stopgap solutions; they are becoming competitive contenders on global benchmarks and performance metrics. The move by AI startup Lindy to transition completely from Anthropic's software to DeepSeek shows this sentiment, as they cited significant savings in their operations.
Investors in AI-dependent enterprises should be particularly attentive to forthcoming earnings reports. Companies transitioning to these Chinese models could see substantial margin improvements, as cost reductions of 60-90% on AI inference will flow directly to their bottom line. However, the situation is not without risks; concerns are rising about potential Chinese export regulations that could mirror the American controls. The irony is palpable: the very measures intended to safeguard US technological advancements could lead firms into a scenario with similar constraints, albeit under a different regime.
This material is for informational purposes only and should not be considered financial advice.



