US companies are increasingly adopting Chinese artificial intelligence models as a cost-effective alternative to premium domestic offerings, driven by soaring AI usage expenses and narrowing performance gaps. This trend, highlighted by the rapid uptake of models like Zhipu AI's GLM 5.2, reflects a broader shift in enterprise AI strategy toward value-driven procurement.
Cost Pressures Reshape Enterprise AI Buying
The financial strain of AI adoption has become a critical issue for many organizations. Reports indicate that Uber exhausted its entire 2026 AI budget within four months due to aggressive use of coding tools, forcing management to impose usage caps. This scenario is emblematic of a wider corporate reassessment, where firms are now routing less critical tasks to lower-cost models rather than relying solely on premium US providers.
Chinese models have emerged as primary beneficiaries of this cost-conscious environment. According to a Citi note, leading Chinese models charge as little as $0.18 per million tokens, compared to roughly $4 per million tokens for top US frontier models, while continuing to narrow the capability gap. On OpenRouter, a platform aggregating multiple AI models, the four most widely used models are now all Chinese, with DeepSeek remaining the most popular. DeepSeek charges only about 3% of OpenAI's GPT 5.5 token price, making it one of the cheapest frontier models available.
Market Share Surge
The growing appeal of Chinese models is evident in adoption metrics. The share of tokens consumed by US companies via Chinese models on OpenRouter has remained above 30% weekly since February 8, peaking at 46%. This marks a dramatic increase from an average of 11% over the prior 12 months and just 4.5% during the first half of 2025.
Zhipu AI's GLM 5.2 Leads the Charge
Among the newest entrants, Zhipu AI's GLM 5.2 has become a standout. According to Artificial Analysis benchmarks, it ranks fifth globally behind three Anthropic models and one OpenAI model, outperforming Google's Gemini series. When hosted by Zhipu in China, its token cost is only about 15% of OpenAI's comparable model. Since its June release, adoption has been rapid. Harpreet Arora, head of agentic infrastructure at Vercel, told CNBC that GLM 5.2 saw the fastest adoption of any model tracked by the company in 2026, with daily token volume growing 27x and customer count rising 80x in its first full week.
“Price is doing the work here,” Arora said. “When a task doesn’t need the best model, teams are beginning to route it to the cheapest one that’s good enough, and the recent wave of models coming out of China is winning that trade.” GLM 5.2 is particularly effective for software development and autonomous AI agents. Rehaan Ahmad, co-founder of Silicon Valley startup alphaXiv, noted that the performance gap has narrowed considerably, especially amid concerns that the Trump administration might regulate the technology. “With Fable restricted, the gap between the US and China is very slim,” Ahmad said.
Bloomberg reported that Zhipu is considering a share sale to raise several billion dollars in Hong Kong, after its stock surged about 2,000% since its January listing. A six-month lock-up expires on July 8.
Cloud Providers Expand Access
Major US cloud providers are facilitating access to Chinese models. Microsoft, Amazon Web Services, and Google Cloud now offer models from DeepSeek, Z.ai, MiniMax, and others. Microsoft has also explored integrating DeepSeek's latest model into its own AI products, according to Axios. For startups, the economics are compelling. San Francisco-based AI startup Lindy recently switched from a US model to a Chinese alternative to cut costs.
While security, regulatory, and geopolitical concerns remain barriers to broader enterprise adoption, the cost advantage and improving performance are driving a significant shift in AI procurement strategies among US companies.
This article is for informational purposes only and does not constitute financial advice.
