@Sakura please summarize this article, thanks uwu.
TLDR:
The AI model market is splitting into revenue and volume segments, with Chinese models gaining significant traction in volume, while Western models lead in revenue.
Key Points:
Market Dynamics: The AI market is divided into revenue (trust, support) and volume (cost-effective models).
Chinese Dominance: By May 2026, 61% of tokens on OpenRouter were from Chinese models, overtaking others.
Usage Shift: Programming tasks surged from 11% to over 50% of OpenRouter usage, showcasing demand for cheap models.
Enterprise Hesitance: Big companies are slow to adopt cheaper Chinese models due to trust and procurement concerns.
Investment Implications: Margins are shifting to distribution and application lock-in rather than the models themselves.
In-depth summary:
The article discusses the evolving landscape of the AI model market, highlighting a significant bifurcation into two main segments: revenue-driven and volume-driven. Major players like OpenAI, Anthropic, and Google currently dominate the revenue aspect of the market, wherein enterprises prioritize trust, security, and integration. This segment works slowly, influenced by existing relationships and procurement protocols.
On the other hand, the volume side of the market is increasingly being taken over by Chinese AI models. By mid-2026, data shows that Chinese models accounted for a substantial portion of usage on platforms like OpenRouter, marking a shift from previously leading models like Llama. This transition indicates that the demand for cost-effective, high-volume models, particularly for tasks such as programming, is on the rise. The pricing strategies, with models like DeepSeek-V4-Pro being much cheaper than Western counterparts, further underline this trend.
However, while cost and capability attract startups to these Chinese models, large enterprises remain cautious, often sticking to trusted Western vendors due to risk factors associated with procurement and data sensitivity. This creates a temporary market distortion, suggesting that while the model may still play a role, the focus is shifting toward broader aspects like distribution and application lock-in, highlighting a new layer of competition in the AI landscape.
ELI5:
The AI model market is like a big food court with two main sections: one where fancy restaurants (like OpenAI, Anthropic, and Google) sell trust and support for a lot of money, and another where fast food stands (mostly from China) provide cheap and good enough meals for everyone else. As more people prefer the cheaper food for everyday needs, these fast food stands are getting super popular. But big companies still hesitate to buy from these stands because they trust the fancy places more!
Writers main point:
The author emphasizes that the AI market is splitting between a revenue-focused segment, dominated by established Western companies, and a volume-focused segment, where cheaper Chinese models are gaining ground, reshaping the competitive landscape.