China’s AI Models Lead Global Token Usage, Challenging US

🔥 Main Takeaway

Chinese AI models are crushing it in global token usage, thanks to cheaper energy and smarter tech, putting US giants on notice for AI spending.

📌 What Happened?

Data from OpenRouter reveals Chinese artificial intelligence models now account for the majority of global token consumption.

DeepSeek V4 Flash emerged as the most popular model, recording 4.63 trillion tokens, followed by MiniMax M3 with 4.13 trillion tokens, and Xiaomi’s MiMo-V2.5 at 3.8 trillion tokens.

These labs benefit from cheaper energy and more efficient models, enabling them to offer tokens at significantly lower costs than their leading US competitors.

Major US models like Google’s Gemini 3 and OpenAI’s GPT 5.5 surprisingly ranked 12th and 13th, respectively, in global token usage.

💰 Why It Matters

This cost advantage is forcing US firms such as Amazon, Cisco, Uber, Walmart, and Meta to implement usage caps or switch to more affordable AI models.

Goldman Sachs projects a staggering 24-fold increase in token consumption by 2030, which will only intensify the existing chip shortage over the next 12 to 18 months.

AI startups like Anthropic and OpenAI, currently eyeing trillion-dollar valuations, face increasing pressure as companies scrutinize AI’s return on investment and pivot towards open-source or cheaper solutions.

The shift signals a potential rebalancing of power within the AI industry, impacting investment trends and business strategies globally.

👀 What to Watch Next

Keep an eye on how US tech giants adapt their AI strategies and pricing models to effectively compete with China’s cost efficiency.

Watch for further developments in the global chip supply chain as the demand for AI tokens continues its massive projected growth into the next decade.

Look for more companies to explore open-source and alternative AI models to manage escalating operational costs, potentially impacting the valuations of leading AI startups.