On Nov 11, TMT Post reported that AI investment in China keeps climbing, leaving tech companies squeezed on two fronts: capital spending and profit delivery. In a separate report the same day, TMT Post examined Token flows, cross-checking GitHub Copilot user and paid-subscription data against the computing costs of AI short dramas to compare traffic-driven and productivity-driven AI and unpack the business logic behind where Tokens go.
Token flows expose US-China AI split
According to TMT Post, 55% of China's Token consumption goes to short-drama scenarios, while 58% of US usage goes to code generation. The gap reflects different paths for the two AI industries: traffic-driven applications and productivity tools differ sharply in payment models and cost structures.
The report said that by cross-checking GitHub Copilot user and paid-subscription data and the computing costs of AI short dramas, it clarified two key figures the industry often conflates and stripped away inaccurate narratives circulating in the sector, restoring a true picture of what AI currently pays and costs.
55% of China's Token consumption goes to short-drama scenarios, while 58% of US usage goes to code generation.
| Country | Main Token consumption scenario | Share |
|---|---|---|
| China | Short dramas | 55% |
| US | Code generation | 58% |
Returns and business models under pressure
As AI investment keeps climbing, tech companies must cope with high computing costs and business models that have not fully proven out. TMT Post reported that the twin test of funding and profitability has become the core challenge for Chinese tech companies.
Looking at Token flows, traffic-driven scenarios such as short dramas soak up large amounts of computing resources, while productivity scenarios such as code generation show relatively clear payment conversion. That structural gap means different AI application paths diverge in commercialization efficiency, and the industry needs a clearer read on return on investment.
What to watch
Industry analysts say that as AI investment continues to rise, whether tech companies can strike a balance between computing costs and commercial returns will be the key thing to watch in the next phase. The real picture of payments and costs reflected in Token flows also offers a reference for judging whether AI business models are sustainable.