The cost of running frontier artificial intelligence models is becoming a serious budget problem for businesses, and companies are responding by shopping elsewhere — including turning to Chinese large language models and open-source alternatives.
According to Tom's Hardware, token costs for top-tier AI models have spiked sharply, pushing firms to seek cheaper options as their AI spending hits what the outlet describes as a "pricing wall." The shift represents a potential threat to the revenue of leading American AI companies, including OpenAI and Anthropic.
The economics of AI subscriptions add another wrinkle. Tom's Hardware reports that utilization rates above 5.7% on flat-rate subscription plans could actually push these startups into losses — meaning the more heavily a customer uses the product, the worse it is for the company's bottom line. That tension makes sustainable pricing a genuine puzzle for providers.
For enterprises, the calculus is straightforward: if a Chinese LLM or an open-source model can handle a given task at a fraction of the cost, the brand prestige of a frontier American model stops justifying the price tag. Open-source models in particular have improved rapidly, narrowing the capability gap that once made premium services easy to defend.
This matters because it signals a potential inflection point in the AI industry's business model — the assumption that enterprises would pay premium prices indefinitely for the best models is now being stress-tested by budget reality, and the winners of the next phase may be whoever cracks affordable, scalable AI delivery.