Foreign media commentary suggests that Microsoft’s recent adjustment of GitHub Copilot’s pricing reflects a shift in the AI industry from subsidy-driven growth to more direct cost pass-through. For some time, many generative AI products expanded rapidly with low monthly fees, but actual compute costs have not declined in tandem; now, enterprise customers are increasingly feeling this pressure.
Microsoft's price adjustment highlights cost issues
The article notes that GitHub Copilot is no longer solely reliant on a fixed subscription model but is placing greater emphasis on pay-per-token pricing. This shift has been dubbed the "Tokenpocalypse" by some users, indicating that AI usage costs are beginning to shift from the platform side to the customer side.
Comments suggest that many current AI products appear inexpensive largely due to investment subsidies. Once these subsidies diminish, manufacturers will need to reflect more accurate model usage costs in their pricing, leading to changes in corporate procurement and internal usage habits.
Companies are beginning to restrict internal use.
The article uses Uber as an example, noting that the company quickly went through a sequence of increasing its investment in AI, then worrying about rapid budget depletion, and finally considering setting usage limits. It concludes that even large enterprises soon face cost-control challenges after大规模 integrating generative AI.
The author believes that the industry's previous rush to pursue token usage rose quickly and cooled just as fast. The growth strategy centered around “using more models and consuming more tokens” has begun to face scrutiny as corporate bills continue to rise.
Business model and regulatory pressures are mounting in tandem.
The comment also noted that as more AI companies prepare for IPOs, risk disclosures in prospectuses may increasingly focus on token costs, pricing sustainability, and customer willingness to pay. The article argues that these risks are evolving too rapidly to be ignored and have become issues the industry must directly address.
Meanwhile, U.S. President Trump signed an executive order this week aimed at giving the government the opportunity to review powerful AI models. The article suggests that AI companies, while grappling with the gap between costs and revenues, also face increasing regulatory scrutiny, which may further accelerate industry adjustments.
Overall, the core judgment of this review is that the AI industry must not only continue to enhance model capabilities but also quickly narrow the gap between product pricing and actual operational costs. If this is not achieved, enterprise customers' enthusiasm for adoption may decline before technological progress continues.