Foreign media commentators believe that Microsoft's recent adjustment to the pricing of GitHub Copilot reflects a shift in the AI industry from subsidy-driven growth to more direct cost transmission. In the past, many generative AI products expanded rapidly with low monthly fees, but actual computing power costs did not decrease proportionally, and now enterprise customers are beginning to feel this pressure more clearly.

Microsoft's price adjustment raises cost issues

The article mentions that GitHub Copilot no longer relies entirely on a fixed subscription price, but instead emphasizes billing by token. This change has been dubbed "Tokenpocalypse" by some users, meaning that the cost of using AI is shifting from the platform to the client.

Commentators argue that the current seemingly low prices of many AI products largely rely on subsidies from investment funds. Once subsidies diminish, manufacturers will need to reflect more realistic model usage costs in their pricing, leading to changes in corporate purchasing and internal usage habits.

Companies begin to restrict internal use

The article uses Uber as an example, noting that the company quickly went through a process of increasing AI investment, then worrying about its budget being consumed too quickly, and finally considering setting usage limits. Based on this, the article points out that even large enterprises will soon face cost control issues after large-scale adoption of generative AI.

The author believes that the industry's pursuit of token usage heated up quickly and cooled down just as fast. The growth strategy based on "using more models and running more tokens" has begun to be questioned after corporate bills continued to rise.

Business models and regulations are under pressure simultaneously.

The commentary also noted that as more AI companies prepare for IPOs, the risk disclosures in their prospectuses may focus more on token costs, pricing sustainability, and customer willingness to pay. The article argues that these risks are changing too rapidly and have become issues the industry must address head-on.

Meanwhile, US President Trump signed an executive order this week that would give the government the opportunity to review powerful AI models. The article argues that AI companies are facing the challenge of balancing costs and revenue while also dealing with regulatory scrutiny, potentially accelerating the industry's adjustment process.

Overall, the core judgment of this commentary is that the AI industry must not only continue to improve model capabilities, but also quickly narrow the gap between product prices and actual operating costs. If this cannot be achieved, enterprise customer enthusiasm for adoption may slow down before technological progress.