Recently, according to foreign media reports, multiple users reported on social platforms that SpaceX quietly adjusted the usage quotas for its popular Grok Bot last weekend, and many users' quotas were directly cut in half. In response to users' complaints and questions, Elon Musk later publicly responded on X, admitting that "the cost of running Grok Bot is unsustainably high relative to its revenue." Although he did not directly admit in his public statement that a "cut" had been implemented, this remark undoubtedly confirms the real pressure currently facing AI agents between commercialization and high computing costs.

From Explosive Growth to Quota Limits

As an advanced feature that can browse the web, handle emails, and log into various accounts on behalf of users, Grok Bot shares the same subscription quota pool as Grok itself and the popular code editor Cursor. Just two weeks ago, Musk had publicly called it its "fastest-growing star product"; by early October, SpaceX President Gwynne Shotwell had also released impressive figures showing that its weekly paying users reached 80,000, more than doubling from the previous month.

However, as the user base and high-frequency usage surged, the issue of computing consumption also came to the surface. In this quota controversy, some users reported that after their quota was used up, they had to wait several days for a new quota to refresh. Facing users' complaints, Musk even asked in response, "Is a higher-tier subscription too expensive for you?"

Industry Under Collective Pressure: High Costs of AI Agents Become a Common Problem

In fact, SpaceX is not the only company facing the dilemma of an imbalance between computing costs and subscription models. Over the past period, the entire AI industry has been entering an intensive period of tightening affordable plans:

OpenAI has announced that it will begin merging Codex and regular chat quotas in December, and is moving to crack down on "freebie abuse" behavior involving frequent cross-account usage.

Anthropic also noted in its calculations that the computing cost required for an AI agent to independently complete complex tasks is often 4 to 15 times that of an ordinary simple conversation.

As AI agents from major tech giants accelerate their move to the forefront, how to ensure user experience while dealing with high backend computing bills that far exceed those of ordinary conversations has become a severe test for all large model companies.