SpaceX Cuts Grok Bot Limits After Costs Outpace Subscription Revenue

Quick Facts

• Elon Musk confirmed Grok Bot usage limits were cut because operating costs were 'unsustainably high relative to revenue.'
• Paid weekly Grok Bot users hit 800,000 in early October, more than double the prior month, while daily active users surpassed 600,000.
• SpaceX's Super Intelligence segment generated $7.56 billion in Q3 2026 revenue, up from $813 million a year earlier, but AI infrastructure consumed 84% of the company's $19.6 billion quarterly capital expenditure.

Over Halloween weekend, Grok Bot subscribers found their weekly usage meters at 100% without having run any AI tasks. Users reported their meters jumping from 50% to 100%, 35% to 78%, and 70% to 100% overnight. Elon Musk confirmed on Sunday the changes were intentional.

'The cost of serving Grok Bot was unsustainably high relative to revenue,' Musk wrote on X in response to a user asking for a usage reset four days before their weekly allowance was due to refresh.

The timing is striking. Just two weeks earlier at a Q3 earnings call, Musk called Grok Bot the fastest-growing product he has ever seen. 'I've never seen anything grow this fast,' he said, adding that users were running more tasks per week, not fewer, as the product improved.

SpaceX President Gwynne Shotwell echoed that optimism at the same call, saying paid weekly users exceeded 800,000 in early October, roughly twice the figure from September. Daily active users topped 600,000 over the same stretch, up from fewer than 100,000 in early September.

That growth is precisely the problem. Grok Bot is included in SpaceX's X Premium ($30 per month), Super ($100), and Ultra ($200) subscription tiers. Usage draws from a shared weekly pool that also covers Grok and Cursor. As more subscribers run more complex autonomous tasks, the per-user cost climbs while subscription revenue stays flat.

SpaceX has not disclosed Grok Bot's operating costs, per-user cost, or profit margins for the product. The company confirmed the cost-revenue gap exists but has not quantified it.

The broader financials offer context. SpaceX reported $14.6 billion in Q3 2026 revenue, with its Super Intelligence business line generating $7.56 billion, up from $813 million in Q3 2025. That segment now accounts for 52% of total company revenue. The AI business produced $431 million in profit, a figure weighed down by $1.2 billion in one-time stock-based compensation tied to the $60 billion acquisition of Cursor parent company Anysphere, which closed August 14.

Infrastructure spending is the underlying pressure. NVIDIA is shipping Vera CPUs to SpaceX AI, and the company directed 84% of its $19.6 billion Q3 capital expenditure to AI-related buildout. That scale of spending makes flat-rate subscription economics difficult to sustain when usage grows as fast as Shotwell described.

User reaction was sharp. A thread on Reddit called the sudden limit changes an 'indirect rug pull.' Several users said the meter moved while they were asleep, which they interpreted as background agent tasks consuming their allowance without explicit instruction.

The backlash mirrors complaints from OpenAI users facing new restrictions on flat-rate plans, which some called the 'end of an era.' Anthropic made a similar move in April 2026, blocking Claude Pro and Max subscribers from using flat-rate plans to power third-party agent frameworks. A single complex multi-step agent session running over a full workday could cost between $1,000 and $5,000 in API compute, an expense Anthropic had been absorbing quietly for months.

The pattern points to a structural problem in how AI agent products are priced. Autonomous agents that browse the web, send emails, and log into accounts on a user's behalf consume far more compute than a single-turn chatbot response. Flat monthly fees designed for light usage break down when power users run agents continuously.

SpaceX has not announced revised pricing or new usage tiers. Musk's post confirmed the throttle was deliberate but offered no timeline for when limits might be restored or restructured.