Running AI is getting expensive: Why Elon Musk says Grok’s costs are unsustainable

Running an advanced AI chatbot can be expensive, and Elon Musk has acknowledged that Grok is facing this challenge. The billionaire said the cost of servicing xAI’s chatbot is “unsustainably high relative to revenue”, suggesting that the company is spending more to operate the service than it currently earns from it.

His comments came after an X user complained that Grok’s usage limits were being exhausted unusually quickly. The response has brought a bigger question into focus: can AI companies turn increasingly powerful chatbots into profitable businesses?

Why are Grok users hitting their limits?

The discussion began when an X user, DogecoinNorway, asked Musk whether Grok’s usage limits could be reset. The user said several people had noticed their quotas running out much faster than usual.

Musk did not promise a reset. Instead, he pointed to the financial pressure behind the restrictions, saying the cost of providing Grok was too high compared with the revenue it generated. AI chatbots require substantial computing resources to answer questions, generate content and handle complex requests. Every interaction uses processing power, while more advanced tasks can require significantly more computing. As usage grows, so does the demand for computing infrastructure. This can make it difficult for companies to offer extensive access through fixed-price subscriptions.

Is Grok's subscription model covering its costs?

Grok is integrated into X's paid subscription offerings, making the chatbot an important part of the platform's premium services. However, subscription revenue does not automatically translate into profit. Companies must also pay for computing hardware, data centres, electricity, infrastructure and the teams needed to develop and maintain their models. The supplied context estimates that Grok has around 800,000 paid users. However, this figure does not establish how much revenue the chatbot generates or how much it costs xAI to operate.

Musk's admission suggests that the current balance between usage and revenue remains a challenge. Usage caps can help companies manage demand and control operational spending.

Is the wider AI industry facing the same problem?

Grok's cost challenge reflects a broader issue for the generative AI industry. Developing an advanced model is only one part of the expense. Companies must also pay to serve it every time a user makes a request. This creates a difficult business equation: users expect powerful AI tools, quick responses and generous usage limits, while providers need to recover the cost of delivering those services. For Grok subscribers, tighter limits could mean fewer queries or more restrictions on intensive tasks. xAI has not, based on the information provided, announced a specific change to subscription prices or a revised usage policy. The bigger test for AI companies is no longer simply building more capable chatbots. It is finding a way to make them financially sustainable as demand continues to grow.