Anthropic is struggling to get compute from AWS, Google and Microsoft

Anthropic is struggling to secure enough computing power from AWS, Google and Microsoft as demand for its Claude AI models grows. Separately, the company has updated its usage policy to let Claude end conversations involving repeated abusive behaviour.

In Short

• Anthropic could spend over $750 billion on computing infrastructure
• Cloud providers are struggling to meet all its computing needs
• Anthropic will let Claude end conversations involving repeated abuse

Anthropic, the company behind the Claude AI chatbot, is struggling to secure enough computing power from Amazon Web Services (AWS), Google and Microsoft. These companies provide the cloud infrastructure needed to train and run AI models, but they are increasingly competing with Anthropic in the AI market and are not providing all the capacity it wants.

The problem comes as Anthropic's business grows rapidly and it prepares for a potential stock market listing. According to a draft IPO filing reviewed by The Wall Street Journal, the company could spend more than $750 billion on computing infrastructure in the coming years, including deals that are still under negotiation.

Anthropic has already committed around $450 billion to cloud computing agreements over the next seven to 10 years. Its planned spending includes $111 billion with Google Cloud, $110 billion with AWS and $31 billion with Microsoft. These long-term deals are meant to secure the computing resources needed to develop and operate its AI models.

However, Anthropic says these providers have not been willing to offer all the capacity it needs. The company is now leasing data centres and working with smaller operators, but it has warned that some of them have a "limited operating history" and may not be as reliable as major cloud providers.

Anthropic spent nearly $9 billion on computing capacity and infrastructure in the first half of 2026. It reported $16.3 billion in revenue during the same period, but its net loss reached $80.7 billion, largely because of accounting charges linked to its fundraising instruments.

Anthropic's struggle to secure enough computing power also points to a wider challenge facing the AI industry, as companies race to build infrastructure to meet growing demand. However, Elon Musk believes the problem is more complicated than electricity shortages alone.

Responding to a post on X discussing data centre constraints, the Tesla CEO said that a "lack of electricity" was an oversimplification. He argued that bringing massive computing capacity online at an extremely fast pace involves several challenges. Musk added that the task would be relatively easy if companies had twice as much time, but the pressure to expand quickly makes it much harder.

Anthropic gets emotional about Claude

In a separate development, Anthropic has updated its usage policy to allow Claude to end conversations if users repeatedly engage in “sustained and needless” abusive behaviour towards the AI chatbot. The decision is linked to Anthropic's research into whether AI models could have some form of welfare. The company has said it remains uncertain whether AI systems can have a moral status that deserves consideration, but believes it is worth taking precautions while studying the possibility.

The company says the rule will apply only in extreme cases of repeated abuse. It will not cover ordinary frustration, criticism, creative writing or AI testing. The change has sparked debate over whether people should be polite to AI chatbots. Some have welcomed the move, while critics argue that AI systems do not have feelings and should not be treated as humans. Microsoft AI chief Mustafa Suleyman, who has previously criticised Anthropic for treating AI as human-like, reacted to the policy update with a melting-face emoji.

- Ends