Enterprise
Anthropic lets enterprises store AI data on their own cloud
The company will maintain its 30-day retention policy for frontier models but allow businesses to control where that data lives.
Key takeaways
- Anthropic will allow enterprise customers to store required AI data on their own cloud infrastructure instead of on Anthropic's systems.
- The company's 30-day data retention requirement for frontier models remains in place, but businesses now control where that data lives.
- The policy applies to Anthropic's Fable and Mythos models and future frontier AI systems.
- Enterprise AI spending faces pressure as some businesses switch to cheaper alternatives like DeepSeek to improve ROI.
- Anthropic reached a $47 billion annualized revenue run rate in May and has filed confidential IPO paperwork with the SEC.
Anthropic is giving enterprise customers the option to store mandated data on their own cloud infrastructure rather than on the AI company's systems, according to a source cited by Reuters.
The policy change preserves Anthropic's 30-day data retention requirement for business customers but shifts control over where that data resides. Instead of storing information on Anthropic's servers, enterprises can now keep it within their own cloud computing environment. Bloomberg News first reported the development on Thursday.
The updated approach applies to Anthropic's Fable and Mythos models, as well as future frontier AI systems. The company introduced the 30-day retention mandate in June, explaining the requirement as necessary to detect and prevent potential cyberattacks leveraging its technology.
Why it matters
Data sovereignty has become a critical concern for enterprises deploying AI systems. Many organizations face regulatory requirements or internal policies that restrict where sensitive information can be stored. By allowing customers to maintain data on their own infrastructure, Anthropic addresses a significant barrier to enterprise adoption while still preserving the security monitoring window it deems necessary. The move comes as AI companies face mounting pressure to demonstrate clear ROI and compete with lower-cost alternatives.
Competitive pressure on enterprise AI
The announcement follows OpenAI's recent introduction of a safety system that can flag potential misuse without retaining customer data. That development, unveiled Wednesday, signals intensifying competition around data handling practices in the enterprise AI market.
Enterprise spending on AI services has faced headwinds in recent months. Some businesses have reduced their use of Anthropic and OpenAI products in favor of cheaper alternatives, demanding clearer returns on investment. Flo Crivello, CEO of AI startup Lindy, recently moved all company traffic from Anthropic's Claude models to DeepSeek, a Chinese company offering lower-cost, open-weight alternatives. Crivello said the switch would save Lindy millions of dollars within months.
IPO preparations continue
Both Anthropic and OpenAI filed confidential IPO prospectuses with the Securities and Exchange Commission in early June. Anthropic's annualized revenue run rate reached $47 billion in May, according to CNBC, as the company prepares for potential public markets.
The data storage policy change represents one element of Anthropic's broader effort to balance security requirements with enterprise customer demands for data control and regulatory compliance.
Reuters first reported the policy change, citing an unnamed source familiar with the matter.