Amazon Now Runs Claude AI Without Your Data Leaving the Country

Amazon and AI maker Anthropic have turned on Claude — one of the leading AI chatbots — inside Amazon's data centers in Seoul and Singapore. The key phrase is “in-region inference,” which simply means your request is handled entirely inside the country you chose. Nothing about your prompt or the AI's answer gets routed anywhere else. Companies in South Korea can use Claude Opus 5 and Claude Sonnet 5; companies in Singapore can use Claude Sonnet 5.

Why does that matter? Banks, hospitals, and government agencies in those countries face strict rules about where customer data is allowed to go. Before this, they had a choice: send sensitive information overseas to use the best AI, or skip it entirely. Now they can use the same AI without breaking the rules. That could speed up everyday work like fraud checks, medical paperwork, and customer service in languages and legal contexts those countries actually recognize.

Getting started is simple for technical teams, and surprisingly approachable for everyone else. Amazon's console has a “playground” where you type a prompt and see the answer — no coding required. Developers can connect through standard interfaces (think of them as universal plugs that different software share), and Amazon adds safety filters called Guardrails, plus a feature that routes simpler questions to cheaper models to keep costs down.

The catch is capacity. Because everything stays in one country, you can only use that country's slice of computing power, and requests are capped by regional limits. Letting data travel between regions usually buys you more speed and scale. So this is a privacy-versus-power trade. For a hospital that legally cannot export patient data, it's an easy trade — the alternative was using no AI at all.