Anthropic's Claude went down hard on April 13, 2026, taking nearly everything with it. The outage hit claude.ai, the API, Claude Code, Claude Cowork, and Claude for Government. Login failures were the main culprit, with users reporting 500 errors across the board. Anthropic posted its first acknowledgment at 15:40 UTC and was still investigating 18 minutes later.
The Hacker News thread told the real story. One developer hit 500s and jumped to Codex for documentation work. Another pointed out that this is what happens when your workflow depends on a single AI provider. Someone joked about the irony of Anthropic's own status page running on Atlassian, a SaaS dependency pileup that's almost too on-the-nose.
Reliability questions came up fast. A commenter noted Anthropic has "only one 9" of uptime, which stings when you're pitching government contracts and enterprise tools. Every major AI provider has faced similar outages. OpenAI went down hard in late 2023. Google stumbled during the Bard-to-Gemini transition. Even Anthropic, which just dealt with Sonnet 4.6 errors across all products last week, isn't immune. The AI industry is building critical infrastructure faster than it can harden it.
When Claude goes down and takes your entire workflow with it, you remember these are production systems now. People depend on them to ship real work. It’s a familiar scenario for developers who recall when Claude went down again earlier this year, disrupting chat and code services. The uptime hasn't caught up to the ambition.