LLM Providers Struggle With Reliability Under Growth

OpenAI and Anthropic status pages and related tweets (March 2 and March 5, 2026) show several services below 99.9% uptime, with ChatGPT at 98.86%. The author attributes outages to rapid user growth or "florescence," producing unpredictable load and saturation given costly GPU capacity constraints. Providers are prioritizing capacity redistribution, load-shedding, and graceful degradation to improve resilience.