Ookla: AI Outages Are Becoming a Business Reliability Problem
As AI platforms become further embedded into daily infrastructure, outages cause major productivity pitfalls.
Zach Stark
WASHINGTON, June 10, 2026 – Artificial Intelligence is transitioning from flashy apps to fragile infrastructure, according to new Ookla data showing a sharp rise in reported AI service disruptions.
Ookla analyzed 471 days of U.S. Downdetector data from Jan. 1, 2025, through April 16, 2026, covering 3.72 million user-reported incidents across ChatGPT, Claude, Gemini, Microsoft Copilot, Amazon Web Services and Microsoft Azure.
The report’s findings suggest that AI reliability is no longer a standalone software issue. As AI becomes increasingly integrated into daily business operations, outages can disrupt writing, coding, research, analytics and customer support workflows that companies are beginning to treat as routine.
The report, based on U.S. Downdetector data, found that major disruption days for leading AI platforms jumped from six in the first quarter of 2025 to 51 in the first quarter of 2026. Ookla defined those as days when user reports were more than 10 times a service’s normal daily volume.
Claude accounted for the largest share of the first-quarter figure, with 39 high-signal disruption days. Gemini had seven, Copilot had three and ChatGPT had two.
The report described AI outages as a layered problem. A failed prompt or stalled workflow may appear to users as a simple platform outage, but the underlying cause could sit in the application, model orchestration system, cloud infrastructure, authentication service, content delivery network or routing layer.
“These patterns are making AI a deeper single point of failure than cloud alone, because enterprises are inheriting both the hyperscaler risk and the provider-specific orchestration risk on top of it,” said Luke Kehoe, lead analyst at Ookla.
The findings suggest the next phase of AI adoption may be less about whether the tools can perform impressive tasks and more about whether they can remain available when businesses build around them.
