Independent · Multi-Cloud · Continuously Measured

The Definitive Source for
Cloud Reliability Intelligence

Executive-grade visibility into how the hyperscalers really perform. We measure every customer-impacting outage across AWS, Azure and Google Cloud — then benchmark, attribute and project it.

Read the 2026 report Compare providers

Featured Research · 2026 Annual Report

The State of Multi-Cloud Reliability

A 60-page analysis of every customer-impacting outage across AWS, Azure and Google Cloud — benchmarked, root-caused, and projected through year-end.

24 min read

708 minaggregate downtime tracked YTD

The Research

Featured research & benchmarks

Peer-reviewed methodology, transparent datasets, and a point of view. The work that informs every figure across the network.

Benchmark

SLA Performance vs. Promise

Where headline availability claims meet measured customer-impact — and the gap nobody publishes.

99.96%

blended availability

Methodology

How We Score an Outage

Our weighting model for blast radius, duration and recovery — the spine of every figure on this site.

74 min

mean time to recovery

Field Note

The us-east-1 Concentration Risk

Why a single region keeps surfacing in post-mortems, and what resilient architects are doing about it.

31%

outages: networking / DNS

Industry Stats · 2026 YTD

The numbers, measured not marketed

A continuously updated read on the three hyperscalers — uptime, incident volume, and where the downtime is coming from.

Tracked incidents YTD: 147
Aggregate downtime: 708 min
Blended availability: 99.96%
Mean time to recovery: 74 min

AWS

12% YoY

99.96%

rolling 12-month uptime

incidents YTD

214m

downtime YTD

Azure

8% YoY

99.94%

rolling 12-month uptime

incidents YTD

318m

downtime YTD

GCP

21% YoY

99.97%

rolling 12-month uptime

incidents YTD

176m

downtime YTD

Outages by root cause

Networking / DNS31%
Power & Hardware22%
Software Deploys19%
Capacity / Scaling16%
Configuration12%

Most significant incidents this year

Azure

Entra ID auth degradation across West Europe

Global auth · multi-region

3h 12m

AWS

us-east-1 DynamoDB elevated error rates

us-east-1 · API throttling

1h 47m

GCP

Cloud Load Balancing config rollback

us-central1 · networking

58m

Azure

Storage account latency spike, East US 2

East US 2 · storage

2h 05m

Trend Analysis

The full year, charted

Recorded actuals through June with a model-projected second half — the data spread behind our reliability call.

2026 Trend Analysis

Downtime trends, the full year in view

Monthly customer-impacting downtime per provider with a model-projected second half, plus the quarter-over-quarter picture across AWS, Azure and Google Cloud.