Parametrix, which provides digital business interruption solutions, said in a new whitepaper that critical cloud service outages declined in 2025 after several years of steady increases. The risk, still, remained material.
Total critical downtime, meaning events causing complete or serious service disruption, dropped by about 28%. After reaching 244.8 hours in 2024, it fell to 175.3 hours in 2025.
The report said lower total downtime did not mean lower exposure across the board. High-impact incidents still hit some of the most important cloud regions and services, including a Google Cloud outage in the us-central1 region in June and an AWS outage in us-east-1 in October.
Parametrix said 2025 stood out because every major cloud provider recorded at least one impactful event. The year also included serious disruptions across digital supply chain services.
These incidents included the global Google Cloud outage on June 12, Cloudflare disruptions on August 21, November 18, and December 5, the AWS outage on October 20, and an Azure failure on October 29.
The AWS incident produced the sharpest spike and the largest single event Parametrix has recorded so far.
The us-east-1 outage stemmed from failures in DynamoDB DNS automation, which cut connectivity for several dependent services. Soon after, Azure suffered a global Azure Front Door outage caused by a configuration change. Rough stretch, no way around it.
The report said this run of incidents shows aggregate duration is not the only useful measure when judging outage severity.
Total downtime was lower in 2025, yet the effect of individual events stayed high. Every major cloud provider, according to Parametrix, experienced at least one impactful outage.
Jonatan Hatzor, co-founder and chief executive officer of Parametrix, said cloud providers improved service continuity in 2025 as businesses and the wider economy leaned harder on cloud-based systems.
He also said the first half of the year stayed relatively quiet, though interruptions returned and intensified in the second half. Parametrix recorded major outage events at data centres run by each of the three largest cloud providers.
If average monthly downtime holds at the same level through the first half of 2026, the market is looking at a record-setting 12-month period from July to June.
As dependence on cloud services rises, and the value at risk rises with it, he said financial protection against digital supply chain outages matters more than ever.
