Critical cloud service outages declined in 2025 following several years of sustained increases, according to the Parametrix's Cloud Outage Risk Report 2025. After reaching a peak of 244.8 hours in 2024, the total duration of “critical” downtime events, those causing complete or significant service disruption, fell by approximately 28% to 175.3 hours in 2025.

Figure 1: Critical Events Overall Duration (Hours), by Single Event Block

Despite this improvement, outage risk remains significant. The report highlights that while overall downtime decreased, high-impact events persisted with major disruptions affecting some of the most critical cloud regions and services, including GCP outage in us-central1 region in June and the AWS outage us-east-1 in October.

Unlike previous years when specific providers suffered no major downtime incidents, 2025 was also notable because every major cloud provider experienced at least one impactful event, along with significant disruptions in critical digital supply chain services:

Figure 2:  Relative Magnitude of Digital Supply Chain Disruptions in 2025

• GCP (June 12): A global GCP outage, triggered by a faulty quota policy update, caused widespread API failures across core services like Compute Engine.
  ‍
• Cloudflare (August 21, November 18, and December 5): Distinct spikes linked to Cloudflare outages highlight the critical role of Content Delivery Networks (CDNs)- when they fail, access to applications can degrade or stop entirely.
  ‍
• AWS (October 20): The most significant spike (and the biggest event Parametrix has recorded to date) was the AWS us-east-1 outage, caused by failures in DynamoDB’s DNS automation that severed connectivity for a number of dependent services.
  ‍
• Azure (October 29): Shortly after the AWS incident, Azure experienced a global outage of Azure Front Door, caused by a configuration change.

This sequence of events illustrates that aggregate duration isn’t the only important factor when measuring the severity of cloud outage events. While total downtime hours were lower in 2025, the impact of specific incidents remained high. Every major cloud provider experienced at least one impactful event.

“As businesses and the world become increasingly reliant on cloud-based services, cloud providers’ ability to keep services operating continuously showed improvement in 2025,” said Jonatan Hatzor, co-founder and CEO of the company. “However, after a relatively event-quite first half, interruptions resumed and reached higher levels during the second half of the year.

“We saw significant outage events at data centers operated by each of the three main cloud providers,” he continued. “If average monthly downtime remains the same for the first half of 2026, we’re looking at a record-breaking 12 months from July to June. As reliance, and therefore the values at risk, continue to escalate, it is more important than ever to secure financial protection against outages of the digital supply chain.”

‍
Download the Cloud Outage Risk Report 2025.