The big tech outages of 2025 and where they leave hyperscaler partners

2025 is set to go down as the year of the hyperscaler outage. After partners grappled with downtime and customer comms, where do the big incidents of the year leave the industry?

The cloud has spent years telling the world it is too big to fail, but 2025 was the year that theory met reality.

From null pointers to DNS race conditions, the industry has been treated to a steady procession of outages that briefly turned the internet into a patchwork of loading screens.

For partners, it has prompted a familiar but increasingly urgent question. If the cloud is the backbone of modern IT, what happens when that backbone keeps creaking?

What follows is a rundown of the year’s most significant outages and a look at what they mean for partners whose customers expect everything to stay up even when the cloud is down.

SentinelOne consoles down, endpoints still protected on 29 May

The first major disruption of the year came from SentinelOne, which experienced a global platform outage that lasted several hours and took customer consoles offline.

The vendor told users it had experienced an outage affecting commercial customer consoles and confirmed that all customers and partners had been notified.

SentinelOne said customer endpoints remained protected throughout, although managed response services lost visibility and threat data reporting was delayed.

The vendor stressed that initial root cause analysis suggested the incident was not the result of a security breach.

Google Cloud and the null pointer problem on 12 June

On 12 June Google Cloud Platform suffered a significant outage lasting more than seven hours.

A null pointer problem brought down core GCP services and disrupted major platforms including Spotify, Gmail and Fitbit, but the ripples did not extend to enterprise software and the channel. For a large portion of the day users found themselves unable to access familiar services.

Cloudflare caught in the aftershock

Cloudflare said a number of its products and services were affected because they relied on Google Cloud. A spokesperson told CRN US this was a Google Cloud outage and that only a limited number of Cloudflare services were impacted. Core Cloudflare services continued to operate normally. By 20:57 UTC the company said all affected services were restored and it was monitoring metrics to confirm stability.

AWS and the US EAST 1 race condition on 20 October

Four months later AWS experienced a severe outage in its US EAST 1 region in Northern Virginia. The disruption lasted more than 15 hours and was caused by a DNS race condition. The outage produced widespread issues across popular applications websites and businesses.

In the UK, the incident added yet more fuel to the repatriation movement. But as partners maintain deep ties with US hyperscalers, an analysis by CRN revealed a more nuanced picture.

Debanjan Saha, former AWS vice president and now CEO of DataRobot, told CRN US the incident was essentially inevitable. He noted that given the global scale and complexity of cloud infrastructure it was remarkable these sorts of disruptions are as rare as they are. In his view the question for any provider is not if an outage will occur but when.

Microsoft Azure and the Front Door failure on 29 October

Microsoft followed with its own global outage only nine days after the AWS disruption. Azure Front Door, the company’s global CDN and application delivery service, failed for more than eight hours and caused issues across Microsoft 365, the Azure Portal and a number of enterprise customers including Alaska Airlines.

Microsoft published a preliminary post incident review setting out the steps it is taking to prevent a repeat event. Zac Paulson, vice president of technology at ABM Technology Group, told CRN the outage took down vendor portals the MSP relies on. He said the team simply waited it out and observed that incidents like this quickly reveal which products depend on which platforms. He also pointed out that some products still went down during the AWS outage the prior week, even in multi cloud environments, which raised questions about how failover is handled in practice.

Cloudflare’s maintenance mishap on 18 November

The most recent major outage originated from Cloudflare. A global disruption lasting roughly two hours took down some of the most heavily used applications on the internet including ChatGPT, Anthropic’s Claude, X, Shopify and Indeed.

Cloudflare first reported internal service degradation at 6:48 a.m. EST. Throughout the morning it issued updates stating that investigation was continuing and that work was underway to restore service. By 9:42 a.m. EST the company reported that a fix had been implemented. Some customers continued to experience issues accessing the Cloudflare dashboard shortly afterwards, but the primary disruption had been resolved.

What this means for partners

The outages of 2025 varied widely in cause and scope yet together they have underlined the same message. Cloud infrastructure, no matter how vast or sophisticated, is not infallible. Partners are now managing customers who assume cloud means constant uptime even though the events of this year have shown how fragile that assumption can be.

For MSPs and integrators the events have highlighted the importance of designing architectures that assume failure rather than hoping to avoid it. Multi cloud strategies can reduce risk but only if the software vendors building on top of hyperscalers have created genuine redundancy. Several partners learned this year that this is not always the case.

Clear and consistent communication has also emerged as a crucial differentiator. SentinelOne’s steady updates and Cloudflare’s rapid public status reports provided customers with reassurance during periods of uncertainty. The same kind of transparency will likely become an expectation during future incidents.

Above all the outages of 2025 have shown how deep customer reliance on cloud infrastructure has become. Partners sit at the centre of that relationship and are increasingly expected to translate complex outages into clear guidance and resilient strategies for the future.

Highlights