Google Cloud Suspends Railway Account, Triggering Massive Service Outage
Table of Contents
An Eight-Figure Relationship, A Sudden Shutdown
For Railway, a platform-as-a-service (PaaS) designed to automate the complexities of code deployment, the cloud is supposed to be the invisible engine that powers their business. But on Wednesday, that engine didn’t just stall—it vanished. Google Cloud unexpectedly suspended Railway’s account, triggering a cascading service failure that left users staring at ‘no healthy upstream’ errors and login failures.
The outage began around 22:00 UTC on May 19. According to Railway, the disruption was absolute. Resources didn’t just malfunction; they appeared to have been deleted from existence. While Google later clarified that the account had been suspended—effectively making the resources invisible rather than destroyed—the result for Railway’s customers was the same: a total blackout of services.
The incident is particularly jarring given the financial scale of the partnership. Railway reportedly spends over $10 million annually on Google Cloud, maintaining its control plane and critical databases within the ecosystem. Despite this high-tier spend, the company claims it took a full hour for Google’s support team to even engage after the suspension began.
A History of Infrastructure Friction
This is not the first time Railway has clashed with its cloud provider. In a move that now seems prescient, the company shifted a significant portion of its infrastructure to colocation services in 2024. That decision was driven by a series of problems at Google that Railway described as an “existential risk” to their business. Yet, the tether to Google Cloud remained necessary for their core management tools and data layers.
Angelo Saraceno, a solutions engineer at Railway, described the atmosphere during the outage as one of frustration and confusion. “Our contacts at Google were confused, customers are irate,” Saraceno noted. While Saraceno speculated that Railway might have inadvertently triggered an automated enforcement rule, the company maintains that the suspension came “out of the blue,” without the standard warnings usually associated with billing disputes or policy violations.
The ‘Invisible’ Risk of Cloud Dependency
The Railway incident highlights a recurring anxiety for modern software startups: the fragility of the “single point of failure” inherent in hyperscale cloud providers. When a provider like Google Cloud decides to flip a switch, there is no recourse for the customer, regardless of their annual spend. The incident mirrors a previous 2024 event where Google wiped out infrastructure for the Australian pension fund UniSuper without sufficient cause.
By the early hours of May 20, Railway began reporting that workloads were returning online, though non-enterprise deploys remained paused for a period during the recovery process. For the engineers at Railway, the aftermath is less about the technical fix and more about the reputational damage.
“Our customers don’t care if it is Google,” Saraceno said, emphasizing the harsh reality of the PaaS market. “We have to own our uptime.”
Google has not yet provided a detailed public explanation for why the account was flagged or suspended, leaving the industry to wonder how a top-tier customer spending eight figures can be shut down by an automated system with such little oversight.