Thursday, May 21, 2026

Last Night, 8 Hours of Darkness. The Reason Your Business Should Care About Cloud Outages.

Last Night, 8 Hours of Darkness. The Reason Your Business Should Care About Cloud Outages.

If you used Railway to host anything yesterday - an app, a website, a client portal, an online store - it went dark at 10:20 PM Eastern time.

And it stayed dark for eight hours.

Not because Railway made a mistake. Not because of a hack. Because Google Cloud accidentally suspended Railway's account. An automated system at Google flipped the wrong switch - hitting "many accounts within Google Cloud," Railway said in their incident report - and everything Railway's customers had built on top of it just... stopped.

This is one of the stranger things about how modern tech businesses are built. And it's worth understanding even if you've never heard of Railway.

What actually happened

Railway is a cloud hosting platform - think of it like renting a chunk of the internet to run your apps or websites. A lot of indie developers, startups, and small tech companies use it because it's simple and affordable.

Railway, in turn, runs most of its infrastructure on Google Cloud (GCP). That's normal. Almost every cloud hosting company does something similar. They rent space from the big providers (Google, Amazon, Microsoft) and resell access to smaller customers with a nicer interface.

Yesterday, Google's automated systems incorrectly flagged Railway's account and suspended it. Railway had their account access restored within 9 minutes of filing a ticket. But their customers - anyone running on Railway's platform - stayed offline for nearly 8 more hours while the underlying systems came back online.

Here's the part that made it worse: Railway's network routing also ran through Google Cloud. So even the customers whose apps lived on Railway's own hardware (not GCP) couldn't be reached once the cached routing tables expired.

Everything cascaded. One switch, one wrong automated decision, and it knocked out everything.

Railway published a detailed incident report at blog.railway.com. The transparency is commendable. But transparency doesn't get your customers back online.

Why this matters even if you've never heard of Railway

"This doesn't affect me - I use Squarespace / Shopify / GoDaddy."

Maybe. But here's the question worth asking: does your hosting provider use Google Cloud, Amazon AWS, or Microsoft Azure under the hood?

If yes, you're one automated system decision away from an event like this.

Squarespace runs on Google Cloud. Shopify runs on Google Cloud. So does Stripe, DocuSign, and hundreds of other services you probably use. AWS is underneath Airbnb, Slack, Adobe, and a huge portion of the internet.

This isn't doom-mongering. These platforms have far more redundancy than Railway, and a big provider like Google is unlikely to accidentally suspend Shopify's account. But the underlying fragility is real: the internet is stacked. Services run on top of services, all the way down.

The $0 audit that could save you later

Here's a simple thing to do this week:

List every tool your business depends on. Not just hosting - email, payments, booking, CRM, payroll, the works. For each one, ask:

  1. If this is down for 2 hours, what happens? Can I still take orders? Respond to customers? Accept payments?
  2. If it's down for 8 hours, what's the financial hit?
  3. Do I have any backup for this function?

Most small businesses have never done this exercise. Most find at least one catastrophic single point of failure - the kind that, if it went down, would mean they couldn't operate at all.

Three practical moves

If you use cloud hosting for anything critical: Check whether your provider offers a status page (like status.railway.com or statuspage.io-powered pages). Bookmark it. Set up alerts if they offer them. You want to know when something breaks before your customers tell you.

If you run an online store or client-facing service: Know how to post a quick "we're experiencing technical issues" message to your social media and email list. Don't wait to understand the problem - just communicate. Customers forgive downtime. They don't forgive silence.

If your whole operation runs on a single platform: Consider a minimal backup. An email address. A phone number. A static landing page hosted somewhere completely different. Something that works even when your main setup doesn't.

The honest truth about the cloud

The cloud is not a building. It's a series of promises between companies, all the way up a chain you can't fully see.

Most of the time, that works great. Yesterday, it didn't - and 8 hours of a small business being unreachable is the kind of thing that costs real money and real customers.

You can't eliminate this risk. But you can know where your weak points are before an automated system somewhere discovers them for you.

Sources: Railway GCP Incident Report - May 19, 2026

Terry Blake owns a landscaping company in Charlotte with 15 employees. He was the last person to try AI. Now he writes about what actually works for people who aren't tech-savvy.

Are you overpaying for AI tools?

Most small businesses waste $150+/month on tools they don't need. Find out in 2 minutes.

Take the Free AI Audit →

Liked this? There's more where that came from.

Every Sunday we send the week's best AI tips for your business. Free. No spam. Ever.