At 6:40 this morning Eastern, people started tapping their phones at the register and nothing happened.
Apple Pay was down. DoorDash was down. Reddit, Hulu, PlayStation, all of it, gone. Not because those companies did anything wrong. Because a single Amazon Web Services region in Oregon coughed.
AWS didn’t officially admit anything until 4:40 a.m. Pacific, which is 7:40 Eastern, a full hour after Downdetector lit up. They called it resolved shortly after 9 a.m. Eastern.
Two hours and twenty minutes, give or take.
And most people are going to read that number and think it means everything’s fine.
What Actually Broke The Internet This Time
The failure hit AWS’s US-West-2 region in Oregon. Amazon started by naming five affected services and then quietly bumped it to seven: Direct Connect, Global Accelerator, Internet Connectivity, IoT Core, API Gateway, Elastic Compute Cloud, and Elastic Container Service.
That list matters, so here it is in English.
EC2 is the virtual servers your apps actually run on. API Gateway is the switchboard that lets different systems talk to each other. Direct Connect and Internet Connectivity are the roads in and out of the region. Global Accelerator picks the fastest route to get you there.
When five of the seven things that broke are plumbing rather than products, you’re not looking at one bad server. You’re looking at multiple layers of a regional network failing at the same time. Amazon flagged it as “impacted,” which is corporate for yes, this is real, please stop emailing us.
Downdetector’s breakdown was ugly. Roughly 30 percent of the AWS complaints pointed at EC2. Another 29 percent named US-West-2 specifically. And 23 percent were coming out of US-East-1 in Northern Virginia, which is a different region entirely, on the other side of the continent.
Read that last part again. The failure was in Oregon and almost a quarter of the pain was showing up in Virginia.
October Was Virginia. July Was Oregon. Nobody Fixed the Actual Problem.
Nine months ago I wrote The Day the Internet Broke after AWS took down US-East-1 at 3:11 in the morning and dragged Snapchat, Signal, Roblox, airline check-in systems, banks, and Ring cameras down with it. Internal network and DNS failures. One company, one mistake, half the internet on its back.
The response from the industry was a lot of blog posts about multi-region architecture and exactly zero structural change.
Amazon still runs about a third of global cloud infrastructure. Add Microsoft and Google and you’re at roughly 70 percent of the internet sitting on three companies’ hardware. Your bank, your camera feed, your medical portal, your kid’s school app, and the payment terminal at the gas station are all renting space in the same handful of warehouses.
That’s not a network. That’s a bottleneck with good marketing.
The Cashless Part Is the Part That Should Scare You
Forget Reddit being down for two hours. Nobody died because they couldn’t scroll.
Apple Pay went down. That’s the one.
There are a lot of people walking around right now with no cash in their pocket, no backup card, nothing but a phone and a whole lot of faith. For two hours this morning those people could not buy anything. Not gas, not groceries, not a bottle of water. The money was still in the account. The account just couldn’t be reached.
Now stretch that from two hours to two days and ask yourself what your town looks like.
We’ve beaten this drum before in Why Preppers Should Be Seriously Concerned About a Cashless Society, and every outage like this one proves the same point. The system doesn’t need to be hacked or seized or weaponized to lock you out of your own money. It can just have a bad Friday morning.
If a company glitch can freeze your ability to pay for two hours, a Central Bank Digital Currency can freeze it for as long as somebody upstairs decides you deserve.
This Was an Accident… But does that make it any less worrying?
Here’s the part that keeps me up.
This morning was a mistake. Some config change, some hardware fault, some routing table that went sideways. No malice, no coordination, no adversary. And it still took down payments, food delivery, and gaming for millions of people across the country before most of them finished their first cup of coffee.
Over 85 percent of American critical infrastructure is privately owned. Banks, telecom, the grid, the data centers, all of it, in corporate hands with inconsistent oversight and security budgets that get cut every time earnings come in light.
Now picture somebody doing this on purpose, to three regions at once, on a Monday, timed with something else. Payment rails down. Comms degraded. Logistics frozen. Grocery stores run about 72 hours of inventory on a good day, and that assumes trucks are still moving and registers still work.
The shit doesn’t have to hit the fan all at once. It just has to stop moving for four or five days.
We covered the physical side of this in Two Government Reports, One Answer: The Grid Runs Short by 2029 and The DHS Power Grid Warning Washington Wants You to Forget. Same story, different wire. Everything’s centralized, everything’s brittle, and everybody in charge is betting nothing goes wrong twice on the same day.
What You Actually Do About It
None of this requires a bunker.
Carry cash. Real bills, small denominations, in your wallet, today. A couple hundred at home in a safe spot. This morning proved the entire case for it in a hundred and forty minutes.
Keep a second payment method that doesn’t route through your phone. A physical card, from a different bank than your primary, is not a luxury.
Get the important stuff off the cloud and onto hardware you own. Documents, photos, medical records, insurance policies, contact numbers. An external drive costs less than a decent dinner out. Print the contact list. Paper doesn’t need a region.
Buy paper maps for your area and the routes you’d actually take. Your phone’s navigation is somebody else’s server.
Have a way to get information when the apps quit. Ham radio is still the answer, and it’s still the thing most people put off. Start with Emergency Communications When TSHTF and the Ham Radio Cheat Sheet.
Cover the boring four: water, food, shelter, protection. Handle those and you’re already ahead of nearly everyone in your zip code. Prepper 101 walks you through it if you’re starting from zero.
Then sit down and run a real Threat Assessment on your own situation, because a cloud outage hits a rural homestead and a downtown apartment in completely different ways.
Amazon will publish a post-mortem in a week or two. It’ll be well written, technically honest, and full of language about improved monitoring and resilience testing. Everyone will nod and go back to building on top of the same seven services in the same three companies’ data centers.
And some morning that isn’t a Friday in July, the fix won’t come by nine.
Related reads:
Read the full article here