In this episode I argue that AWS is the latest incarnation of a centralized mainframe architecture. Let me explain.
Welcome to Lead Prompt // executing leadership from the root. I’m your host, John Collins.
We all love the word "cloud."
It’s brilliant marketing: it sounds ethereal, it sounds decentralized. It evokes this invincible, invisible network of computing power floating above us, everywhere and nowhere at once. The cloud is omniscient.
But here is the reality: there is no cloud. It’s just someone else’s computer. And the truth is...we’ve been here before.
Picture the 1960s. The era of "Big Iron." Corporate offices relied on massive, room-sized computers, called “mainframes”, that could be found humming away in aggressively air-conditioned rooms. IBM ruled the world. And the employees? They didn't have powerful computers on their desks. They had "dumb terminals" that were just a keyboard and a screen. Every keystroke, every calculation, was sent back to the central mainframe to be processed.
Then, the microprocessor came along. We put powerful PCs on every desk. We decentralized! We broke free from the mainframe!
...Or did we?
In my opinion we haven't evolved past the mainframe at all. We’ve just given it a new name. Today, Amazon Web Services (AWS) is this generation’s mainframe.
Think about it. You might be holding a thousand-dollar smartphone, but when you open Netflix, Venmo, or load up a multiplayer game, your device is doing almost none of the heavy lifting. It’s just rendering what’s fed to it over the network. It’s a beautifully designed, incredibly expensive dumb terminal. The real work is happening in a massive, centralized server farm, likely in Northern Virginia.
And that centralization? It’s a massive vulnerability.
In distributed systems, a failure is local. But with a centralized mainframe, you have a single point of failure. You have an enormous "blast radius" when something goes wrong.
We’ve seen exactly what happens when that core cracks. Think back to October 2025, when a cascading bug in AWS’s vital us-east-1 region took down 113 services simultaneously. In a matter of minutes, 70,000 organizations went dark. Major banks, trading platforms, communications tools, even people's smart doorbells - all went dead. That single 15-hour outage cost the global economy an estimated $2.8 billion.
Or remember December 2021, when an AWS failure went so deep it crippled Amazon’s own physical logistics network. Delivery drivers couldn’t get routes, warehouse workers were frozen, and half the internet simply vanished.
We traded the IBM mainframe for the AWS cloud, and in the process, we surrendered control, we locked ourselves in, and we tied the entire global economy to a few massive server clusters.
So, how did we get here? How did we let the dumb terminal make a comeback? And more importantly, what happens when the modern mainframe goes down again?
Let's dive in.
1. The Era of the Mainframe
Firstly, we need to understand what the original mainframe era looked like.
- The Big Iron (1950s–1970s): Before personal computers, computing was entirely centralized. Organizations bought massive, expensive, room-sized computers known as mainframes. IBM dominated this era, particularly with the release of the IBM System/360 in 1964, which standardized hardware and software.
- The "Dumb Terminal": Users did not have computers on their desks. They had "dumb terminals" which were keyboards and screens with no processing power of their own. Every keystroke was sent to the central mainframe, which did all the computing and sent the visual result back to the screen.
- The Shift (1980s–1990s): The invention of the microprocessor led to the Personal Computer (PC). Computing power became decentralized. People suddenly had powerful machines on their own desks, leading to the client-server model and a shift away from absolute centralization. Though mainframes never truly died; they still quietly process billions of financial transactions today.
2. The Weaknesses of Centralized System Architectures
The centralized architecture of mainframe computing was fundamentally risky.
- The Single Point of Failure (SPOF): This is the fatal flaw of centralization. If the central node goes down, the entire network goes dark. There is no route around a dead core.
- Vendor Lock-In and Monopoly Pricing: When all your data, processing, and proprietary software live on one company's hardware, leaving becomes incredibly expensive and technically gruelling. The vendor holds all the leverage (side note: Amazon do the same today).
- Blast Radius: In a distributed system, a failure is usually localized (e.g., one user's PC crashes). In a centralized system, a single bad software update or routing error has a massive "blast radius," taking out thousands of dependent processes simultaneously via a cascade effect.
3. The Argument: Why AWS is a Mainframe
The tech industry loves to use the word "cloud" to evoke an image of decentralized, ethereal, and distributed networks. But logically, the cloud is just someone else's computer. Here is why I believe that AWS is the modern mainframe:
- Logical Centralization: While AWS physically owns millions of servers across the globe, the control plane consisting of the APIs, identity management, and underlying infrastructure is entirely centralized under Amazon's roof. It’s “house rules”, just like at the casino.
- The Return of the Dumb Terminal: We have come full circle. Today, powerful smartphones and tablets frequently act as modern "dumb terminals." When you open the Netflix app, Venmo, or Fortnite, your device is doing very little heavy lifting. It is simply rendering data processed and served by the AWS "mainframe."
- Regional Hegemony (The us-east-1 Problem): AWS has a massive data center cluster in Northern Virginia known as us-east-1. It is the oldest and most vital region in their network. So many foundational services and global companies rely specifically on this one physical location that it acts as the centralized heart of the internet.
- The New IBM: Just as employees at companies in the 1970s couldn't get fired for buying IBM, modern startups build natively on AWS. The tight integration of AWS's proprietary services like DynamoDB, Lambda, and S3, creates a level of vendor lock-in that mirrors the mainframe era perfectly.
4. When the Modern Mainframe Fails
Now let’s look at some real-world consequences of treating the cloud like an infallible utility.
The "Amazonk" Outage (October 20, 2025)
- What Happened: A cascading failure triggered by a rare software bug and "faulty automation" in AWS's DynamoDB database system crippled the us-east-1 region for roughly 15 hours.
- The Blast Radius: Because DynamoDB is a foundational "layer one" service, its failure took down 113 other AWS services. This wiped out roughly 70,000 organizations globally. Snapchat, Fortnite, Venmo, Coinbase, Ring doorbells, and major banks all suffered simultaneous outages.
- The Business Impact: Research estimates the global economic loss of this single outage at roughly $2.8 billion, with cyber insurers facing an estimated $38 million to $581 million in claims.
The Kinesis Data Streams Outage (July 30, 2024)
- What Happened: A newly upgraded architecture intended to improve scalability failed in the us-east-1 region, causing a 7-hour disruption.
- The Blast Radius: It broke the backbone of real-time data processing for thousands of applications, proving the risk of architectural complexity. While data wasn't lost, applications simply stopped behaving correctly.
The December 2021 Outages
- What Happened: Two massive outages hit us-east-1 within weeks of each other.
- The Blast Radius: The failure was so deep it actually took down Amazon's own logistics network. Delivery drivers couldn't get routes, warehouse workers couldn't scan packages, and consumers couldn't use their smart home devices or watch Disney+.
Conclusion
As an older engineer, I am often perplexed with the degree to which my younger colleagues outsource their server architecture to cloud providers like AWS. I often suspect they have no idea how the infrastructure works that runs their code in production every day, it’s a sad state of affairs.
The growth of the cloud has not just dumbed down end users, but to some degree it has dumbed down software engineers also. “Leave it to AWS to worry about” is a phrase I hear often, how about you?
Sources
Revealing the Cascading Impacts of the AWS Outage - https://www.ookla.com/articles/aws-outage-q4-2025
AWS Outage Analysis: October 20, 2025 - https://www.thousandeyes.com/blog/aws-outage-analysis-october-20-2025
The History of AWS Outage - https://statusgator.com/blog/aws-outage-history/
Major outage at Amazon disrupts businesses across the US - https://english.alarabiya.net/News/world/2021/12/08/Major-outage-at-Amazon-disrupts-businesses-across-the-US
Recent AWS Outages: Should You Panic? - https://www.globaldots.com/resources/blog/recent-aws-outages-should-you-panic/
Download audio
File details: 15.4 MB MP3, 10 mins 49 secs duration.
Title music is "Apparent Solution" by Brendon Moeller, licensed via www.epidemicsound.com
Subscribe
Sponsor
Five.Today is a highly-secure personal productivity application designed to help you to manage your priorities more effectively, by focusing on your five most important tasks you need to achieve each day.
Our goal is to help you to keep track of all your tasks, notes and journals in one beautifully simple place, which is highly secure via end-to-end encryption. Visit the URL Five.Today to sign up for free!
