Around the time I received my degree, the majority of computers were about the size of a Volkswagen bus and larger. You could connect with a terminal, either directly wired or call it up on a modem. Everything was based on accounts, and you paid for the CPU time you used. I spent a lot of time playing text based games (Zork) at 300 baud (30 characters a second), which, at the time, felt cutting-edge… mostly because we didn’t know any better.
Big, bulky, expensive, and slow. So slow that you usually ran your bookkeeping or inventory programs overnight and crossed your fingers nothing crashed before morning. Innovation was slow; the manufacturers made hardware and sold CPU time, why improve the experience when the billing model was already perfect?
When some bright engineer got the idea of using a calculator chip to create a desktop computer, the home computer revolution was born, and just like that, the meter started slipping out of their control.
I won’t bore you with the details, only that I was fortunate to be part of that revolution. I was able to see firsthand how hard the mainframe companies fought the home computer.
These became mantras for the mainframe salesmen:
- “It’s only a toy”
- “Nobody needs a computer on their desk”
- “They’re so limited”
The last one was especially amusing because, until the personal computer showed up, programs were outrageously expensive and limited. When you sell CPU cycles you have no need to innovate. Word processors and spreadsheets first showed up on home computers. An IBM salesman told my company that spreadsheets were unnecessary and IBM had no intention of selling them. I found it hard not to laugh when six months later, I was sitting in a classroom learning the IBM mainframe version of a spreadsheet. Easy class, by the way. I’d already been using a spreadsheet on my personal computer for almost a year, well ahead of IBM’s decision to join the party.
The thing is, a lot of innovations had to come together to create the computer you’re familiar with. And many of those innovations came from individuals and small companies because they finally had the ability, and the freedom to tinker on these home computers without asking permission or filing a report in triplicate.
This revolution disrupted constant income stream model for the mainframe makers. People bought home computers and the software they needed. No more money was spent until upgrades were necessary or new functions were desired. Customers deciding when to spend money, an absolutely reckless idea from a corporate perspective.
To the companies, this meant continual development and a rapidly declining income stream until the next update. Worse, while you’re working on the next generation of hardware or software, you’re at risk of your customers deserting you for the competition. Even worse, some people saw no reason to upgrade just to gain a marginal increase in speed or a handful of features that only mattered to 10% of the power users. The nerve.
This has pushed more and more companies toward a subscription-based model. You can’t buy the software; you have to rent it, indefinitely, preferably. Of course, they’ll tell you that you get all the updates as they happen. Just don’t look too closely at what those updates actually contain. Funny thing, even when there are discrete intervals between releases, there are always those few bugs they somehow never tested for. Miraculously consistent.
Moving to a subscription-based paradigm also allows more of the computing to take place on internal servers. This is important, especially as AI enters the picture. Now there’s a compelling reason to subscribe.
With higher bandwidth available to the average home, we’re seeing more and more applications hosted on servers rather than locally. Your computer is no longer doing the calculations on that spreadsheet; your keystrokes are sent to a remote computer, it does the calculations, and sends them back to you in a visual format. Efficient, convenient, and just a little reminiscent of something we already went through.
Well yes, this does require you to log into the spreadsheet server. That pesky issue of customers keeping their old programs for years is gone. Your subscription has to be current in order to use the spreadsheets you created. Customer loyalty? That disgruntled customer who wants to switch to a competitor isn’t going anywhere, you have all his files. It’s amazing how loyalty improves when leaving becomes difficult.
Privacy, that’s gone. You’re on someone else’s computer. They know exactly what you are doing, how often you hit delete, how often you save, and, apparently, how often you hesitate. Don’t worry, they’ll continue to store your data after your subscription lapses, at least two milliseconds before they allocate all that valuable storage space to a paying customer.
If you think I’m being an alarmist, then consider the vast numbers of data centers being planned. Sure, AI requires a lot of computing power, but that’s not the whole story. Those data centers are nothing more than the mainframes of yore, just repackaged, rebranded, and much better lit. And like the mainframes of yore, the providers intend to squeeze every penny out of you they can. Some business models never really die, they just change clothes.
Innovation, not so much. Don’t forget, the home computer revolution was driven by hobbyists. These were the people who innovated, small enough to take risks and not buried under layers of approval. Ever try to get a concept started in a large company? ROI, core business, customer penetration, and a stack of other reports thick enough to stop a bullet have to be produced before the executive board will even glance at it. The big companies know it’s cheaper to let small companies develop both the products and the markets. Once success is proven, they swoop in and buy the smaller companies. Works until there’s no longer any innovators.
Consider all the innovations made possible by local computing, 3D printing, for example. Can you imagine a data center running your printer? Unfortunately, if politicians have their way, it will happen and not for the better, but that’s a story for another day.
While writing this, I discovered a major drawback to the mainframe concept. With distributed computing hackers have a huge number of targets. As China recently discovered, storing all your data in one place allows the hackers to concentrate on only one target. I had to look up how large a petabyte is, an astounding one thousand terabytes. That hacker has to have his own data center to store all the data he absconded with.
I’ll leave you with one hopeful thought for the future. My first computer was a Southwest Technical Products 6800:
8-bit processor running at 1 MHz
4K ROM
32K max at $100 /4K
One serial port
Storage was provided using an audio tape recorder interface from Percom Data
Power required was 5 Volts at about 20 amps.
Graphics were limited, to say the least, but limited as it was, it could still run a full-featured word processor and a spreadsheet. By the time it was all said and done, I had over $2000 in it, more than I spent on my first car.
Today, I’m working on a project incorporating a XIAO ESP32C3:
RISC-V single-core 32-bit chip processor up to 160 MHz
400KB SRAM
4MB Flash
1x UART
Power required is 3.3 Volts at 75 milliamps, and it costs less than $5.
There is a difference, though. My first computer mainly ran programs written in assembly language. That was the only way to get them to fit. As complexity grew, more programs were written in high-level languages. High-level languages are far easier to write, debug, and maintain, but the trade-off is that they are memory-intensive and slower, though we’ve spent decades throwing hardware at that problem.
High-level languages came about because of human limitations. We think in words not registers or memory locations. AI doesn’t have those limitations. It’s not there yet, but it’s coming fast. The people betting everything on data centers may not have fully considered just how fast technology moves, or how often centralized control eventually gives way to something smaller, cheaper, and harder to contain.
Do I really need a data center and a subscription to power my spreadsheet, word processor, email, or edit my vacation photos? There are millions of innovators out there, and a few of them are already building the next revolution. If history is any guide, it won’t ask for permission, and it probably won’t require the equivalent of a small city to power it.
And of course, today’s song from Songer… Pay, Pay, Pay
© 2026, Byron Seastrunk. All rights reserved.
