Last updated: September 25, 2026
Every company is now a technology company because the way it takes orders, ships product, bills customers, and figures out whether it made money all runs on software. So technology decisions are business decisions now, whether leadership treats them that way or not. That doesn’t mean you should go build an app. It means somebody senior has to own the stack.
“We sell irrigation fittings. We are not a tech company.”
The owner said it with his arms crossed, which is how most of these meetings start. Fine. I asked how an order gets in the door.
Shopify for the contractors who buy online. EDI from two big-box retailers. Reps who email PDFs of purchase orders to a shared inbox, where a woman named Deb retypes them into the ERP every morning before nine. Inventory lived in the ERP, mostly, except for the overflow warehouse, which lived in Deb’s head. Freight quotes came from a carrier portal. Commissions came from a spreadsheet that one person on earth understood, and she was on maternity leave.
I counted eleven systems before we got to accounting. He sells irrigation fittings. He also runs a small, badly integrated software company that happens to have a warehouse attached.
He didn’t love hearing that. Nobody does.
The Line Is Old. People Still Read It Wrong.
I didn’t coin this. Gartner’s Peter Sondergaard used “every company is a technology company” in the firm’s 2013 Symposium keynote, and Marc Andreessen had already made the longer version of the argument two years earlier in “Why Software Is Eating the World,” where he described FedEx as “a software network that happens to have trucks, planes and distribution hubs attached.” Satya Nadella got on a stage in Barcelona in 2019 and said computing was now “a core part of every industry.”
So it’s a keynote line. Which is exactly the problem.
Keynote lines get repeated until they stop meaning anything, and this one got flattened into “you need to act like a startup.” Hire developers. Build a mobile app. Put AI in the investor deck. Rename the IT department “Digital.” I’ve watched a $70M distributor spend most of a year and a lot of money building a customer app that about forty customers ever downloaded, because somebody on the board read a Forbes piece on a plane.
That is not what the phrase means. At least not the part of it that’s true.
Quick disclosure before I go further. This is on KORE1’s site, and KORE1 places technical people for a living through its IT staffing services, so they’re happy when anybody concludes they need more of them. I sell consulting. Everyone here has an angle. Mine is that most of the fix is cheaper than you think and doesn’t start with a hire.
What “Every Company Is a Technology Company” Actually Means
“Every company is a technology company” means that a business’s core operations, from taking an order to closing the books, now run through software and the data moving between systems. Technology stops being a support function. It becomes the operating model, and decisions about it carry the same weight as decisions about pricing, inventory, or headcount.
Read that again and notice what isn’t in it. Nothing about writing code. Nothing about apps. You can be a technology company and never employ a single developer.
What you can’t do is treat the stack as plumbing that somebody in a back room keeps running while the “real” business decisions happen upstairs. Because the real decisions go through the plumbing. When a CFO tells me “we need a system to tell us if we’re making any money,” that’s not an IT request. It’s the most important business question there is, and the only honest answer runs through the ERP, the order channels, and whatever glue sits between them.
Most of the time the glue is Deb.
Look Who Actually Employs the Developers
If you want proof this isn’t just consultant talk, look at where software developers actually work. The Bureau of Labor Statistics counted about 1.7 million software developer jobs in 2025, and its Occupational Outlook Handbook breaks down the largest employers.
| Industry | Share of software developers (BLS, 2025) |
|---|---|
| Computer systems design and related services | 29% |
| Finance and insurance | 11% |
| Software publishers | 10% |
| Manufacturing | 8% |
| Management of companies and enterprises | 6% |
Add up the two rows you’d call “tech.” Software publishers plus IT services firms come to 39%. That leaves roughly six in ten developers working somewhere else: banks, insurers, factories, corporate headquarters, hospitals, retailers, and a long tail of companies whose websites don’t mention software at all.
Those companies didn’t hire developers for fun. They’re expensive. They hired them because the business stopped working without them.
Now the uncomfortable half. The Census Bureau’s Business Trends and Outlook Survey found that 19.8% of U.S. businesses were using AI as of May 2026, against 37% of firms with 250 or more employees and 39.7% in the Information sector, while retail trade sat around 14%. Big firms and actual tech firms are pulling away, and the companies in the middle, the $50M to $500M manufacturers and distributors I spend my weeks with, are mostly watching from the parking lot.
That gap is the whole reason this phrase matters for a mid-market company. It isn’t about whether you’re a technology company. You already are one. It’s about whether you’re a good one or a bad one, and the bad ones don’t find out until a competitor quotes faster, ships faster, and knows its margins by Tuesday instead of by the 20th of next month.
Count Your Systems. Seriously, Count Them.
Here’s the exercise I run in the first meeting. It takes an afternoon and costs nothing, which consultants aren’t supposed to admit.
- Write down every system that touches an order, from the first click or phone call to the cash landing in the bank. Include the spreadsheets. Especially the spreadsheets.
- Draw a line between any two systems that share data. Then mark which lines are automated and which ones are a person copying and pasting.
- Next to each manual line, write the name of the human doing it. If the same name shows up three times, you’ve found your single point of failure, and she probably wants a raise.
- Ask one question of every system on the page. Who decided we’d use this, and who would decide to stop?
That last one is the killer.
At the irrigation company, the answer for about half the list was “it came with the guy who used to run the warehouse” or “the marketing agency set that up in 2019.” Nobody chose the stack. It accreted. Every system was a reasonable decision made by somebody solving a local problem on a Tuesday, and eleven reasonable local decisions added up to one unreasonable company-wide mess. Untangling it is a business systems consolidation project, and it goes faster with people who’ve done one before.
Not broken, exactly. Unowned.
What It Doesn’t Mean
I want to be clear here, because this is where the keynote version sends people off a cliff.
It doesn’t mean you should build your own software. Most mid-market companies should buy almost everything and customize carefully, and I’ve written a whole piece on when to build versus buy because the urge to build is strong and usually wrong. Your competitive edge is almost never a custom order-entry screen.
You also don’t need a full-time CTO. A $60M distributor with no software product doesn’t need a Silicon Valley executive with stock options and opinions about Kubernetes. It needs judgment, a few days a month, applied to decisions that currently get made by whoever complains loudest.
And it doesn’t mean the technology is the hard part. This is the myth I push back on hardest. People tell me ERP is complicated. It isn’t, really. The software does what it’s configured to do. Your people, your processes, and the three departments that each define “an order” differently, those are complicated. Nadella said something close to this at that same 2019 conference, that the transformation of technology “was not the hard part” and “the business model was the tough part.” Microsoft’s CEO, of all people, telling a room of executives the software wasn’t the problem. Funny how that quote never makes the slide deck, while the “every company is a software company” half ends up on every vendor’s homepage for the next five years.
Buying a new system to fix a process problem is the most expensive way to learn you had a process problem.
Somebody Has to Own It
So what does being a good technology company actually take, if not developers and an app?
Ownership. That’s most of it.
Somebody with enough authority to say no has to own the stack the way your CFO owns the books. Not the person who resets passwords, and not the managed service provider who patches the servers, though you need both of those too. I mean somebody who decides which systems exist, how they talk to each other, what data is the official version, and what gets retired. Right now in most mid-market companies that job belongs to nobody, which in practice means it belongs to whichever vendor sold something most recently.
| Question | Unowned stack | Owned stack |
|---|---|---|
| Who picks a new system? | The department with budget left in Q4 | One person, against a written plan |
| Where does the real inventory number live? | Depends who you ask | The ERP, and everyone knows it |
| What happens when a key person quits? | A process quietly stops | It’s documented and automated |
| When does the month close? | Around the 20th | A few business days in |
| Who retires old tools? | Nobody, so you pay for all of them | The owner, on a schedule |
Who should that person be? Depends on your size, honestly. Above a few hundred million in revenue, it’s usually a real CIO. Below that, a full-time one is hard to justify and hard to hire, which is why I spend my days as what I call a rentable CIO. Our clients generally can’t afford me full time, but a slice of the right judgment goes a long way. If you want the plain-English version of that model, KORE1 also runs fractional CIO services.
Owning the stack also doesn’t mean doing the work yourself. The owner decides. Somebody else builds the integration, cleans up the data, and runs the migration, and that’s usually a contract hire for the project rather than a permanent seat, because a lot of this work has an end date. If that contractor ends up reporting to you, here’s how to manage developers when you’re not technical.
The Irrigation Company, Eight Months Later
You’re wondering how it ended, so here’s the short version, with the caveat that the boring parts took longer than the interesting parts, like they always do.
We didn’t buy a new ERP. The one they had was fine. We killed four of the eleven systems, including a freight tool that two people were still paying for on separate credit cards. The emailed PDFs went through an automated capture step that lands them in the ERP as draft sales orders, so Deb reviews them instead of retyping them, and she got her mornings back. The overflow warehouse became a real location in the system with real bins, which sounds boring and was the single biggest improvement to their stockout problem. Commissions moved into the ERP where more than one person can see the math. Getting everyone to actually work in it afterward was its own project, and I’ve written up how to drive ERP adoption after go-live for that stretch.
The owner still says they’re not a tech company. He says it less often, though. And the month now closes around the sixth business day instead of the 20th, which means he finds out whether he made money while he can still do something about it.
Nice.
None of that required an app, a data scientist, or a single use of the word “transformation” in a board deck. It required someone to own the list.
What CFOs Push Back With
So does my company need a CTO now?
Probably not, unless software is the thing you sell.
A CTO builds technology products. Most mid-market companies need someone who runs technology as an operating function, which is closer to a CIO, and below a certain size that can be part time. Hire the title that matches the job, not the one that sounds impressive on LinkedIn.
Who actually said it first?
Gartner’s Peter Sondergaard gets the most credit, from the firm’s 2013 Symposium keynote.
Marc Andreessen made the same argument in 2011 with “software is eating the world,” and Satya Nadella has said versions of it for a decade. Honestly, whoever said it first matters less than the fact that it keeps getting said by people selling technology, which is a good reason to be a little skeptical of the flashy version.
We’re a $60M distributor. Isn’t this a big-company problem?
It’s worse for you, not better.
Big companies have IT departments that absorb the mess. At $60M, the mess lands on three or four people who already have other jobs, and one of them leaving can stop a process cold. The Census data showing smaller firms trailing on AI adoption tracks with what I see on basic integration, too. The gap usually isn’t budget. It’s that nobody owns the decisions.
“Every company is a software company.” Same claim?
Close cousin, and more misleading.
“Software company” makes people think they need to write code. “Technology company” is broader and more accurate, because most businesses run on software they buy and configure rather than build. If a vendor pitches you the software version to justify a custom build, ask them to show you what off-the-shelf can’t do first.
Where does AI fit into this?
At the end of the line, not the beginning.
AI is a high speed idiot. It’s fast and it’s dumb, and it does exactly as well as the data and process you point it at. If your order data lives in four places and disagrees with itself, AI will confidently summarize the wrong answer. Clean up the stack, then point AI at simple, boring, well-defined tasks. There’s a longer version in where practical AI for business actually starts.
What’s the first thing to fix?
The manual line with the most names next to it.
Run the system count above. Whatever process depends on one person retyping data between two systems is both your biggest risk and usually your cheapest win. Automate that one, prove it works, then do the next. Resist the urge to replan the whole company in one go, because that plan will sit in a folder for a year.
You Already Are One
There’s no opting out. The phrase went from keynote slogan to plain description of reality somewhere around the time your customers started expecting to see stock levels online and your bank started asking for monthly financials by the tenth. You’re a technology company. The only open question is whether anyone is running the technology part on purpose.
Count the systems. Name an owner. Kill what nobody uses. Automate the retyping. Let’s graduate your tech stack from the 90’s to at least the 2000’s.
Did the count and hated the result? Hit me up. And if the fix needs people, whether that’s an integration developer for a project or a permanent systems lead, KORE1 has been recruiting technical talent since 2005 across 30+ U.S. metros, and you can talk to a recruiter there directly.
