Linda ParkSomebody Has to Decide What 'Software-Defined Everything' Means for Our Company
February 17, 2026
About three weeks ago, Derek forwarded me an article about software-defined everything with the subject line "thoughts??" He does this periodically - sends me something dense and vaguely alarming and expects me to come back with a verdict. The article was from a trade publication. It was fourteen paragraphs long and not once did it explain what a person at a real company should actually do about any of it. I told Derek it was fine. He seemed satisfied. I was not.
So I've been sitting with this topic for a few weeks now, and I have a real opinion about it, and the opinion is this: the "software-defined everything" wave is not overhyped in terms of the actual technology, but it is wildly overhyped in terms of how it's being communicated - and that gap is going to cost mid-size and smaller businesses real money when they start chasing it without knowing why.
What This Actually Is
The software-defined anything (SDx) market refers to the abstraction of physical hardware into programmable software systems that enable dynamic management, configuration, and optimization of IT, networking, and storage resources. That's the clean version. The longer version is that for decades, if you wanted your company's network to do something different, you went and physically changed hardware. A router, a switch, a server configuration. Now, increasingly, all of that is controlled through software layers. The hardware becomes, in theory, generic. The intelligence lives in code.
SDE technologies encompass key areas like Software Defined Networking (SDN), Software Defined Storage (SDS), Software Defined Data Centers (SDDC), and Software Defined Wide Area Networks (SD-WAN), enabling organizations to virtualize and centrally manage complex resources. Each of those is its own world. Each of those has its own vendors, its own acronyms, its own conferences where people in lanyards tell you it will change everything.
And here is the thing: it probably will change a lot. More than 62% of new enterprise IT investments in 2023 focused on software-defined components, compared to only 41% in 2020. That's not a blip. That's a direction. The global Software-Defined Networking market alone was estimated at $24.5 billion in 2023 and is projected to reach $60.2 billion by 2028, growing at a CAGR of 19.7%. The money is moving. The question is whether your company needs to move with it right now, next year, or not for another five years - and nobody in the marketing materials is helping you figure that out.
The Vendor Communication Problem Is Real and It Is Deliberate
I want to be direct about something. The reason "software-defined everything" is hard to parse is not because the technology is inherently obscure. It's because vendors benefit from the confusion. When you don't know exactly what you need, you tend to buy more than you need. You buy the enterprise tier. You buy the add-on. You buy the professional services engagement to explain the thing you just bought.
Gerald and I were at Costco last fall and he picked up a blender that had seventeen settings. He used two of them the whole first year. Software-defined infrastructure is like that, except the blender costs six figures and someone from the blender company comes to your office for four days to set it up and then sends you a follow-up invoice.
Adoption of cloud-native, software-defined infrastructure has led to a 27% reduction in manual IT tasks, while automation features in SDE platforms have helped reduce average incident response times by 34%. Those are real numbers and I believe them. But those outcomes don't land the same way for a 40-person distributor as they do for a regional bank with 600 endpoints. The research firms selling these market reports aren't really differentiating between those two companies - and neither are the platform vendors.
SMBs are let down at the implementation stage, while enterprises experience regret afterward - SMBs are more likely to blame their purchase regret on poor implementation than enterprises. That's the Capterra 2025 Tech Trends data. It matches exactly what I've seen. The problem isn't that small and mid-size businesses make bad decisions. It's that they make decisions without enough context, because the context being served to them is designed for a much larger organization.
Where the Trend Is Actually Solid
Here is where I'll give the SDx world its due. The networking piece - particularly SD-WAN - is probably the most immediately relevant for businesses that have multiple locations or a lot of remote workers. A 1 Gbps MPLS circuit runs about $1,439 a month. A managed SD-WAN setup covering the same bandwidth runs closer to $1,066. That's roughly 25% less, and that's before you factor in the agility difference - MPLS circuits can take months to provision; SD-WAN brings a new site online in days. If you have five branch offices still running MPLS circuits, and your people are spending half their time in Salesforce or Teams or whatever SaaS pile you've assembled, that is a legitimate problem with a legitimate software-defined solution. That's not a theoretical enterprise architecture project. That's Tuesday.
SDE solutions have reduced average IT operational expenditures by 23%, with downtime incidents declining from 7 hours to just 2.8 hours annually for organizations utilizing software-defined architectures. Again, real. But I want to be careful here because 23% of a large IT budget is a material number and 23% of a small one might not even cover the implementation cost. The math is not neutral.
The other piece that's genuinely worth paying attention to is security. If you've been stitching together your security posture with overlapping point solutions and prayers, a more unified software-defined approach to your network control layer is probably overdue. The centralized control plane in SDN lets you push a policy change to every node at once instead of logging into individual devices one at a time - which matters when you're responding to an incident at 11pm and you're the only one who knows the admin password. 61% of small businesses have experienced at least one cyberattack, and the answer to that number is not another SaaS tool sitting in a silo - it's something more architectural.
The Part That Nobody Talks About
The skills gap. I'm going to say it plainly: most companies implementing software-defined infrastructure at the mid-market level do not have people on staff who can fully manage it once it's in place. For SMBs, the problem is people - SMBs are 13% more likely to cite the availability of labor as a significant external factor shaping their business goals than enterprises, and they're also more likely to consider staff acceptance or training a top challenge when planning investments in new software.
I've tested enough platforms to know that the gap between "we deployed this" and "we are actually using this correctly" is enormous and almost never discussed in the sales process. I once spent an afternoon going through the documentation for a network management tool that our team had been running for eight months. Nobody else had read it. There were three features we'd been paying for and not using, and one misconfiguration that had been quietly affecting our traffic since month two. The vendor's onboarding process had not caught it. Neither had anyone else.
Most small businesses and many core midmarket firms lack formal IT budgets, using operating revenue to address urgent remediation needs and respond to new opportunities where possible. That is the actual environment in which "software-defined everything" is supposed to land. Vendors building for that environment need to reckon with it more honestly.
What I Think Is Actually Going to Happen
The Fortune 1000 is already deep into this. As of 2024, an estimated 73% of Fortune 1000 companies have integrated SDE frameworks to optimize their IT environments. That ship has sailed. The interesting question is what happens in the next three to five years as those architectures filter down through managed service providers into smaller businesses that couldn't have afforded to build any of this themselves.
That is the actual path for most of you reading this. Not a greenfield software-defined data center. Not a six-figure Cisco or HPE engagement. The managed service version of all this is coming - MSPs are already bundling SD-WAN with AI-assisted orchestration and multicloud visibility into monthly contracts that don't require you to hire a network architect. It will be priced more accessibly, and it will be the realistic entry point for companies that don't have a full IT department.
Tory - who I genuinely feel for on a personal level, given everything going on with him right now - made the point last month that everyone wants to lead with transformation but nobody wants to manage the transition. He was talking about something else entirely, but it applies perfectly here. The transformation story is easy to sell. The transition - which involves your existing systems, your existing staff, your existing vendor contracts - is where deals fall apart and budgets get eaten and people get frustrated.
If your company is somewhere in the middle of the market - 50 to 500 employees, multiple locations, a real mix of cloud and on-prem, some legacy stuff you haven't replaced yet - my honest read is this: you do not need to architect a software-defined everything strategy right now. You need to get very clear on where your current infrastructure is actually failing you, and then evaluate whether a software-defined approach to that specific problem makes sense. SD-WAN for branch connectivity? Probably worth looking at. Software-defined storage for your data center? Depends on how much of a data center you still actually run. Software-defined everything because someone at a conference told you the future is software-defined? No. That's how you end up three years from now with a half-implemented platform, a vendor relationship that's become complicated, and a team that's resentful because the tool never did what the slide deck promised.
The terminology matters less than the problem you're solving. I've been saying this about tools for years - about project management platforms, about lead generation software, about anything where the category gets hot and everyone starts buying category instead of solution. The same principle holds here, just with higher stakes and longer contracts.
The Question Someone Actually Has to Answer
My suggestion is boring but I mean it: before that meeting happens, make someone - an internal person, an MSP you trust, somebody - produce a one-page map of your current infrastructure. Where your traffic actually goes. What hardware you own versus what you're renting from a cloud provider. Where the friction is. Where the security gaps are. That document will tell you more about whether SDx is relevant to your company right now than any market research report, including the ones I've cited above.
I made Gerald's chicken casserole the night before I started writing this. He took a second helping without commenting, which in 31 years I've learned means he thought it was good. Sometimes the thing you've been doing for years still works fine and you don't need to say anything about it. Sometimes you do need to change it. The difference is knowing which situation you're actually in.
Know which situation you're actually in. Then decide.