Paul Lachance, Chief Industry Strategist & Technologist, Software Equity Group.
I spent a significant part of my development career working with languages such as Visual Basic, Pascal and C++. At that point, building software still required specialized knowledge.
Over the decades, I’ve watched the barriers to building software steadily fall, from more accessible programming languages to low-code platforms, and now, AI-assisted development and “vibe coding.”
Of course, AI is much broader than coding acceleration and generative AI. Machine learning, natural language processing, computer vision and predictive and prescriptive analytics have been part of software for years. What’s different today is the accessibility and sophistication of these technologies, and the pace at which they’re advancing. Today, someone with an idea and limited programming experience can build a working application pretty quickly.
That is an incredible technological shift, but it raises an obvious question for anyone building, investing in or acquiring software companies: If software is becoming easier and cheaper to build, where does enterprise value come from?
The ‘So What?’ Test For AI-Era Software Value
I think about that question from two perspectives. I’ve spent much of my career in technology and software, and now I evaluate software companies through an M&A lens. In a recent discussion with my team about what makes a software company durable in the AI era, one of my colleagues offered a simple test: “So what?”
You added an AI capability. “So what?”
You have an AI road map. “So what?”
Your employees are using AI every day. “So what?”
None of these, on their own, tell us much about the durability or enterprise value of the business. For that, we need to go a layer deeper. What does this company have that gives it real staying power?
What Sits Below The Waterline
An iceberg is a useful way to think about software defensibility. Everyone can see the user interface and features. Below the waterline may be sophisticated architecture and data structures, years of domain expertise, proprietary first-party data, deeply embedded workflows, compliance requirements, integrations, customer relationships and other capabilities that are much harder to replicate.
AI can make it easier to write code and build basic functionality, but true software development goes far beyond the code itself. It encompasses product design, architecture, integration, testing, security, deployment, scalability and ongoing maintenance.
And building the code is different from replicating everything that sits below the waterline. Those less visible, but critical, attributes have always contributed to the value of a software business. As the barrier to creating software comes down, they will play a larger role in separating durable companies from products that are relatively easy to reproduce.
Where Software Defensibility Lives
Historically, many durable software companies have become deeply embedded in one or more layers of a customer’s business. They hold critical data and context, support the workflows where work gets done, and help customers turn that information into decisions. You might think of those as systems of record, systems of action and systems of intelligence.
AI doesn’t make those layers irrelevant. If anything, it raises the bar for each of them.
Established systems of record and action aren’t disappearing. In fact, I think the market’s initial fear that frontier AI could quickly displace large categories of established software underestimated how deeply many applications are embedded in businesses.
At the same time, AI can expand the intelligence layer. We’re seeing another model emerge that creates value from the intelligence surrounding data, even when a company doesn’t own the primary system where that data resides.
I think of these businesses as domain intelligence companies. They pull information from multiple sources, normalize and enrich it, combine it with proprietary information or methodology, and apply deep domain expertise to produce intelligence customers can’t easily generate themselves.
AI Is Changing Where Software Value Comes From
AI may broaden our definition of valuable software rather than replace the software models that came before it. And the changes aren’t limited to the product itself. AI is also beginning to challenge some of the economics that have defined SaaS.
If software helps a customer accomplish the same work with eight people instead of 10, that’s real value. But if the software company charges per seat, it may have reduced its own revenue in the process. That may push some companies to rethink how they capture value, including through consumption-, usage- or value-oriented pricing models.
At the same time, AI can improve the vendor’s own economics. If a software company can onboard and support more customers without adding resources at the same rate, it can create operating leverage. That potential becomes another factor in how we think about a company’s future economics and enterprise value.
Can Established Software Companies Win In The AI Era?
The public markets have shown how quickly AI can change perceptions of software companies, and I’ve seen that scrutiny carry into M&A, as well.
But the conversation is becoming more nuanced. The initial reaction to generative AI seemed to assume that if software became dramatically easier to build, established software would become dramatically easier to replace. Those aren’t the same thing.
Buyers are still looking several years ahead. Will the software remain relevant as AI capabilities improve, new competitors emerge and customer expectations change? An AI road map alone won’t answer that.
I don’t interpret today’s environment as evidence that established software has lost its value. Capital continues to back systems of record and other deeply embedded software businesses. What has changed is the level of scrutiny and the differentiation between companies.
Established companies may have advantages of their own. They start with years of customer relationships, workflow ownership, domain knowledge, historical data and real-world context, then use AI to make those assets more valuable.
That’s where I think the enterprise-value conversation is headed. The strongest software companies will be those that can show not only what their technology does, but why customers will continue to need them, and how AI makes that position stronger rather than weaker.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
