Tom Snyder: What the market knows about the future of software

Over the past week, I have been watching the earnings reports from big tech, as I do regularly as background for this column. Major software companies have been beating expectations. Revenues are up. Guidance is solid. And yet, their stock prices are sliding.

That’s not how markets usually behave. Wall Street doesn’t punish success unless it suspects something deeper is changing. When fundamentals look good but confidence erodes anyway, it’s often a sign that the future no longer resembles the past.

For the AI behemoths, the “excuse on the street” is that investors are concerned about the rapidly increasing budgets for AI infrastructure. As big as these giant companies are, they are taking on hundreds of billions of dollars of debt to build out data centers. That their stock value is reflecting doubt in the shadow of these investments makes sense. And that many of these investments have an almost incestuous circularity, with the major players all investing in each other, gives logical reason for pause.

But the much more interesting story, in my opinion, is to look at what is happening in enterprise software. Companies like Salesforce are showing strong earnings and still losing significant stock price erosion. I believe this is not due to a cyclical slowdown or a temporary loss of faith. It’s the early signal of a structural shift. One that suggests the enterprise software industry, as we’ve known it for the last three decades, is running headlong into a great disruption.

The market is flashing a warning light

Look more closely at the recent earnings cycle and the pattern sharpens. Microsoft posted strong results, with cloud and AI revenues continuing to climb. Salesforce beat expectations again, pointing to improved margins and disciplined cost controls. Atlassian’s revenues are up 23% while its stock is down 7%. Palantir’s stock is down in 2026 despite achieving 78% year-over-year earnings growth.

Many enterprise software firms told similar stories: steady demand, resilient customers, improving efficiency. And yet, across much of the sector, stock prices drifted down or failed to respond.

Now for full disclosure, I’ll admit that I’m not a huge fan of the stock market. I believe that company valuations are complete fiction, no longer actually tied to the real value of a company’s physical assets, human capital, bank account balance or intellectual property. We live in a hype-driven society where high-speed, bot-driven trading moves the value of companies minute by minute based on social media opinion, when actual business value changes slowly over months or years. The concept of auditing the actual value of a company’s assets and liabilities is almost a quaint concept at this point.

But I do think this recent market behavior is interesting. I may complain that markets are noisy, but they remain directionally informative. I believe in this case that the market isn’t reacting to how well these companies are running their businesses. It’s reacting to the growing possibility that the business itself is becoming less essential. That disconnect matters. Markets don’t ignore good news without a reason. When revenues are solid but valuations soften, it usually means investors believe the next chapter will look very different from the last one.

This isn’t about a single company or a single quarter. It’s a broader loss of conviction that today’s enterprise software leaders will own tomorrow’s growth.

Why is this happening?

At first glance, none of this feels intuitive. If enterprise software companies are executing well, expanding margins and embedding AI into their products, why isn’t the market rewarding them?

The answer isn’t hidden in any one balance sheet. It sits at a deeper structural level, and it has less to do with how well these companies are running their businesses and more to do with what kind of technology they are now producing. To understand what’s unfolding, it helps to revisit a framework that has explained nearly every major technology transition of the last century.

Clayton Christensen’s warning revisited

In The Innovator’s Dilemma, Clayton Christensen drew a critical distinction between sustaining innovations and disruptive innovations. Sustaining innovations make existing products better. Faster. More capable. They align perfectly with what current customers want and what established business models reward. Market leaders almost always excel at them.

Disruptive innovations take a different path. They begin by doing many things poorly. They underperform on traditional metrics. They often look like toys compared to the polished incumbents they eventually replace. And because of that, they’re easy to dismiss.

The catch is that disruptive technologies don’t stay inferior for long. Once they cross the threshold of being good enough, they tend to improve rapidly — and when they do, entire categories can collapse with stunning speed.

This pattern has repeated itself again and again, long before anyone talked about artificial intelligence.

The birth of enterprise software

Once upon a time, work was done with specialized tools. Typewriters for writing. Calculators for math. Drafting tables and T-squares for engineering drawings. Each tool was refined relentlessly. Typewriters became faster and more ergonomic. Calculators gained memory, programmability and specialized functions. Drafting instruments grew more precise.

Then the personal computer arrived, and it was terrible at all of those jobs.

Early PCs couldn’t match the speed of a good typist, the reliability of a high-end calculator, or the precision of professional drafting tools. Even when I was in college, I was still required to buy an expensive calculator for my engineering courses, despite having a computer on my desk. The interface was clumsy. The software was immature. Word processors were primitive. CAD tools barely existed.

But the computer had one overwhelming advantage. It could eventually do all of those things.

Once PCs became good enough for writing, calculating and designing, entire categories of tools disappeared. The computer didn’t just replace them, it absorbed them. And out of that absorption came the rise of modern enterprise software.

CRM systems replaced rolodexes and spreadsheets. ERP programs replaced ledgers and manual reconciliation. Enterprise software became the digital operating system of the firm. Since then, the industry has lived almost entirely in the world of sustaining innovation, adding features, improving performance and refining platforms built on that original computing revolution.

Enterprise software vendors today are masters of sustaining innovation. Every release adds more automation, more analytics, more features layered on top of the same fundamental model. The dashboards get prettier. The workflows get tighter. The licensing gets more creative.

But the abstraction hasn’t changed. It is still enterprise software, now with AI bolted on.

The problem is that this is just an evolutionary improvement. The software is still capable, and customers are still asking for improvements. But a new paradigm is emerging that is so immature that enterprise software companies are discounting it -- exactly as Christensen would predict. Soon, however, the new paradigm will be “good enough” and it could be game over for traditional software companies.

When software starts writing itself

So what is the new pattern emerging inside organizations? Teams are using large language models to build their own tools. Not toy demos. Not hackathon experiments. But real internal systems. CRMs tuned to a specific sales motion. Operations dashboards built around a company’s actual data flows. Compliance tools shaped by how work really gets done, not how a vendor imagined it might.

This is often dismissed as “vibe coding,” and to be fair, that criticism isn’t entirely wrong. These systems are rough, they break, they lack polish. And they definitely make CIOs and CISOs nervous. But that’s exactly the point.

In Christensen’s framework, this is what early disruption looks like. New tools that underperform on the metrics incumbents care about, while excelling on dimensions the market hasn’t fully learned to value yet. The real comparison isn’t between an AI-generated CRM and Salesforce today. It’s between a custom, AI-built system and a generic platform once the AI crosses the threshold of being good enough.

The power of “good enough”

That threshold is arriving faster than most people expect. Once AI-generated software becomes reliable enough, secure enough, and explainable enough, the value proposition flips almost overnight. A tool built specifically for your business, using your data, shaped around your workflows, becomes impossible to justify replacing with a generic alternative.

There’s no per-seat pricing logic. No feature gating. No vendor roadmap dictating how fast you’re allowed to evolve. When the business changes, the software changes with it. At that point, traditional enterprise software stops looking like infrastructure and starts looking like friction. Not because it’s bad, but because it’s misaligned.

Why the market is losing patience

This brings us back to the recent earnings reports. Revenue numbers are backward-looking. They reflect contracts signed years ago, renewals locked in by switching costs and organizations still operating under old assumptions.

Stock prices, on the other hand, reflect a forward-looking mindset. And what investors increasingly see is a future where the moats around enterprise software are shallower, margins are under pressure, and lock-in is far less durable than it once was.

This isn’t panic. It’s pattern recognition. The market understands something that earnings calls can’t easily admit: when software becomes something you generate instead of something you buy, the entire industry has to be rethought.

The end of software as a product

This isn’t the end of software. Far from it. It’s the end of software as a product.

In the coming years, software won’t live in licenses and modules. It will live in agents, workflows and continuously adapting systems. Capability will be assembled on demand, not procured through procurement cycles. The winners won’t be the companies that add the most features. They’ll be the ones that make it easiest for businesses to turn intent into working systems.

Enterprise software giants didn’t fail. They succeeded spectacularly in a world that’s now giving way to a new one. Disruption rarely announces itself with collapse. More often, it shows up as a quiet loss of confidence. Solid numbers. Slipping stock prices. And a growing sense that the future is being written somewhere else.

The next generation of enterprise software won’t be installed. It will be imagined, and then built, by machines that finally understand your business as well as you do.