AI Apocalypse for Software stocks?

The recent surge in Agentic AI and Generative AI has stirred considerable discussion about the future of software as a service.

Vibe coding, in particular, has lowered the friction involved in building applications. Today, a developer—or even a technically curious individual—can describe a product in a prompt and generate a working application. In many cases, this can be done without a dedicated front-end developer, back-end engineer, or UI/UX designer.

The results are often surprisingly functional. They may not be perfect. Depending on complexity, they can still be buggy. But they work.

Markets, as they often do, have reacted quickly. Over the past few quarters, several major software companies have seen meaningful erosion in their market values.

This naturally raises a question worth considering:

Is it really over for software?

What makes Software?

To think clearly about this question, it helps to step back and understand what software actually consists of.

Most applications can be broken down into three basic components: the back-end, the front-end, and the UI/UX layer.

The back-end is everything that happens behind the scenes. When you open an Uber app and request a ride, the app sends your GPS location, finds nearby drivers, calculates pricing based on supply and demand, allows the driver to accept the request, and processes payment. All of that logic takes place in the back-end.

The front-end is what the user interacts with. The buttons you click, the reviews you read, the ratings you see, the map where you enter your destination—all of these elements are part of the front-end.

Then there is UI/UX. This governs the visual and experiential layer: how the fonts look, which buttons stand out, the colors, the themes, and the overall look and feel of the front-end.

In essence, most software can be reduced to these three pieces.

Once we understand that, another question naturally arises.

Is it impossible for a new small team to disrupt the back-end, front-end, or UI/UX of an existing software company?

The answer, in most cases, is no.

There is nothing inherently sacred about these layers. Given sufficient time, talent, and capital, they can be rebuilt. And in the age of AI, that process may become even faster.

If a company’s advantage rests primarily on an extensive feature set, broader functionality, or a cleaner interface, that advantage may be more fragile than it appears.

The real durability of a software company does not lie in whether someone can replicate its product. It lies in whether customers can realistically and economically switch away from it.

Software was never about being the best

It’s tempting to think that the best software wins. But historically, that hasn’t really been the case.

The major software companies of the world—whether Microsoft, Oracle, or hybrid software-hardware companies like Cisco and Apple—did not dominate primarily because they had the best back-end or the most elegant interface.

They built ecosystems.

And ecosystems tend to be sticky.

As is usually the case with technology, advantages tend to narrow over time. Tesla, for instance, once appeared to produce electric vehicles that were two or three times better than the competition. That edge did not last forever. Competitors eventually caught up.

The same pattern has appeared again and again throughout technological history.

Software, in the long run, was never really about having the best back-end, front-end, or UI/UX.

What makes Software durable?

If superior technology alone is not the moat, then what is?

Historically, durable software companies benefited from other factors.

Switching costs.

Access to valuable data.

Stickiness driven by the difficulty of migrating systems.

Integration into workflows.

Network effects.

And in some cases, ecosystem lock-ins with hardware.

It’s also worth noting that the success of a software product has never depended solely on how good its features are or how intuitive its interface feels. Those things help, of course. But even historically, small teams have been able to replicate features.

AI tools may simply accelerate that process.

But when software is mission-critical, reliability matters more than novelty.

After all, no business wants a system that works 70 percent of the time and throws major bugs the rest of the time—even if that system was cheap and built in-house in five days.

Consider something as simple as a website. If your website disappears from search engines, if the fonts render incorrectly, if users can’t log in, or if contact forms stop working, the damage extends beyond the technical problem. It affects brand perception.

Great software requires constant maintenance. Every time new features are added—whether to the back-end, front-end, or UI/UX—new bugs inevitably appear. That is simply the nature of complex systems.

As a result, mature software platforms rely on continuous cycles of testing, fixing, and updating.

A one-and-done application can not replicate that.

Commoditized vs Mission critical?

Not all software is created equal.

Some applications are used casually and can be replaced easily.

Food delivery apps provide a simple example. If one platform doesn’t offer a good discount, users switch to another. They search for the same restaurant, build the same meal, and compare prices instantly.

This is what low switching costs look like.

Software that behaves like this will likely struggle in a competitive environment.

Mission-critical software operates differently. When a company has stored years of operational data inside a system—when that system connects to payroll, accounting, compliance, and internal workflows—switching becomes difficult.

Employees must be retrained. Data must be migrated. Processes must change.

That friction creates durability.

High net revenue retention over time is typically a great indicator of durability.

AI's impact on Software stocks?

From an investor’s perspective, AI should be evaluated through the lens of outcomes rather than excitement.

Ultimately, the value of any business is the future cash flows it will generate over its lifetime. If superior AI technology leads to superior and sustainable cash flows, then it deserves attention.

But the real question is not simply whether a company is using AI.

The real question is whether its AI is meaningfully superior to competitors—and whether that superiority can endure.

Going back to Tesla, there was a time when Tesla’s vehicles were dramatically better than competing EVs. Over time, however, competitors closed the gap.

The probability that technology alone—whether EV technology then or AI technology today—will function as a long-lasting moat is unlikely.

Technology advantages tend to compress.

From an investor’s perspective, this has practical implications.

If a portfolio is full of software companies that are unprofitable, lack defined moats, and were purchased at steep valuations, then AI may indeed represent a serious risk. It could potentially wipe out a large portion—if not all—of the market value of those businesses.

But investing without understanding what you are buying has always been a bad idea. It was a bad idea before AI, and it remains a bad idea to invest in businesses outside your circle of competence.

There are also specific metrics investors can track.

The most important ones are churn and net revenue retention.

How many users who signed up one or two years ago are still active today?

If they remain customers, that’s a strong indication of switching costs.

Of course, the next step is to understand why retention is high.

Is it because of ecosystem lock-ins with hardware?

Is it because the software bundles multiple mission-critical tools together?

Is it because years of operational data are embedded in the system?

Whatever the reason, an investor should understand it clearly and then further evaluate if the advantage is threatened by AI.

Two Buckets?

In the end, it may help to place software investments into two simple buckets.

Bucket A consists of mission-critical software businesses with high switching costs and strong net revenue retention.

Bucket B consists of speculative software businesses with no clearly defined or proven moat.

Ideally, a portfolio is concentrated in the first bucket.

If the switching costs are real and the moat is understood, those businesses may prove resilient even in the face of rapid AI-driven change.

But if a portfolio is concentrated in the second bucket, the situation looks very different. An environment where technological change is accelerating, that deterioration may not happen gradually.

It can happen very quickly.

By:

Paul Gray; Chief Executive Officer

Siddharth Singhai; Chief Investment Officer

Disclaimer

This White Paper expresses the views of the author as of the date indicated and such views are subject to change without notice. Ironhold Capital has no duty or obligation to update the information contained herein. Further, Ironhold makes no representation, and it should not be assumed that past investment performance is an indication of future results. Moreover, wherever there is the potential for profit there is also the possibility of loss. This White Paper is being made available for educational purposes only and should not be used for any other purpose. The information contained herein does not constitute and should not be construed as an offering of advisory services or an offer to sell or a solicitation to buy any securities or related financial instruments in any jurisdiction.

Certain information contained herein concerning economic trends and performance is based on or derived from information provided by independent third-party sources. Ironhold Capital Fund 1, L.P. (“Ironhold”) believes that the sources from which such information has been obtained are reliable; however, it cannot guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based.