The AI software freakout is a massive overreaction. Here's why.

• AI-driven software disruption fears are overblown; SaaS firms may thrive with AI.
• Enterprise DIY software projects often fail, benefiting established SaaS vendors.
• AI startups struggle against SaaS incumbents due to scale and integration challenges.

The recent AI-inspired software meltdown is an enormous overreaction. SaaS companies are going to do just fine. In fact, they might see their businesses boom as a result of AI.

After working for decades in the IT and cloud sectors, I know what it takes to build, sell, and maintain complex enterprise software. And it's clear to me that AI is not the threat that many investors fear.

Let's look at the three main concerns that the market is obsessing over. All of them have shortcomings that will keep established SaaS vendors doing well.

Enterprise DIY

Software development has been transformed by AI. It's a near-perfect use case for generative AI: applying established patterns to a well-bounded use case, resulting in incredible productivity improvements.

This coding revolution has led some commentators to predict that companies are going to write their own software, rather than buy it from SaaS vendors.

This makes me wonder if these commentators have ever actually spent any time working in enterprise IT. Even if one accepts that software creation is much easier now, bringing a full software product to market requires much more than code:

• Domain expertise (regulatory requirements, industry practices, supply chain expectations).
• Education (marketing, presales support, prototype creation).
• Support (upgrade help, customer use case enablement).
• Product management (selecting user personas to address, feature definition and prioritization, prioritizing customer groups to address).
• Financial negotiation (special deals for volume, discounts for public endorsement, and so on).
• Legal 'immunization' (indemnification, public policy advocacy to shape process guidelines).

Geoffrey Moore, a renowned management consultant and organizational theorist, wrote a book called "Crossing the Chasm" on what mainstream enterprises need to adopt new technology. He never mentioned the cost of writing software code as a gating factor.

Over the course of my long career, I have witnessed countless DIY failures by enterprise IT organizations that fail to understand the difference between an internal software project and a real product. I can already see another wave of failures, fueled by misplaced AI coding enthusiasm.

AI startup disruption

AI prognosticators see cheaper startups displacing large software incumbents. This overlooks the reality that established SaaS providers already have smaller, cheaper competitors yet somehow remain dominant. The challenges for new entrants to a software sector include:

• The obvious ones inherent to any digital sector, such as a lack of network effects and scale.
• Less obvious ones like the need for geographic vendor reach ("Do you have local, native-language consulting available in XYZ country?").
• Custom integration enablement for large enterprises. Big companies often demand bespoke software setups.
• Weird contractual requirements difficult for a small vendor to support. Again, big customers often want special treatment.

It's incredibly difficult for a small startup to displace an incumbent vendor. As Clayton Christensen observed in "The Innovators Dilemma," innovative startups usually begin by solving use cases incumbent vendors are unable or unwilling to serve.

Left unaddressed in this scenario is why incumbents wouldn't just use AI themselves to improve engineering efficiency to address any price pressure from smaller new rivals.

Going vertical

This is the idea that AI model companies will extend their nascent software products into vertical offerings, thereby killing off incumbent vendors.

OpenAI made a big splash with a healthcare initiative, for example. Anthropic caused a bunch of software stocks to drop with its plugins.

It's understandable why these companies launched these initiatives. Many industry-specific software companies are highly profitable, so AI labs would love to get a piece of this business.

Pursuing this, while working on other initiatives, could spread AI labs too thin, though. Startups often fail due to a lack of focus, and this is especially apropos for AI model makers. They face enormous, unprecedented opportunity, and getting distracted by bright, shiny vertical SaaS offerings is a terrible idea.

Going back to the DIY section above, there's huge complexity and cost to shipping and maintaining real enterprise software. Now multiple this by all the various industry verticals that exist, such as healthcare, financial services, and manufacturing. Addressing the idiosyncratic requirements of each sector would require huge numbers of employees, along with management time and attention. The model makers are already growing at breakneck speed; trying to add enough people to become vertical software providers would be a Sisyphean task.

Beanie Babies

If I were advising these boards, I would argue for focus: win the horizontal AI model layer in what is likely to become a small oligopoly.

And yes, expand AI coding capabilities that lower the cost of development and increase the global population of software creators. That dynamic could trigger Jevons Paradox — cheaper software leading to vastly more of it — enriching the model providers without forcing them into every software vertical.

The SaaSpocalypse will, in retrospect, come to be thought of like the Beanie Babies mania: a short-lived phenomena that now seems inexplicable to comprehend.

Bernard Golden is CEO of Navica, a Silicon Valley-based technology analysis, consulting, and investment firm.