Wall Street is far too pessimistic. When software stocks corrected 10-25% last week following the latest Claude and ChatGPT releases, analysts predicted a significant decline of enterprise software sales. Their reasoning? If companies can now create applications quickly and cheaply through AI code generation, why would they continue purchasing expensive solutions from leading enterprise software vendors?
That logic sounds compelling but is incomplete.
After 30 years in IT, I think they are overreacting—and missing 90% of the picture.
The Total Cost of Ownership Reality
The cost of creating an application is typically only 10-20% of the total cost of ownership. The lion's share? Long-term maintenance, enhancements, bug fixes, integration complexity, cyber security controls, compliance and the teams that implement them.
Companies do not buy CRM and ERP solutions because building them is expensive. They buy them to avoid the ongoing burden of maintaining, securing, and evolving them over time.
Organizations have limited bandwidth. They purchase third-party solutions for non-core functions so they can focus on their core business and revenue opportunities. This fundamental reality will not change, regardless of how cheap initial implementation becomes.
Investment banks, insurance companies, healthcare organizations and manufacturers are not going to build and support their own CRM, HR platform, ticketing system, or core financial application—no matter how cheap the initial implementation. The ongoing costs and distraction take away from their primary mission.
30 Years of Buy vs. Build Decisions
I have participated in countless buy versus build discussions throughout my career. In less than 10% of the cases, cost was a primary factor.
What mattered most? The need for functionality unavailable in the market, or proprietary solutions that provide strategic advantage. Where established third-party solutions exist with reasonable functional coverage in non-core areas, organizations overwhelmingly choose to buy.
The reason is simple: vendor products spread the costs of maintenance, enhancements, and strategic advancement across dozens, hundreds, or even thousands of customers. When you build a custom solution, there is only one customer—you—bearing all of those costs.
You also cannot outrun the market. Vendors, with their many clients, get a bird's-eye view of what the market requires next. Single-customer solutions risk tunnel vision regarding future requirements.
Where AI Can Create Real Disruption
There is one legitimate risk to incumbent software companies: dramatically lower barriers to entry for startups.
By dropping the cost of building a competing solution from $3 million and a 6-12 month timeline to a few thousand dollars and a couple hours, AI enables entrepreneurs to bring their vision to life far more easily. Many startups with superior ideas have historically failed due to lack of initial funding or running out of money before finding success.
AI will allow many more startup companies to take root and flourish within a greatly abbreviated timeline.
However, building your "superior" solution is only 10% of the work. Significant time, money, and energy are required to take that product to market and overcome entrenched solutions. We have all seen examples of nimble upstarts disrupting incumbents, but it remains a small percentage of the landscape. Winning enterprise trust, passing security reviews, building integration ecosystems and sustaining roadmap velocity will remain difficult tasks. Companies only take risks on new ventures when solutions are significantly more advanced and/or costs are much lower.
My Predictions
It will be fascinating to watch this unfold over the next few years. I predict a much more energetic startup environment as the time and cost of market entry plummets thanks to AI platforms.
I also predict far less disruption to incumbent enterprise software companies than the market is currently pricing in. Why? Because solution implementation cost is only 10-20% of the problem to solve.
What do you think? Are you seeing similar patterns in your organization? I would especially like to hear from CTOs and CFOs who are evaluating AI-generated internal solutions versus traditional enterprise software. Please share your perspective in the comments.