The AI investment narrative has been remarkably one-sided. Capital has poured into chips and infrastructure, while investors have aggressively discounted the future of software companies and technology services firms. The assumption appears to be that AI will make software development and implementation so efficient that these industries will steadily lose relevance.
However, I believe that conclusion misunderstands what is actually happening inside enterprises. Like Mark Twain’s famous observation that reports of his death were greatly exaggerated, the oft-discussed demise of software and technology services is also being misrepresented. AI will fundamentally reshape both industries, but it is creating two distinct markets, rather than eliminating one. Understanding that distinction is critical for investors, technology providers, and enterprise leaders alike.
The Traditional Technology Stack is Not Going Away
In previous columns, I have argued that AI dramatically improves software engineering productivity, but productivity is not the same as elimination. Writing code faster does not get rid of the need to understand complex business processes, architect systems, or manage enterprise technology estates.
Large enterprises are simply not going to replace decades of investment in their existing technology stacks anytime soon. The risks are too great, and the effort required is often underestimated. Before AI technology can be implemented as replacement applications, organizations must first understand every function their existing systems perform in sufficient detail to describe them. Building that understanding remains an enormous undertaking.
As a result, I expect the traditional technology stack to continue evolving, rather than disappearing. These platforms may no longer attract the same level of investor excitement as emerging AI-native technologies, but they will continue to generate value and experience modest growth for years to come.
Task Compression Does Not Mean Market Collapse
The same misunderstanding exists in technology services. There is no question that AI creates meaningful productivity improvements at the task level. Coding, testing, documentation, and many development activities can now be completed more efficiently. I refer to this as “task compression”, where the revenue associated with an individual task declines because the work can be completed more rapidly.
A year ago, I believed this would significantly reduce the overall market for software development and technology services. After watching the market evolve over the past 12 months, I have changed my view.
Task compression is only one factor shaping demand. Other forces continue to support spending on maintaining, extending, securing, governing, and modernizing enterprise systems. Those forces appear strong enough to offset much of the productivity improvement. Instead of seeing a shrinking market, we are observing a traditional services market that remains relatively stable and may even experience modest growth.
The Real Growth Story is AI Native Services
At the same time, an entirely new market is emerging. The native agentic AI environment is fundamentally different from extending today’s enterprise applications. Rather than maintaining an existing “house”, so to speak, organizations are continually redesigning their architecture while they are living in it. Agents, ontologies, orchestration layers, governance frameworks, and supporting infrastructure will all evolve continuously.
This creates substantial opportunity for technology providers. These environments require ongoing engineering, implementation, monitoring, and refinement. They demand close collaboration between business teams and technical specialists. Instead of reducing dependence on third-party expertise, native AI is likely to increase it. For engineering firms and technology service providers, this represents a fast-growing market with significant long-term potential.
Two Markets Require Two Operating Models
One of the biggest mistakes I see today is treating these two markets as though they operate the same way. Maintaining and extending the traditional technology stack continues to fit existing delivery models built around offshore talent, Global Capability Centers, and established managed services relationships. AI will improve productivity within those models, but it does not fundamentally change how they operate.
Native agentic AI is different. These projects favor closer collaboration, greater business intimacy, and more engineers working alongside business teams. They require operating models that can accommodate constant change rather than long periods of stability. Pricing models will also need to evolve, with stronger links between commercial arrangements and business outcomes rather than simply labor inputs.
Perhaps most importantly, software and services become far more tightly integrated than they have been historically. The industry may ultimately need an entirely new category that combines technology products with engineering services in much closer partnership.
Separating the Old from the New
The temptation is to assume that adding AI capabilities to an existing application automatically transforms it into a native AI business. It does not.
Most organizations can enhance existing technology stacks with AI while continuing to operate under familiar governance models, sourcing approaches, and delivery structures. The more fundamental transformation occurs only when organizations begin building AI-native operating environments that continuously evolve alongside the business itself.
That distinction has important implications for both investors and enterprise leaders. For investors, I believe current market sentiment has become too pessimistic toward software and technology services companies. These industries are changing, but they are not disappearing.
For enterprise leaders, success depends on recognizing that the traditional technology estate and the emerging agentic environment require different operating models, different commercial relationships, and different expectations. Trying to force one model onto the other risks creating unnecessary cost, complexity, and disappointment.
The future is not a story about one market replacing another. It is a story about two technology markets growing side by side, each with its own economics, operating model, and opportunities. Understanding this will place enterprises and providers in a better position to truly leverage AI’s potential.
