Artificial intelligence is shifting enterprise software economics by accelerating custom application prototyping, forcing chief information officers to weigh maintenance and security risks against traditional SaaS subscriptions.
For decades, commercial software subscription models offered predictable execution and maintained enterprise capabilities across predictable three-to-five-year roadmaps, according to former Intel CIO Motti Finkelstein cionews.com. Yet as AI tools make software development faster, enterprises face complex software portfolios featuring overlapping capabilities and rising subscription costs.
Finkelstein said that this era of predictability has ended, leaving executives operating in what Axios CEO Jim Vandeheli terms post-processable velocity cionews.com. Consequently, the build-versus-buy decision has evolved from a one-time procurement task into a continuous discipline of managing organizational optionality cionews.com.
Development productivity metrics present mixed signals. A 2025 randomized controlled study by METR found that experienced open-source developers using AI tools took approximately 19 percent longer to complete assigned tasks.

While natural-language programming tools enable rapid prototyping, enterprise applications demand rigorous security protocols. A 2025 Veracode report evaluating AI-generated code revealed that 45 percent of the tested samples introduced security vulnerabilities.
Internal infrastructure responsibilities expand significantly when organizations build applications instead of purchasing them. Maintenance obligations, vulnerability management, authentication, encryption, and regulatory compliance remain internal duties regardless of initial generation speed.
Finkelstein highlighted how architectural strategy shifted at Intel, where an enterprise architecture team eliminated 321 in-house applications and 1,400 commercial products, avoiding $13 million in costs cionews.com. Simultaneously, the company rebuilt its internal AskIT support tool on its own AI platform cionews.com.
That internal tool now handles 26,000 conversations for 11,000 employees while reducing basic user tickets by 30 percent, cionews.com reported. When platform users surged from 500 to 1,600 in a single month, Intel built Intel Forge and Intel GPT to establish governance guardrails for federated AI workflows cionews.com.
Capital One similarly developed VulnHunter, a set of AI agents paired with a falsification engine to hunt exploits directly in source code, addressing security threats faster than external vendors could respond cionews.com.
CIOs evaluating enterprise software portfolios must determine whether a capability offers strategic differentiation or simple departmental support. Standardized infrastructure and routine CRM workflows remain suited for commercial vendors who amortize development costs across thousands of customers cionews.com.
Specialized applications addressing narrow requirements represent immediate opportunities for internal development or extended platform usage. Organizations must calculate total cost of ownership, including integration, security, architecture standards, and ongoing maintenance.
Former Intel CIO Motti Finkelstein, who served from July 2022, led the Global IT organization spanning manufacturing IT, data analytics, architecture, and infrastructure for the semiconductor manufacturer.
His background includes prior roles as CTO of Strategy and Planning and CTO of the Americas at Bank of Montreal Capital Markets, where he focused on cloud, big data, and robotic process automation, plus earlier experience as a Major overseeing technology infrastructure in the Israeli Air Force computer unit.
The core question facing technology leaders is no longer solely whether custom code can be generated more cheaply than a commercial license. The decision centers on whether building, buying, or extending existing technology delivers the best business outcome over an application’s entire life.