McKinsey: 32% of Companies Skipped Buying Software Because Agentic Coding Tools Could Build It

McKinsey's State of AI 2026 global survey, fielded May 4 through June 8, 2026 across 1,719 participants in 97 countries, found that 32% of organizations have decided against buying at least one software product or feature because it could be built internally using agentic coding tools instead.

The Build-vs-Buy Shift

This finding represents a specific, quantified data point on how agentic coding tools are directly affecting software purchasing decisions — nearly a third of surveyed organizations have made at least one concrete build-instead-of-buy decision attributable to these tools, rather than simply reporting general interest or experimentation.

Agent Scaling Varies Sharply by Company Size

• 40% of respondents from large organizations (annual revenue over $1 billion) report scaling AI agents across one or more business functions, up from 27% the prior year
• Respondents from smaller organizations reporting agent scaling remained flat at 22% year-over-year
• This divergence suggests scaling AI agents currently correlates with organizational resources and scale rather than broad-based uniform adoption

The Productivity-Profitability Gap

Eight in ten respondents reported that AI has improved their own personal productivity. However, the share of respondents reporting that AI has contributed to their organization's EBIT remained essentially unchanged year-over-year, at 37% — indicating a persistent gap between individually-experienced productivity gains and measurable organization-wide financial impact.

What This Means for Software Vendors and Buyers

For software vendors, the 32% build-vs-buy figure is a concrete signal that agentic coding tools have moved from developer productivity aids to a genuine competitive alternative to purchasing software for at least some categories of functionality — a dynamic worth factoring into product and pricing strategy. For enterprise buyers, the widening gap in agent-scaling rates between large and small organizations, combined with the flat EBIT-impact figure, suggests that AI agent adoption alone is not a reliable proxy for measurable financial return, and that scale and specific implementation approach matter more than adoption rate alone.