McKinsey’s latest State of AI report reveals a critical shift for developers and CTOs: agentic coding tools are fundamentally altering procurement strategies. Nearly one-third (32%) of surveyed organizations have decided against purchasing specific software products or features because they can now build the functionality in-house using agentic coding tools. This isn't just a productivity boost; it's a direct threat to the bottom line of SaaS vendors who assumed AI would only enhance their own products, not replace them.

The Build vs. Buy Reversal

The data shows that this trend is most pronounced in technology, healthcare, professional services, and energy sectors. While chatbots remain the most widely scaled AI tool at 47%, software coding agents are catching up, with about two in ten respondents reporting enterprise-wide scaling. For large enterprises with over $1 billion in annual revenue, the adoption rate for scaling these agents jumps to 31%. This suggests that big tech and large incumbents are leveraging internal dev teams and AI agents to bypass external licensing fees, potentially reshaping the enterprise software market's revenue model.

The ROI Gap and Cost Constraints

Despite individual productivity gainsβ€”80% of respondents say AI improved their personal outputβ€”enterprise-level financial impact has stalled. Only 37% of organizations attribute positive EBIT impact to AI, a figure unchanged from last year. Furthermore, cost is becoming a real bottleneck for infrastructure teams. About 20% of respondents report that AI-related operating costs, including token usage, have constrained their AI usage. High performers, who make up just 6% of respondents, are spending more than twice as likely to allocate over 15% of their ICT budget to AI, indicating that true ROI requires heavy, sustained investment rather than casual experimentation.

Key Takeaways

  • 32% of organizations are cancelling software purchases to build features in-house with coding agents.
  • Enterprise EBIT impact from AI remains flat at 37%, despite 44% of orgs scaling AI across the enterprise.
  • 20% of respondents cite AI operating costs (tokens) as a primary constraint on usage.
  • High performers are 3.3 times more likely to fundamentally transform their business with AI within three years.

The Bottom Line

If your SaaS product doesn't offer a moat that coding agents can't easily replicate, you're in trouble. The era of buying generic enterprise software is ending; the era of building specific, AI-generated internal tools is beginning.