Updated
Updated · O'Reilly Media · Jun 17
Enterprises Abandon Internal AI Builds as In-House Share Falls to 24%
Updated
Updated · O'Reilly Media · Jun 17

Enterprises Abandon Internal AI Builds as In-House Share Falls to 24%

3 articles · Updated · O'Reilly Media · Jun 17

Summary

  • Enterprise AI teams are being warned that internal “agent platforms” are usually mis-scoped projects, because memory, governance, evaluation and orchestration each behave like separate product categories rather than add-on features.
  • Menlo Ventures data underpins the case: the share of enterprise AI solutions built internally dropped to 24% in late 2025 from 47% in 2024, suggesting the build-versus-buy market flipped within 12 months.
  • The biggest hidden costs sit in agent-specific requirements such as persistent memory, action-level governance and trajectory-based evaluation, with 78% of executives lacking confidence they could pass an AI governance audit within 90 days.
  • Orchestration is also still unsettled across competing frameworks and emerging protocols, making custom stacks expensive to maintain as models, tool standards and design patterns keep shifting.
  • The article argues companies should build business-specific agents on top of bought components, especially with the EU AI Act fully enforceable for high-risk systems in August 2026 and Gartner expecting 40% of agentic AI projects to be canceled by 2027.

Insights

Why are 40% of corporate AI agent projects doomed to fail before 2028?
Your new AI agent has 'excessive agency.' How do you prevent it from going rogue?

76% of Enterprise AI Now Vendor-Built: The Rapid Shift, Risks, and Strategies for Sustainable Adoption (2025-2026)

Overview

In 2025, enterprise AI adoption experienced a dramatic and rapid transformation. Companies shifted sharply from building AI solutions internally to purchasing them from external vendors, with only 24% of use cases developed in-house compared to 47% the previous year. This change marked a clear preference for external solutions, as 76% of AI deployments were vendor-supplied. The move away from resource-intensive internal development allowed businesses to access specialized AI capabilities more quickly and efficiently, signaling a pivotal moment in how enterprises approach AI integration and setting the stage for a new era of vendor-led innovation.

...