GEP-UVA Darden Report: Up to 95% of GenAI Supply Chain Projects Fail, Process Maturity Cited as the Real Culprit
GEP (an AI-driven procurement software vendor) and the University of Virginia Darden School of Business published a new report measuring AI readiness among senior supply chain executives. Of the 180 surveyed, more than half are deploying AI in some manner, but fewer than one in ten have scaled AI pilots into enterprise-wide operations. 74% of respondents have not even moved beyond the planning stage or set out a roadmap for proceeding. The study was launched explicitly to unpack the structural reasons behind the widely-cited “95% of AI investments are failing” sentiment that has been circulating across the industry.
GEP's global head of strategy Michael DuVall said researchers began hearing nearly a year ago that AI, robotics and related tech initiatives were stalling out and that clients had shifted into a “wait-and-see” posture. The project was led by Tim Laseter, professor of practice at UVA Darden, whose team drew on case studies from Amazon.com, logistics services provider C.H. Robinson and Harvard Business School, and surveyed C-suite, vice president, director and manager-level leaders across 12 industries. The headline finding: the culprit behind the lack of AI progress is not the technology itself, but the same lack of proper business processes that bedevils almost every new technology rollout. Most AI projects are driven from the top down but lose energy as they descend the organizational chart, with messaging becoming vague and everyday operating concerns intruding. On top of that, many companies are layering AI over old, broken processes; per Laseter, they treat the effort as a routine software installation rather than as an “operational transformation” that requires sharp change-management focus.
On the success side, the “performance elite” firms share a recognisable recipe: onsite PhD-level AI expertise coupled with process-savvy frontline staff and the routine use of automated data cleansing, real-time dashboards and digital audit trails. A “portfolio approach” also delivered value: leadership teams that nurse along multiple AI initiatives on different timelines spread the investment risk, and a win in one function becomes the platform for the next. Crucially, organizations that scaled AI fastest relied on a dedicated steering committee staffed by experts from multiple functions and disciplines; a third of those lacking such a structure had no systematic view of opportunities at all. Even the performance elite struggles to absorb AI into the organization, with stakeholder engagement and talent management still the weakest links — in Laseter's words, the next competitive advantage “will come from a better-prepared workforce, not a better model.”
Key Takeaways:
1. GEP + UVA Darden survey of 180 senior executives across 12 industries: more than half deploy AI in some form but fewer than one in ten have scaled it enterprise-wide.
2. 74% of respondents have not even moved beyond the planning stage, exposing the structural causes behind the widely cited '95% of AI investments are failing' narrative.
3. The real bottleneck is not the technology but business processes; AI is being layered over old, broken processes and treated as routine software rather than operational transformation.
4. Success recipe among 'performance elite' firms: onsite PhD-level AI expertise + process-skilled frontline + automated data cleansing, real-time dashboards and digital audit trails.
5. A third of organizations without a dedicated cross-functional steering committee have no systematic view of AI opportunities; the next competitive advantage will come from a better-prepared workforce, not a better model.
[44048]
