New Research: Why Enterprise Agentic AI Stalls Before It Scales

Teradata/Teradata-commissioned Wakefield Research (n=1,000) finds only 7% of global enterprises have reached the “operationalizing” stage for agentic AI, while 68% remain in “experimenting” or “developing” due to context fragmentation and data governance gaps. Despite 90% of senior leaders expecting higher investment in the next 12 months, 63% report little to no ROI so far, and 40% of pilots fail to reach production because infrastructure wasn’t built for autonomous use. The study’s Agentic AI Maturity Index frames a path forward via “Autonomous Knowledge” (context, lineage, governance) to enable enterprise-level workflows, implying incremental but not immediately market-moving relevance for TDC.

Analysis

This reads more like category creation than fresh information: Teradata is trying to reframe AI spend from “model enthusiasm” to “data plumbing,” which is directionally right but only monetizes if CIO budgets migrate from pilots to architecture work. The nearest beneficiaries are not necessarily the most visible AI app names; they’re the control-plane vendors that sit on data governance, lineage, access control, and hybrid deployment. That favors broad infrastructure owners more than pure agentic-software plays, because the bottleneck is trust and integration, not model quality.

The immediate market risk is that this is a narrative event, not a revenue event. For TDC, the trade only works if the survey converts into pipeline, then bookings, then cRPO/ARR over the next 1-2 quarters; otherwise it’s just marketing noise and any pop should fade. In contrast, if enterprises really do shift from experimentation to production, the second-order winners are data platform and cloud vendors with larger ecosystems and better distribution, while point solutions that assume autonomous workflow adoption may see longer sales cycles.

Contrarian angle: consensus likely overestimates how quickly agentic AI budgets translate into P&L, but underestimates the budget reallocation toward governance and data modernization. That said, TDC is not the cleanest expression because it needs proof of share gains, not just agreement with the thesis. Falsifiers are simple: no acceleration in bookings or RPO in the next two quarters, or management commentary showing the AI narrative is not changing deal size / win rates.

Key Decisions for Investors

• Do not chase TDC on this PR alone; treat any 5-8% post-news strength as a fade unless the next earnings print shows bookings/cRPO acceleration.
• Watch SNOW and ORCL as cleaner 6-18 month beneficiaries of enterprise data-governance spend; initiate only if upcoming guidance or partner commentary confirms a shift from pilots to production.
• Relative-value idea: long data-infrastructure basket (SNOW/ORCL/MSFT) vs short high-multiple AI application software over 1-3 months; thesis is budget migration from agents to the data layer.
• Set a hard alert for TDC: if the next quarter does not show an inflection in new bookings or remaining performance obligations, the AI narrative should be discounted rather than capitalized.