72% of Organizations Say Process-Related Challenges Have Caused AI Initiatives to Fail and it is Costing Them Millions

Camunda research found that 72% of organizations have had AI initiatives fail because of process-related challenges, at an average cost of $1.55 million per business. A further 72% of respondents said AI costs could spiral without better process management, highlighting execution and cost-control risks in enterprise AI deployments.

Analysis

This is a weak near-term trading signal, but it reinforces an investable distinction within enterprise AI: model spend is increasingly constrained by the cost of integrating workflows, data permissions, exception handling, and human approval loops. Over the next 1-3 quarters, this favors incumbents with embedded systems of record and workflow control planes—NOW, CRM, SAP, ORCL, and MSFT—over standalone application-layer AI vendors whose ROI depends on customers rebuilding fragmented processes first.

The second-order effect is a longer enterprise sales cycle for AI copilots and agents, not necessarily lower AI budgets. CIOs are likely to redirect portions of experimentation budgets toward process mining, integration, governance, and observability; this supports NOW and SAP’s process/workflow positioning, ORCL’s data and application stack, and potentially DDOG/ESTC only where AI deployments create measurable monitoring and data-governance needs. It is less favorable for high-multiple software names priced for rapid seat-based AI monetization without a clear implementation-services or workflow attachment.

Consensus may be too focused on inference-cost deflation as the route to AI ROI. The binding constraint for large enterprises is often organizational redesign rather than token cost, which means reported AI revenue can remain lumpy even as usage grows. Falsification: broad-based software earnings showing AI products converting from pilots to paid production deployments with meaningful net-revenue-retention uplift, without corresponding professional-services growth or elongated implementation durations.

Company-sponsored survey evidence is not independently sufficient to establish sector-wide demand deterioration. Treat this as a diligence prompt ahead of the next enterprise-software earnings cycle: monitor backlog conversion, professional-services utilization, remaining-performance-obligation growth, and management commentary on pilot-to-production conversion rather than taking a directional sector position solely on this release.

Key Decisions for Investors

• Maintain a 3-6 month quality tilt toward NOW and SAP versus unprofitable application-software AI beneficiaries: both can monetize workflow redesign and governance before autonomous-agent deployments scale. Reassess if subscription backlog/RPO growth decelerates materially or AI attach rates fail to appear in the next two earnings reports.
• Use CRM as a watch item rather than a fresh AI-monetization long: validate whether Agentforce production deployments translate into incremental recurring revenue rather than services-heavy pilots. A clean catalyst would be disclosed paid-agent growth and durable margin expansion at the next earnings update.
• Avoid broad short exposure to software based on this item alone. If subsequent earnings calls reveal elongated AI implementation cycles across multiple vendors, consider a 1-3 month pair trade long IGV constituents with workflow/system-of-record exposure (NOW, ORCL) versus a basket of higher-multiple, AI-narrative application software names; require valuation and revenue-exposure data before specifying legs.
• Monitor Accenture (ACN) and Cognizant (CTSH) for a 6-18 month services spillover: process remediation can enlarge implementation demand, but only initiate on booking acceleration because fixed-price delivery risk can offset revenue upside.