Companies are investing heavily in AI for supply chains, yet most networks still rely on fragmented systems, inconsistent data, and manual oversight. New research from Gartner and Infor shows that only a small share of organizations are redesigning workflows around AI orchestration, while the majority remain constrained by integration costs, governance gaps, and limited trust in autonomous decision-making.
AI Ambition Outpaces Operational Readiness
Gartner surveyed 140 senior decision-makers in late 2025 and found that only 17% are actively rebuilding supply chain workflows around AI-enabled orchestration. The remaining 83% are limiting AI to discrete use cases or layering it onto existing processes, which keeps legacy decision patterns in place even as new tools arrive.
Gartner describes the destination as AI-powered orchestration that can monitor network events, simulate response options, and support accelerated human and machine decision-making. In that model, control infrastructure behaves like an enterprise nerve centre, linking planning, logistics, procurement, and finance around shared data and coordinated response. The gap between that vision and current practice is explained less by lack of interest than by structural constraints.
Multiple studies point to the same friction points. Infor’s global AI Adoption Impact Index reports that 80% of business decision-makers believe their organization could manage an AI implementation, yet 49% remain stuck in the early stages of deployment, often unable to move beyond pilots or partial rollouts. The tension between stated confidence and actual progress shows up in where the blockers sit: 36% cite data security, sovereignty, or compliance as the primary barrier, 25% point to insufficient in-house AI expertise, and 23% struggle to define clear business value.
Gartner highlights a parallel set of operational hurdles. Many organizations still wrestle with basic master data alignment, so the inputs that feed planning and execution tools lack consistency across regions and functions. Data from suppliers, logistics partners, and other ecosystem players is often incomplete or unreliable, which undermines AI models that depend on accurate external signals. Process maturity is another constraint: workflows, roles, and standard data models are not always defined tightly enough for algorithmic orchestration to run at scale.
Vendor fragmentation introduces an additional layer of complexity. AI-enabled orchestration typically spans planning, visibility, transportation, and analytics solutions that were procured and implemented at different times. Gartner notes that there is no single one-stop platform that covers end-to-end orchestration for most enterprises today. Integrating these pieces is expensive and slow; industry estimates cited by Infor put integration spending at roughly 30–40% of total technology budgets, which forces difficult trade-offs between new capabilities and the plumbing required to connect them.
Human capability and trust sit behind many of these numbers. Infor’s research shows that nearly half of AI-generated insights and workflows currently require manual review by subject matter experts before they are allowed to drive regulated or high-risk processes. Roughly one-third of respondents express discomfort with autonomous agents executing critical tasks without oversight. Gartner reinforces that AI is not displacing operational judgment; organizations still need employees who understand their networks deeply and can work with AI outputs over time, rather than deferring to them blindly.
From Visibility Projects To Decision Systems
The persistence of these barriers does not diminish the scale of the opportunity. Gartner’s analysis argues that the capabilities underpinning AI-powered orchestration will deliver transformative business value once they are in place, including faster disruption response, more precise trade-offs across cost, service, and carbon, and tighter linkage between network decisions and financial outcomes. The challenge is sequencing investment so that foundations and orchestration advance together, rather than waiting for a perfect platform to arrive.
Recent product moves from vendors point toward how this might evolve. Infor has introduced an Agentic Orchestrator layer designed to coordinate ‘Supervisor Agents’ across multi-step workflows, with industry-specific task agents plugged in for functions that range from planning to deployment. These supervisors maintain context across connected tools, escalate anomalies, and keep a human in the loop where required. The same vendor is pushing open standards such as its Model Context Protocol to simplify how AI systems access and act on data across different applications, including non-Infor environments. That approach targets two of the hardest problems identified by Gartner and Infor: interoperability and observability.
On the ground, AI is already moving into execution in narrow but meaningful ways. Infor’s new Velocity Suite for warehouse management applies machine learning to optimize pick paths, guiding workers along faster routes and reducing travel time. Other providers are testing similar ‘copilot’ patterns in transportation planning, inventory allocation, and supplier collaboration. These deployments remain constrained in scope, yet they illustrate a path forward: embed AI where the data is mature, governance is clear, and impact is measurable, then extend outward.
The research also shows that expectations are shifting toward more autonomous behavior once those foundations are secure. Infor reports that 32% of respondents rank the ability for AI to perform tasks autonomously as a top-three priority for long-term success, even as many remain wary today. At the same time, 87% say fixed and predictable AI pricing is important, which indicates that economic clarity is becoming as critical as technical capability when committing to multi-year orchestration programs.
Industry reports outside these studies underline a similar pattern: capital is available for AI, but boards increasingly demand proof that initiatives reduce working capital volatility, shrink disruption response times, or support regulatory compliance, not just create dashboards. That expectation pushes network leaders to move beyond visibility projects and design AI as an operating layer that governs flows, not just monitors them.
Why AI Spending Still Fails To Change Network Behavior
Many supply chains already possess enough AI tools to automate narrow decisions, yet their planning and execution models still depend on fragmented ownership, delayed data reconciliation, and manual escalation paths built for slower networks. That mismatch helps explain why orchestration programs often stall after pilot success. As tariffs, compliance pressures, and inventory volatility continue to compress response windows, companies may find that the next advantage comes less from adding another AI layer and more from reducing the organizational friction that prevents systems from acting on the information they already have.
