A new Gartner survey shows AI-native supply chain ambitions running into two old barriers – legacy technology and scarce skills. As pressure rises to show returns from digital investment, many organizations are discovering that outdated systems and unchanged roles are limiting AI to small pilots and local gains.
Legacy Foundations are Capping AI Payoff
Gartner’s latest research points to a structural constraint rather than a tooling problem. In a sample of 140 senior supply chain executives at companies with at least $250 million in annual revenue, 56% cited integration with legacy systems and processes as a major obstacle to AI deployment. Half said they lack internal expertise to implement and manage AI at scale. The result is a wave of projects that sit on the edge of the operation instead of reshaping how it runs.
These findings echo a pattern seen across recent digital programs. Enterprises have spent heavily on control towers, planning platforms, and analytics, yet many still rely on spreadsheets and manual workarounds for core decisions. Reference data from recent consulting interviews shows advanced systems often used as data pipes rather than decision engines, with teams reverting to familiar workflows when complexity rises. AI is now landing on this same brittle foundation: fragmented data, overlapping systems of record, and processes that were never designed for algorithmic decision-making.
Gartner defines an ‘AI-native’ supply chain as one architected around AI from the start, where data, governance, and workflows are built to support machine-driven recommendations. That design is rare. Most current deployments still resemble bolt-on enhancements to analog-era designs. Signals surface in dashboards, models flag risks, but ownership for acting on those insights is fuzzy and cross-functional alignment is slow. Industry interviews highlight a recurring execution gap: visibility improves, yet service, cost, and working capital metrics move only marginally because decisions remain episodic and person-dependent.
Legacy integration is part technical and part structural. Many ERP, warehouse, and planning systems run on rigid data models and batch processes. AI tools that depend on clean, timely, and consistent inputs struggle in this environment. Data quality issues and process variance force teams to spend time reconciling feeds rather than acting on recommendations. One consulting leader described continuous improvement staff spending more time as ‘continuous analytics’ teams with stopwatches than as operators of self-improving systems. Until underlying flows are simplified and standardized, AI remains a sophisticated overlay on top of operational noise.
Operating Models, Not Algorithms, Define AI Leaders
The same Gartner report highlights a smaller cohort already converting AI into tangible value. These ‘AI leaders’ share three traits: they are redesigning how decisions get made end to end, reshaping team structures around AI-centric work, and upgrading technology in modular layers instead of attempting big-bang replacements. Their advantage comes less from novel algorithms and more from operating discipline.
First, they treat AI deployment as an operating model redesign. Decision rights, escalation paths, and service policies are revisited so AI-generated signals map to clear actions. Industry analysis shows that when every disruption or demand shift routes through a named owner with predetermined playbooks, the time from detection to execution drops sharply. These organizations are also more willing to retire redundant reports and legacy KPIs that anchor teams to backward-looking performance rather than forward risk and opportunity signals.
Second, they are altering roles to reflect AI-era work. Gartner notes that companies further along in their journey are moving away from narrow functional positions and introducing new roles focused on orchestration, AI interpretation, and scenario design. This shift lines up with broader market trends. Reference reports describe emerging positions centered on managing cross-functional command centers, curating decision frameworks, and governing AI output quality. In these environments, planners spend less energy building spreadsheets and more time deciding which automated recommendations to accept, override, or escalate.
Third, AI leaders are tackling the technology stack as a layered renewal instead of locking into a single transformation event. Gartner highlights an approach that builds new AI-capable layers over time, decoupling data, analytics, and workflow orchestration from the deepest legacy cores. That allows incremental de-risking: new capabilities can be proven in one lane or region while the rest of the network continues to run on existing platforms. Reference commentary on process mining and simulation highlights how such tools help expose bottlenecks and validate changes before they scale, improving the odds that AI initiatives produce measurable operational results rather than isolated proofs of concept.
Execution Discipline Becomes The Real AI Advantage
The most under-reported consequence of this research is what it implies about future competitive separation. Industry data suggests AI currently delivers modest productivity gains in the single digits when layered onto unchanged processes, yet the gap between organizations that align operating models to AI and those that do not is compounding year by year. As more disruption risk, ESG pressure, and working capital scrutiny converge on supply networks, the differentiator will not be who has AI, but who has rebuilt governance, roles, and system layers so that AI can act as the default engine of execution rather than an add-on experiment.
