Most MES programs are stuck at pilot scale, and that’s now an AI and cyber risk decision

Manufacturing execution systems are everywhere. Standardized, enterprise-grade execution systems are not.

A Rockwell Automation survey of 1,560 manufacturing and industrial professionals across 17 countries found that 93% of respondents have MES running in at least one facility, yet only 23% say the technology is fully integrated across all of their sites, according to Manufacturing Dive’s reporting of Rockwell’s July 2026 report. In other words, most organizations have proven MES value locally, then stalled when it comes time to scale.

For plant and IT leaders, the headline is not “MES adoption is high.” It’s that the hard part, the operating model required to make MES a multi-site system of record, is still the bottleneck, right as AI use cases and cyber risk are pushing manufacturers to tighten control of shop-floor data and workflows.

The number to benchmark against is 23%, not 93%

The 93% figure is useful as a reminder that MES is no longer niche. But it can be a misleading yardstick for peers because it includes organizations running MES in a single plant, a single facility, or even a limited set of lines.

The 23% metric, “fully integrated across all sites,” is the more operationally relevant benchmark for manufacturers with multiple plants, co-manufacturers, or regional networks. It is a proxy for whether a company has built repeatable rollout mechanics: standard process models, common integrations, identity and access patterns, and a governance structure that survives plant-to-plant variability.

Manufacturing Dive notes that integration of separate, patchwork systems has become one of the biggest challenges as AI adoption rises and cyber risk grows, a framing that shifts MES from a productivity tool into a control plane problem.

Patchwork execution systems become an AI cost center fast

Manufacturers are under pressure to operationalize AI in ways that go beyond isolated pilots. But AI depends on consistent definitions and repeatable context. When MES is implemented differently by plant, the organization inherits multiple ways to define a “good” unit, a “downtime” event, a “rework” path, or a “recipe change.”

Rockwell’s numbers, as reported by Manufacturing Dive, suggest that many manufacturers are in exactly that state: MES exists, yet the enterprise still lacks a unified execution layer. That situation does not block AI outright, but it changes the economics. Data engineering and integration work expand, model governance gets harder, and any cross-site KPI program becomes an exercise in reconciliation rather than continuous improvement.

For organizations running high-mix production or operating across acquired facilities, this matters more. Those environments typically have the widest variance in equipment, operators, and process flows, which makes “common MES” both more valuable and more difficult to implement.

Scaling MES is becoming a cyber and governance requirement

The same fragmentation that makes analytics difficult also complicates cyber defense. Manufacturing Dive ties the scaling challenge to “growing cyber risk,” and the linkage is practical: multiple MES instances often mean different patch levels, different integrations to historians and ERP, and inconsistent user provisioning and remote-access controls across sites.

A standardized MES footprint, or at least a standardized security model around MES, makes it easier to enforce segmentation, logging, and access controls. Fragmentation does not guarantee risk, but it increases the number of exceptions security teams must understand and manage, especially when plants rely on third-party integrators and local admin practices.

Rockwell’s survey design also matters operationally. Because it spans 17 countries and includes both manufacturing and industrial professionals, according to Manufacturing Dive, the data is broad enough to be directional across verticals, not a single-industry snapshot. That breadth reinforces that the barrier is rarely “Does MES work here?” and more often “Can we scale change management and integrations here?”

Where this lands in 2027 capital planning and vendor selection

The immediate implication for procurement and operations is to stop treating MES as a software line item and start treating it as an enterprise rollout capability. The constraint highlighted by Rockwell’s 23% figure is less about licensing and more about deployment repeatability.

Manufacturing organizations that are budgeting for AI initiatives should stress-test their MES roadmap against a simple requirement: can a model trained on Plant A data be validated, monitored, and audited in Plant B without rebuilding the entire pipeline? If the answer is no, the organization is likely paying the “patchwork tax” that Rockwell and Manufacturing Dive are flagging.

That also shifts how MES vendors, systems integrators, and internal IT teams should be evaluated. A strong demo in one facility is table stakes. The differentiator is the vendor’s ability to support templated deployments, upgrade cadence, and governance across a fleet of plants, including integration patterns that survive equipment differences.

Questions to take into MES rollout scoping, integration SOWs, and security reviews

• When an MES instance goes live at a new site, what is the “golden template” (data model, workflows, role definitions), and who owns changes to it after local exceptions are discovered?
• What is the vendor or integrator’s priced rollout unit, for example, per site or per line, including integration to ERP/quality/maintenance systems, and what must be true for that price to hold?
• How will identity, access, logging, and patching be standardized across all MES-connected assets, and which team (IT security, OT engineering, or an integrator) is accountable for ongoing control verification?
• Which AI use case is being used as the scaling test, and what cross-site KPI definition will be enforced through MES rather than handled later in BI?