Cybersecurity is now the biggest obstacle to scaling industrial AI, according to Cisco’s survey of more than 1,000 decision-makers at industrial companies with more than $100 million in annual revenue.
In the survey, 40% named cybersecurity a top barrier as 61% said AI is already in broad or mature deployment.
Infrastructure that wasn’t built for this
Those deployments are landing on infrastructure many organizations still need to upgrade. Cisco found 97% expect AI workloads to change industrial network requirements, with 51% citing higher connectivity and reliability needs, 44% more edge compute capacity and 42% more bandwidth.
Reliable wireless has also moved into the critical path, with 96% saying it is vital to enabling industrial AI, while 48% named security and segmentation their biggest networking challenge.
Why companies are pushing ahead anyway
The report also shows why companies are still pressing ahead. Productivity was the top driver of industrial AI interest at 63%, followed by cost reduction at 42% and improved security at 36%, while 87% said they expect AI outcomes within two years.
OT’s collision with AI timelines
The near-term expectations are colliding with the realities of operational technology, where systems interact directly with the physical environment and carry distinct performance, reliability and safety requirements, as NIST’s Operational Technology (OT) security guidance notes.
That collision is also showing up in U.S. government guidance. In December, NSA, CISA and partner agencies said integrating AI into OT systems introduces new risks to safety and security.
They outlined four principles for critical infrastructure operators: understand AI risks, assess whether AI is appropriate for the OT use case, establish governance and assurance frameworks, and embed safety and security practices into AI-enabled OT systems
The guidance also calls for human oversight and fail-safe mechanisms in critical decisions.
IT/OT gaps and the visibility paradox
Cisco’s survey suggests governance is lagging behind deployment. Forty-three percent of respondents said IT and OT still operate with limited or no cooperation. Among those siloed organizations, 90% reported wireless instability, versus 61% where teams collaborate, and confidence in scaling AI was lower.
Cisco also found that reported cybersecurity concern rises as IT/OT collaboration increases, which the report says may reflect greater visibility into risk as teams work more closely together.
Executive ownership appears to be moving in the same direction. Fortinet said in July 2025 that 52% of organizations now place OT security under the CISO or CSO, up from 16% in 2022. That shift does not by itself prove maturity, but it does show OT risk moving higher in enterprise governance as AI and connected operations expand.
Confident, but not transforming
Cisco’s sharpest finding may be the gap between confidence and transformation. The report said 93% of organizations are confident they can scale AI, yet only one-third expect enterprise-wide or end-to-end operational transformation in the next three to five years.
That disconnect leaves cybersecurity, network readiness and IT/OT alignment looking less like supporting issues and more like the conditions that determine whether industrial AI remains a set of productivity projects or becomes a broader operating model.
In Deloitte’s broader 2026 enterprise AI survey, 34% of organizations said they were using AI to deeply transform the business, while 30% were redesigning key processes and 37% were still using it at a more surface level (figures may not total 100% due to rounding).