Cisco Industrial AI Report 2026: Cybersecurity Is the Top Barrier to AI Adoption

Key Takeaways

1. Cybersecurity Now Leads AI Barriers. 40% of industrial decision-makers cite cybersecurity as the top obstacle to AI adoption, rising from third place in just two years.
2. Network and Security Challenges Converge. 48% of organizations name security and network segmentation as their biggest networking issue amid expanding AI-driven attack surfaces.
3. IT/OT Silos Undermine Readiness. 43% of firms still operate with limited IT/OT collaboration, correlating with 90% wireless instability rates and major visibility gaps.
4. AI Security Hopes Lack Foundations. 85% expect AI to boost cybersecurity, yet only 43% have centralized AI data governance to support it.

Two years ago, industrial organizations treated cybersecurity as one concern among many when evaluating AI readiness. Skills gaps and integration complexity ranked higher. Budget constraints were on the same level. Cybersecurity was a consideration, not a blocker.

That’s over. The Cisco 2026 State of Industrial AI Report documents the shift in stark terms. Among more than 1,000 decision-makers across manufacturing, utilities, and transportation in 19 countries, cybersecurity now ranks as the number-one obstacle to AI adoption. Forty percent cite it as the top barrier. Forty-eight percent name it their biggest networking challenge. The rise from third-place concern to first-place blocker happened in less than two years.

The cause is structural, not perceptual. Industrial AI requires connecting more assets, more systems, and more data flows across environments that were never designed for broad connectivity. Every new sensor feeding an AI vision system, every edge computing node processing quality inspection data, every machine-to-machine decision loop—each one expands the attack surface. Traditional security architectures built around perimeter defense and north-south traffic monitoring were not engineered for this reality. Organizations deploying AI fastest are often creating risk fastest, because their security infrastructure hasn’t kept pace with their AI ambitions.

The report is clear about what’s at stake: 61% of organizations are already deploying AI at scale across multiple sites. Only 14% are still in the exploring or piloting phase. This is not a future problem. It is a current exposure—one that grows with every deployment that lacks the security architecture to support it.

5 Key Takeaways

1. Cybersecurity has overtaken every other barrier to industrial AI adoption—and it’s not close.

The Cisco 2026 State of Industrial AI Report, based on 1,000+ decision-makers across 19 countries, finds 40% now cite cybersecurity as the single largest obstacle to AI adoption in manufacturing, utilities, and transportation. In 2024, cybersecurity ranked third. The rapid escalation reflects what happens when organizations connect more assets to support AI without rethinking how those expanded attack surfaces are protected.

2. Network infrastructure and security are now the same problem—and most industrial organizations aren’t solving either.

Forty-eight percent of respondents identify security and network segmentation as their greatest networking challenge. AI workloads demand connectivity and edge computing capacity that legacy industrial networks were never designed to deliver. When security depends on network architecture that can’t keep pace with AI deployment, every new connected asset becomes both a productivity investment and a risk exposure.

3. IT/OT silos are measurably degrading AI readiness—and 43% of organizations haven’t fixed them.

Nearly half of industrial organizations operate with limited or no collaboration between IT and OT teams—unchanged since 2024. The consequences are measurable: 90% of siloed organizations report wireless instability versus 61% among collaborative ones. The Kiteworks 2026 Forecast Report found manufacturing organizations report the highest data loss prevention visibility gaps of any industry at 67%, with 52% citing third-party AI data handling as a top concern.

4. Organizations are betting on AI to solve the cybersecurity problem AI created—but the foundation isn’t in place.

Eighty-five percent expect AI to improve their cybersecurity posture, and industrial cybersecurity ranks as the second-highest priority for AI investment. But only 43% of organizations globally have a centralized AI Data Gateway. In manufacturing specifically, blind spots are pervasive. You cannot secure what you cannot see.

5. The gap between AI deployment confidence and actual operational transformation is the defining risk of the next three years.

Ninety-three percent of organizations say they’re confident in scaling AI. Only one-third expect enterprise-wide transformation. That gap is where security failures, compliance violations, and operational disruptions live. The organizations furthest along share modernized networks, mature cybersecurity practices, and collaborative IT/OT governance—conditions not yet widespread.

The Network Problem and the Security Problem Are the Same Problem

One of the most consequential findings in the Cisco report is how completely network readiness and cybersecurity have converged in industrial environments. Most decision-makers say reliable wireless networks are vital for enabling AI. Half expect significant increases in connectivity and reliability requirements as deployments scale. And 48% report that security and network segmentation represent their greatest networking challenge.

This convergence creates a compounding problem. AI workloads are generating demands for bandwidth, edge computing, and real-time data processing that legacy industrial networks cannot meet. At the same time, those networks lack the segmentation, visibility, and zero trust architecture controls required to prevent a security event on the office network from reaching the plant floor. When the infrastructure can’t deliver reliable connectivity and can’t enforce security boundaries, organizations face a choice between slowing AI deployment and accepting risk they cannot quantify.

AI already accounts for 13% of networking budgets, and 83% of organizations plan to increase that allocation. Edge computing capacity, AI vision systems, and industrial connectivity rank among the highest-priority technology investments. But budget allocation without architectural clarity produces spending without security. Organizations that treat network modernization and cybersecurity as separate line items are building infrastructure that will require expensive remediation later—if a breach doesn’t force the issue first.

IT/OT Silos: The Organizational Failure That Makes Every Technical Problem Worse

The Cisco report confirms what industrial cybersecurity practitioners have been saying for years: the IT/OT collaboration gap is not closing. Forty-three percent of organizations operate with limited or no cooperation between IT and OT teams—a figure that has not meaningfully improved since 2024 despite widespread recognition that it’s a problem.

The practical consequences are measurable and severe. Organizations with siloed teams report wireless instability at 90%, compared to 61% among those with collaborative structures. Confidence in scaling AI tracks directly with organizational alignment. And here’s the finding that should give every CISO pause: organizations with closer IT/OT collaboration are actually more likely to cite cybersecurity as a primary obstacle—by 12 percentage points. The report attributes this to visibility. Closer collaboration surfaces risks that siloed teams never detect.

This visibility gap extends directly to data security. The Kiteworks 2026 Forecast Report found that 67% of manufacturing organizations cite visibility gaps as a top data security concern—21 percentage points above the global average. Manufacturing’s complex, multi-tier supply chains generate sensitive data flows that move through email, secure file sharing, SFTP, MFT, APIs, and AI integrations. When IT and OT teams operate independently, no one has unified visibility into how sensitive data moves across these channels. Supply chain risk management concerns around third-party AI data handling sit at 52% in manufacturing—the highest of any industry. The silos that prevent network collaboration are the same silos that prevent data governance.

AI as the Security Answer—But Only With the Right Foundation

Despite cybersecurity’s role as the primary constraint on industrial AI, organizations are simultaneously betting heavily on AI to strengthen their defenses. Eighty-five percent of respondents expect AI to improve their cybersecurity posture. Industrial cybersecurity ranks as the second most important area for AI investment. The expectation is that AI will deliver detection, monitoring, and response at a scale and speed that manual approaches cannot match.

The logic is sound. The execution prerequisites are where most organizations fall short. Cisco’s Samuel Pasquier identifies the path forward in three steps. First: visibility. You cannot protect data or feed it to an AI system if you don’t know what’s on your network—and visibility must extend all the way to the network edge to capture east-west traffic between devices. Second: network segmentation to isolate AI workloads so that a security event on the office network cannot reach the plant floor. Third: unified IT/OT governance that treats OT cybersecurity as a shared baseline, not a separate domain.

These three requirements—visibility, segmentation, and unified governance—map directly to the data security challenges the Kiteworks 2026 Forecast Report documents across industries. Only 43% of organizations have a centralized AI Data Gateway. The remaining 57% operate with distributed, partial, ad hoc, or nonexistent controls over how AI systems access sensitive data. In manufacturing, the problem is particularly acute: supply chain data, proprietary process information, quality inspection records, and operational telemetry all flow through AI systems that most organizations cannot monitor, audit, or control from a single governance plane.

The Confidence-Transformation Gap: Where Industrial AI Stalls

Perhaps the most revealing finding in the Cisco report is the disconnect between confidence and outcomes. Ninety-three percent of organizations express confidence in their ability to scale AI. Yet only one-third expect enterprise-wide operational transformation over the next three to five years. Most continue using AI to improve existing processes rather than redesign how operations work.

That gap is not a marketing problem. It is an infrastructure, security, and organizational structure problem. The organizations furthest along in AI deployment share a common profile: modernized networks, mature cybersecurity practices, and collaborative IT/OT governance. Those conditions are not yet widespread. And without them, AI at industrial scale remains the exception.

For manufacturers still operating on legacy infrastructure, Pasquier says the path forward does not require replacing existing systems. It requires layering visibility, segmentation, and governance onto what already exists. That’s a pragmatic message—but it carries a crucial caveat. The tools you use to implement visibility and governance must themselves be secure, auditable, and capable of operating across the fragmented channel landscape that industrial data actually moves through.

How Kiteworks Helps Industrial Organizations Secure the Data That AI Depends On

The Cisco report identifies three foundational requirements for industrial AI security: visibility into data flows, segmentation to isolate AI workloads, and unified governance across IT and OT. These are the same requirements that govern how sensitive data should be exchanged, shared, and controlled across the channels industrial organizations actually use.

The Kiteworks Private Data Network addresses these requirements at the data exchange layer—where sensitive information moves between systems, partners, AI models, and operational environments. For industrial organizations navigating the security challenges the Cisco report documents, Kiteworks delivers capabilities that fragmented tools cannot:

• Unified data governance across all exchange channels: One policy engine applies consistent role-based and attribute-based access controls across secure email, secure file sharing, SFTP, secure MFT, APIs, web forms, and AI integrations. When manufacturing data flows through multiple channels to reach AI systems, governance follows the data—not the channel.
• Immutable audit trails with real-time SIEM delivery: Every data exchange event is captured in a single, consolidated audit log—with zero throttling and zero dropped entries. For the 63% of organizations that have not integrated file transfer with SIEM/SOC platforms, this eliminates the blind spot where attackers exploit unmonitored channels.
• Security-by-design architecture: Kiteworks deploys as a hardened virtual appliance with embedded firewalls, WAF, intrusion detection, double encryption at rest, and zero trust data protection—maintained by Kiteworks, not your infrastructure team.
• Network segmentation support: Single-tenant deployment means no shared databases, file systems, or runtimes. AI data exchanges are isolated by design, ensuring a compromise in one environment cannot propagate to another—precisely the ring-fencing the Cisco report recommends.
• Third-party and supply chain data governance: Full external user lifecycle management with access control enforcement and complete audit trails for every vendor data exchange. For the 52% of manufacturers concerned about third-party AI data handling, Kiteworks provides the visibility and control that fragmented tools leave as blind spots.
• AI-ready integration: The Kiteworks Secure MCP Server enables AI systems to interact with sensitive operational data while respecting existing governance policies—extending zero trust controls to AI workflows without requiring separate infrastructure for every new use case.

The result: industrial organizations can pursue the AI deployment velocity the Cisco report documents while maintaining the security, visibility, and governance that prevent every deployment from becoming a new exposure.

What the Cisco Report Means for Your Industrial AI Security Strategy

The Cisco 2026 State of Industrial AI Report does not describe a future problem. It documents a current one. Organizations deploying AI at scale—and 61% already are—are simultaneously expanding attack surfaces, increasing data flow complexity, and creating governance gaps that traditional security tools cannot close.

Five priorities emerge from the report’s findings:

First, treat network security and AI security as a unified problem. The finding that 48% cite security and segmentation as their greatest networking challenge confirms these are not separate budget lines—they are the same infrastructure requirement.

Second, close the IT/OT collaboration gap before scaling AI further. The 43% still operating in silos are not just missing operational efficiency—they are missing the security risks that only collaborative visibility reveals.

Third, implement visibility into east-west data traffic at the network edge. North-south monitoring is necessary but insufficient. AI workloads generate lateral data flows between devices that perimeter-focused security architectures do not capture.

Fourth, establish a centralized AI Data Gateway. The finding that 57% of organizations lack centralized governance over AI data access is a direct liability for manufacturers running AI across quality inspection, process automation, and supply chain risk management. Distributed controls do not scale.

Fifth, demand unified governance over every channel through which sensitive data reaches AI systems. Secure email, secure file sharing, SFTP, MFT, APIs, and AI integrations each represent a potential governance gap. Organizations managing these through separate tools are creating the exact fragmentation the Cisco report identifies as a scaling barrier.

The industrial organizations that close these gaps in 2026 will scale AI with confidence. The ones that defer will discover that their AI deployments have been building technical debt measured not in inefficiency but in unquantified risk—risk that compounds with every new connected asset, every new data flow, and every new AI use case deployed without the governance to support it.

To learn more about deploying AI at scale securely, schedule a custom demo today.

Additional Resources

• Blog Post
  Zero‑Trust Strategies for Affordable AI Privacy Protection
• Blog Post
  How 77% of Organizations Are Failing at AI Data Security
• eBook
  AI Governance Gap: Why 91% of Small Companies Are Playing Russian Roulette with Data Security in 2025
• Blog Post
  There’s No “–dangerously-skip-permissions” for Your Data
• Blog Post
  Regulators Are Done Asking Whether You Have an AI Policy. They Want Proof It Works.