AgenticOps and the race to control enterprise AI

Are you ready to be a part of the premier event trusted by enterprise leaders for almost twenty years? VB Transform is the place where the minds behind real enterprise AI strategy come together. Find out more

By the year 2032, it is projected that spending on Enterprise AI infrastructure will soar to a staggering $309 billion. The key players in this arena will not be determined by who has the best models, but rather by who controls the infrastructure layer that allows AI to function at scale.

Security vendors are taking bold steps in this direction. Companies like Palo Alto Networks, CrowdStrike, and Cisco are witnessing a significant surge in AI-driven security revenue, with growth rates of 70 to 80% year-over-year, while traditional infrastructure sales are on the decline. The trend is clear: Security is evolving into the control center for enterprise AI.

Ali Ghodsi, CEO of Databricks, highlighted in a blog post the challenges posed by the complexity of AI workloads on existing infrastructure. He emphasized the need for innovative approaches to manage AI operations at scale.

Evidence suggests that 73% of enterprises identify infrastructure inadequacy as the primary obstacle to adopting AI. Moreover, adversaries are harnessing AI faster than enterprises can fortify their defenses. The battleground for infrastructure supremacy has been set.

AgenticOps Emerges as the New Battleground

AgenticOps is not merely a single vendor’s vision; it signifies the industry’s acknowledgment that traditional IT operations are ill-equipped to manage AI agents operating at machine speed with human permissions. While Cisco introduced this concept at Cisco Live 2025, other major players like Microsoft’s AI Orchestration, Google’s Model Operations, and startups such as Weights & Biases are racing to dominate this space. The battle lines have been drawn.

Enterprises deploying a significant number of AI agents require infrastructure that can handle cross-domain data access, real-time governance, and multi-team collaboration. Traditional tools falter when the agent count reaches 5,000. The need for a new approach is evident.

According to Jeetu Patel, Cisco’s president and CPO, security is now a catalyst for adoption, rather than a hindrance. Security teams are transitioning to facilitators of AI deployment, rather than impediments.

Forrester research emphasizes the importance of multi-domain visibility, with three pillars defining enterprise-grade AgenticOps: unified data access across all domains, collaborative environments for NetOps and SecOps teams, and purpose-built models governing agent actions. Vendors that excel in all three components are poised to lead the pack.

The Death of Perimeter Security

Conventional firewalls are inadequate for safeguarding AI workloads. Leading security solutions like Palo Alto’s Prisma Cloud, Fortinet’s Security Fabric, and Check Point’s Infinity operate on the premise that perimeter defense is no longer effective.

The introduction of Extended Berkeley Packet Filter (eBPF) has revolutionized security enforcement by eliminating the performance overhead associated with traditional methods. Cisco’s acquisition of Isovalent for $2.8 billion validated this approach, with the open-source project Cilium now securing critical workloads for major enterprises. The shift towards security at the hardware level ensures minimal latency and a significant leap in performance.

Cisco’s VP and CTO of security, Craig Connors, emphasized that security policies now extend across all layers, from workloads to silicon. This integration underscores the evolution of security from an add-on feature to an integral component of infrastructure.

Hardware-based security offers a substantial advantage, with silicon-embedded security operating at nanosecond latencies, far outperforming software-defined security in terms of speed and efficiency.

The 72-Hour Exploit Window

In the realm of cybersecurity, adversaries exploit vulnerabilities within 72 hours, whereas enterprises typically take 45 days to patch these vulnerabilities. This delay accounts for the majority of successful breaches. Therefore, security vendors are in a race to narrow this crucial window.

Leading solutions like CrowdStrike’s Falcon Prevent, Qualys VMDR, Tanium Patch, and Cisco’s Live Protect are designed to block exploits before patches are available and deliver real-time vulnerability management with swift automated responses.

Research from the Ponemon Institute reveals that delayed patching incurs significant breach risks, with each hour of delay costing an average of $84,000. Automated platforms offer a rapid return on investment, with CISOs recognizing the imperative nature of swift cybersecurity responses.

Shlomo Kramer, CEO of Cato Networks, stressed the critical role of automation in cybersecurity, particularly in responding to attacks that exceed human response capabilities.

The Observability Wars Intensify

The recent $28 billion acquisition of Splunk underscores the pivotal role observability plays in determining success in the AI infrastructure domain. Companies like DataDog, New Relic, and Dynatrace are processing vast amounts of data to provide essential visibility into AI operations.

Enterprises venturing into AI without robust observability mechanisms are operating blindly, as Etay Maor, senior director of security strategy at Cato Networks, aptly puts it, “You can’t secure what you can’t see.” The evolution towards generative UI promises dynamic interfaces that adapt in real-time to address specific challenges, with industry players like ServiceNow, Splunk, and Observable at the forefront of this innovation.

Market Consolidation Accelerates

The landscape of AI infrastructure is undergoing rapid consolidation through acquisitions by major players. Companies like Cisco, Palo Alto, CrowdStrike, and Broadcom are fortifying their positions through strategic acquisitions, reshaping the competitive landscape.

The velocity of platform development now serves as a crucial determinant of survival in the market, with unified architectures reducing deployment timelines significantly. Gartner’s prediction that the AI infrastructure market will streamline from over 200 vendors to fewer than 20 platforms within three years underscores the fierce competition and the imperative for full-stack control.

The Bottom Line

AgenticOps signifies a monumental shift in architecture, with enterprises adapting to continuous compromise, infinite identities, and rapid cyber threats. Those embracing these changes will thrive, while those clinging to outdated defense mechanisms risk obsolescence.

The vendors at the forefront of this evolution, whether established giants like Cisco and Microsoft or emerging disruptors, will shape the next decade of enterprise technology. The race for dominance is underway, and the winners are already emerging.