Cisco’s AI Strategy: Controlling the Economics, Complexity, and Trust of Agentic AI
Jeetu Patel’s keynote at Cisco’s Partner Summit delivered a message that goes well beyond another AI infrastructure investment cycle.
My biggest takeaway: Cisco is positioning itself not simply as a networking and security provider, but as the foundational infrastructure and operational platform for the agentic AI economy.
I believe this is an important distinction.
For the past several years, much of the industry has focused on GPUs, compute capacity, and building larger AI data centers. Those investments remain critical. But as AI transitions from chatbots responding to human prompts to autonomous agents continuously executing tasks, the economics and operational requirements of infrastructure fundamentally change.
Four market forces
Jeetu framed the transformation around four forces: agents as the new workforce, tokens as the new currency, control as the new competitive moat, and trust as the catalyst for adoption.
From an ACG Research perspective, these forces are interconnected. Together they create several important implications for enterprises, service providers, and Cisco’s partner ecosystem.
Five strategic observations
1. Agentic AI changes the economics of infrastructure consumption
Traditional AI interactions are largely transactional. A user asks a question, infrastructure resources are consumed, and the interaction ends.
Agentic AI changes that consumption model. Agents operate continuously, interact with other agents, access multiple systems, and execute complex workflows. This creates persistent infrastructure demand, increasing network utilization, compute consumption, and operational complexity.
Jeetu cited Cisco research suggesting that agents can consume 450% more network bandwidth than humans performing comparable tasks.
The implication is significant: enterprises must begin evaluating infrastructure not simply on capacity, but on cost per completed task, network efficiency, token consumption, and business outcomes. This is where I see the next major opportunity for infrastructure economics and optimization.
2. Token economics will become as important as compute economics
One of the more insightful parts of the presentation was the emphasis on token consumption. As organizations deploy thousands of agents, AI costs become increasingly difficult to forecast and control. Using expensive frontier models for every task is neither economically sustainable nor operationally necessary.
Cisco’s introduction of Model Foundry, with general availability targeted for June 2027, is strategically interesting because it enables partners to customize domain-specific models using customer data. The demonstration with WWT showed a smaller, networking-specific model outperforming a general-purpose frontier model on selected networking tasks.
The future of AI economics will not be determined solely by who has the most powerful model. It will be determined by who can deliver the required intelligence at the lowest sustainable cost. The ability to intelligently route workloads across frontier models, smaller specialized models, private infrastructure, and edge environments will become a competitive differentiator.
However, model efficiency must ultimately be validated against production workloads: accuracy, inference latency, total operating costs, and ongoing model maintenance.
3. Cisco Cloud Control could become a strategic differentiator, but execution matters
Cisco has historically faced a challenge common to large technology companies: an extensive portfolio of powerful products accompanied by operational complexity across management platforms, licensing, and integration. Jeetu made it clear that simplification is now a strategic priority.
Cisco Cloud Control is an effort to unify management, observability, security, support, and agent-driven operations into a more consistent operational experience. From my perspective, this could be one of Cisco’s most consequential initiatives.
Why? Because customers do not need more management tools. They need fewer operational silos, faster troubleshooting, improved automation, and lower operating expenses. If Cisco delivers integrated, cross-domain operations while maintaining openness to third-party environments, the value proposition moves beyond product consolidation. It becomes an opportunity to materially improve infrastructure economics.
The critical measurement will be whether Cisco can demonstrate quantifiable reductions in operational costs, mean time to resolution, and complexity, not simply a more unified interface.
4. Trust becomes the foundation of autonomous operations
Perhaps the most important strategic point was the discussion of trust. The more autonomy an AI agent receives, the more valuable it can become. But granting access to enterprise applications, data, and infrastructure also expands security and operational risk.
Traditional security architectures were designed around human users, devices, and applications. Agentic AI introduces another dimension: autonomous entities making decisions and taking actions at machine speed. This requires organizations to move beyond identity and access management toward identity, action control, behavioral monitoring, and real-time policy enforcement.
Cisco’s combination of AI Defense, Splunk, network telemetry, and observability creates an opportunity to correlate agent behavior with infrastructure and security events. The strategic value is not simply detecting malicious activity. It is distinguishing between an external attack, an agent operating outside its intended boundaries, and an agent making a poor decision. That distinction will become increasingly important as enterprises move toward autonomous operations.
My view is that trust will become an economic enabler of AI adoption, not simply a security requirement. Organizations that cannot govern autonomous agents will struggle to scale them safely.
5. Cisco’s vertically integrated strategy lets partners move up the value chain
Cisco’s investments in Silicon One, optical technologies, AI-ready data centers, networking, security, and observability reflect a strategy to deliver a tightly integrated infrastructure platform. The advantage is the potential to optimize performance, power consumption, operational efficiency, and security across multiple technology layers.
But the real opportunity for Cisco’s partners is not simply selling more infrastructure. It is helping customers understand how these architectural changes affect their businesses.
Partners can evolve from technology resellers and systems integrators into strategic advisors delivering AI infrastructure assessments, economic modeling, security modernization, agentic operations, and measurable business outcomes. This requires a different conversation with customers, one focused on ROI, TCO, operational efficiency, risk reduction, and time to value.
The bigger picture: three tests for Cisco
I see three areas that will ultimately determine whether Cisco can translate this strategy into sustained competitive advantage.
Economic differentiation. Can Cisco demonstrate lower cost per AI transaction, reduced infrastructure TCO, improved power efficiency, and measurable operational savings?
Operational simplification. Can Cisco Cloud Control unify the operational experience across Cisco and multivendor environments without introducing another layer of complexity?
Trusted autonomy. Can Cisco integrate security, observability, and agentic operations so enterprises can delegate more decisions safely and with appropriate governance?
Final perspective
Jeetu presented a compelling vision for Cisco’s role in the AI economy, particularly around the convergence of networking, compute, security, observability, and autonomous operations.
I agree with his broader premise that the industry may be underestimating the infrastructure implications of agentic AI. However, significant demand does not automatically translate into profitable growth. Power availability, capital intensity, model efficiency, integration complexity, and enterprise adoption will determine how this opportunity develops.
What stood out to me was Cisco’s recognition that the next phase of AI will be defined as much by operational economics and trust as by raw infrastructure performance. The companies that win will not necessarily be those deploying the most GPUs, building the largest networks, or running the most sophisticated models.
For Cisco, the opportunity is substantial. But the next phase must be about proving those outcomes through measurable customer results, not announcing products too early.
AI is no longer just a compute challenge. It is becoming a network, security, operational, and business economics challenge.
And that is where I believe the next major wave of industry differentiation will occur.
Ray Mota
Principal Analyst, ACG Research
rmota@acgcc.com
If you found this useful, subscribe for more analysis on the economics of networking, AI infrastructure, and autonomous operations.
Originally published on Ray Mota's Substack. Subscribe to get new analysis by email.



