How Cisco’s platform mindset is meeting the AI era

Cisco is no longer trying to be just a software, hardware, or silicon company. It’s now positioning itself as a systems or platform company that delivers the critical infrastructure needed for the AI era.

This week, Cisco held its annual Cisco Live EMEA event and announced its fiscal second-quarter earnings. During the Amsterdam keynote, Jeetu Patel, Cisco’s president and chief product officer, was blunt: “AI isn’t coming. AI is here, and it’s moving at breakneck speed.” This sentiment was reiterated on the earnings call, during which CEO Chuck Robbins reported record revenue and a massive acceleration in AI infrastructure orders.

** Related:[More Cisco news and insights]**

Here are six trends that stood out to me after Cisco Live EMEA and the company’s Q2 performance. (See also my earlier column on Cisco’s promising 2026)

1. AI‑ready infrastructure becomes the main act

The keynote at Cisco Live EMEA was meant to send a message of urgency to the audience: AI is no longer a sidecar workload; it’s the design center for networks, data centers and security architectures.

Patel framed Cisco as a key supplier of technology, emphasizing that “tokens per dollar per watt is the new currency” for data centers, not just speeds and feeds.​ This is something I’ve discussed at length with Nvidia, as well. AI TCO must be calculated at a token level, not the cost per switch port or GPU.

On the earnings call, Robbins backed that message with numbers: AI infrastructure orders from hyperscalers hit $2.1 billion in Q2 alone—which is equal to Cisco’s AI infrastructure orders from hyperscalers for all of fiscal 2025. For fiscal year 2026, Cisco now expects AI orders in excess of $5 billion and more than $3 billion in AI infrastructure revenue from hyperscalers. Robbins stressed that this $5 billion outlook does not include the newly announced G300 102.4 Tbps switching silicon nor the latest P200-class scale‑across products, meaning upside remains if those ramp faster than planned.​

For enterprise and service‑provider customers, the key takeaway is that AI networking is no longer experimental. Cisco is committing billions into Silicon One, high‑density optics, and end‑to‑end architectures that span scale‑up in the rack, scale‑out across rows, and scale‑across between data centers separated by hundreds of kilometers.

As Patel reminded the Cisco Live audience when he walked through the evolution of training clusters, the network is what allows multiple racks and entire data centers to behave “like one large computer” for AI training runs—where restarting due to congestion or packet loss can cost millions. This validates Cisco’s decision to build its own silicon. By controlling the chip, the system, and the optics (via Acacia), Cisco is offering hyperscalers and neoclouds the “token generation factory” efficiency they require and can do that across their customer sentiment.

2. Platform‑first networking replaces product silos

A second major theme is Cisco’s insistence that it is now a platform company, not a holding company of loosely related products. This is something Patel has been working on since he took the role 18 months ago and described two “seismic” cultural shifts inside Cisco: building a platform where each new component compounds value across the stack, and committing to an open ecosystem even when that means integrating with competitors.​

That platform story was highlighted in several ways:

• Nexus One and HyperFabric : Cisco introduced Nexus One as a unified, cloud‑managed data center platform that ties on‑prem Nexus Dashboard with cloud‑based management, supports Cisco and third‑party silicon, and uses HyperFabric templates to standardize and automate AI and standard data center designs.​
• Unified campus and branch : On the campus side, Cisco has refreshed Wi‑Fi, switching, and routing, and is finally converging Catalyst and Meraki into a single cloud management plane, with new Secure Connect routers tailored to European access needs.​
• Observability and Splunk : The new Cisco data fabric, built around Splunk, is intended to unify logs, metrics, and events from networking, security, applications, and infrastructure into a common telemetry plane for AI operations and SecOps.​

CFO Mark Patterson reinforced this during the earnings call, noting that networking product orders grew more than 20% in Q2, driven by customers who are refreshing their entire “operational stack” rather than just buying individual switches. Robbins added that the campus refresh is “the top of the first inning”—a multiyear, multibillion‑dollar opportunity that is just beginning to hit the P&L.

One of the keynote quotes that best echoes Cisco’s platform mindset came from Patel when he said, “Delivering these capabilities at pace is what you need to thrive… you do that with platforms, not with products.”

3. Agentic AI drives AgenticOps and security rethink

If 2024 and 2025 was about chatbots, Cisco Live made 2026 about agents. Patel argued that we are moving from AI as a Q&A tool to AI agents as digital coworkers, soon followed by physical AI in the form of robots and intelligent devices in every workplace.​

That shift, he said, creates three structural constraints:

• Infrastructure constraints (power, compute, bandwidth, shells, memory).
• Trust deficits (safety, security, sovereignty).
• Data gaps (shortage of human‑generated training data, growing importance of enterprise and machine data).​

Cisco’s solution is AgenticOps—an AI‑driven operations model in which fleets of agents continuously watch telemetry, correlate anomalies, test fixes against digital twins, and only escalate to humans when necessary. At Cisco Live, the company showed early AgenticOps capabilities focused on troubleshooting, configuration, optimization, and validation across campus networks.​

On the earnings call, Robbins extended that concept to security operations, describing a “security premier edition” that uses agents to handle malware reversal, triage, and threat playbooking, with early capabilities targeted for alpha in April 2026.

AI also forces a re‑architecture of security. Cisco is fusing enforcement directly into the network fabric via “smart switches” that combine traditional packet forwarding with on‑box inspection and enforcement, and by using products like Secure Access and Hypershield to enforce zero‑trust policies, not only for people but also for AI agents and their interactions. At Cisco Live EMEA, Cisco extended its AI Defense offering to scan models and repositories for vulnerabilities, produce AI bills of materials, and apply semantic inspection to agentic interactions in SASE, blocking context‑dependent threats rather than just matching signatures.

For those on the fence regarding agentic, it’s important to heed the words of Patel: “Each one of us, at some point in time, will be the supervisors of these agents… knowledge work will happen with these digital coworkers.”

4. Sovereignty, trust, and the rise of sovereign AI

In EMEA, trust and sovereignty were more than talk—they were central to almost every discussion. This came across loud and clear at the event and in Davos in January. Cisco emphasized four dimensions of trust: security, innovation, execution, and sovereign control.​

The keynote highlighted Cisco’s “sovereign critical infrastructure” for EMEA, which includes a legal commitment that specific on‑prem products are fully air‑gapped, with no back doors, and will continue to function during crises. Robbins told investors that European customers are showing “strong interest” in this sovereign portfolio to keep sensitive data and critical infrastructure under local control amidst growing privacy, data governance and regulatory concerns.

On the AI side, sovereignty is becoming a meaningful growth vector. Beyond hyperscalers, Cisco booked 350 million in AI orders in Q2 from neocloud, sovereign, and enterprise customers, and it sees a pipeline over $2.5 billion for high‑performance AI infrastructure in these segments. Cisco also announced plans for a joint venture with AMD and Humane.

Patel framed sovereignty as one more reason Cisco can differentiate: Its combination of in‑house silicon, optics, systems, security and observability, coupled with local engineering centers and legal commitments, is pitched to give governments and large enterprises a vendor that can meet stringent sovereignty requirements without sacrificing performance.

This concept came up in an analyst Q&A session with Cisco’s Guy Diedrich, senior vice president and global innovation officer. Diedrich explained that the work Cisco does with governments and big brands around the world has created a situation where it’s one of the most trusted tech brands, and I agree with this sentiment. The more political distrust there is, the more it favors a company like Cisco. Robbins summed up that thought: “As AI adoption accelerates, concerns over privacy, data governance, and regulatory compliance are top of mind for our customers, making sovereign solutions an essential foundation for building digital trust.”

5. Security woven into the fabric, not bolted on

The “trust deficit” was identified as a major impediment to AI adoption. In Amsterdam, Cisco demonstrated how it is fusing security directly into the network fabric. The lead innovation is Hypershield and the ability to run security enforcement on a smart switch without adding latency.

Robbins confirmed that this security-centric approach is winning customers. Even as Splunk undergoes a transition to cloud subscriptions (which creates a temporary revenue drag), Cisco’s refreshed security portfolio—including XDR and AI Defense—is seeing rapid adoption.

“Over 1,000 new customers purchased these new security products in Q2, representing more than 100% growth quarter-over-quarter,” Robbins said.

6. Customer experience and talent as AI force multipliers

A less flashy, but strategically important, theme from Amsterdam is the focus on talent and customer experience as competitive weapons in an AI world. Cisco argued that AI agents don’t replace human potential, they unlock it—provided organizations rethink skills and operating models.​

The company is leaning heavily on its training and certification ecosystem in EMEA, with a commitment to train 10 million more people over the coming years to close the IT skills gap and diversify the tech workforce. On stage, customer executives—from global manufacturers to Real Madrid—spoke about using Cisco plus AI to redesign processes, improve observability from “business process down to the last switch port,” and modernize complex global networks under tight maintenance windows.​

On the earnings side, Cisco’s CX organization (services and support) is already using AI at scale. Patterson noted that AI‑assisted support has contributed to Cisco’s highest‑ever customer satisfaction scores, and that AI is actively deployed across sales, security and trust, supply chain, and corporate functions to drive efficiency and cost savings. Cisco also refreshed its CX portfolio with support tiers promising 15‑minute response times for critical incidents and data‑driven methodologies that, in internal studies, have shown 30%–50%reductions in unplanned downtime and implementation time.

Investor concerns and watch points

For investors, the Cisco Live and earnings commentary provide a compelling narrative, but the stock was down mid-single digits after hours. It’s important to note that even with the slight step back, the stock is up 11% year to date and 37% in the past 12 months. However, the earnings Q&A surfaced a few risks worth noting.

• Gross margin pressure from memory and mix. Patterson was explicit that rising memory prices and hardware mix are driving near‑term gross‑margin compression. Q2 non‑GAAP product gross margin fell 130 basis points year over year to 66.4%, and Q3 guidance embeds further pressure from memory before mitigation efforts and pricing changes fully flow through. While Cisco is raising prices and revising terms with partners and customers, there is a timing gap that will weigh on margins in the near term.
• Non-linear hyperscaler demand. While Cisco raised its AI orders outlook to “in excess of $5 billion” for FY26, Robbins repeatedly emphasized the “lumpy” nature of hyperscaler spending and the small number of customers driving these orders. That concentration introduces volatility in quarterly results and complicates traditional seasonality models, something Robbins pointed out when asked why Q4 sequential growth looks modest versus historical patterns.
• Software and ARR growth lagging the narrative. Despite the platform and subscription story, total software revenue grew just 2% and total ARR grew 3% year over year in Q2, with product ARR up 6%. Patterson suggested that software growth should improve beyond FY26 as cloud transitions mature, but for now, Cisco’s strongest growth is still in hardware networking rather than high‑margin software.

Final thoughts

I have been a Cisco watcher for decades, and I saw the company fumble its way around cloud and security in the past. This version of Cisco, looking to catch the AI wave, is different. By leaning into silicon, sovereign clouds, and agentic operations, Cisco has positioned itself as the essential foundation for AI. The challenge now is to manage the macro-headwinds of component costs while ensuring the campus refresh—which Robbins says is in the “top of the first inning”—continues to gain steam. If Cisco can execute on both, fiscal year 2026 will indeed be its strongest yet.

AI will create a rising tide, and I believe Cisco will be one of the big infrastructure winners. The company is now thinking “platform,” and the political infighting, like Catalyst versus Meraki, has been eliminated. Cisco customers should feel empowered to buy what they need today but know they won’t be locked into a particular product line should they decide to choose a different path in the future.

More Cisco news:

• Cisco extends AgenticOps across networking, security, observability products
• Cisco amps up Silicon One line, delivers new systems and optics for AI networking
• Takeaways from Cisco’s AI Summit
• Cisco: Infrastructure, trust, model development are key AI challenges
• AI, security tailwinds signal promising 2026 for Cisco
• Cisco adds intelligent policy enforcement to mesh firewall family
• Actively exploited Cisco UC bug requires immediate, version‑specific patching
• Cisco’s 2026 agenda prioritizes AI-ready infrastructure, connectivity
• Cisco finally patches seven-week-old zero-day flaw in Secure Email Gateway products
• Cisco routers knocked out due to Cloudflare DNS change
• Cisco identifies vulnerability in ISE network access control devices