Hook
A Cisco engineer loads a datacenter, eyes fixed on the network switch. The AI order book is bursting—7 billion in the last quarter alone. Yet the stock price is bleeding.
This is the paradox of the AI supercycle: record revenue, tumbling shares. The market is punishing what it once celebrated.
I’ve seen this pattern before—in the 2017 ICO fever, when every protocol with a whitepaper pumped, then dumped. The difference now? Cisco’s numbers are real. The AI orders are real. But the story is more complex than a headline.
Context
Cisco is the world’s largest networking hardware company. It builds the switches, routers, and optical modules that connect data centers. For decades, it was the backbone of the internet. Then came the cloud era, and giants like Arista, Juniper, and Broadcom eroded its dominance.
Now, the AI supercycle is reshaping the narrative. The explosion of GPU clusters—from 1,000 to 100,000 cards per installation—demands a new generation of network infrastructure: faster, lower-latency, and scalable. Cisco’s Nexus 9000 series and 800G optical modules are positioned to serve this demand.
But the company is also in the middle of a strategic transformation. The $28 billion acquisition of Splunk in March 2024 moved Cisco from a “hardware box” company to a “software + subscription + security” hybrid. The AI order book is a mix of hardware for hyperscalers and software for observability and security.
Core
Let’s cut through the noise. The article is based on Cisco’s FY2025 Q2 earnings, likely released on February 12, 2025. The numbers: total revenue of $14 billion, up 9% year-over-year; non-GAAP EPS up 4%; full-year guidance raised to $56.0–$56.5 billion. AI orders hit $700 million in the quarter, up from $500 million in the prior quarter.
But here’s the kicker: the stock fell after the earnings call. Why?
First, the “record” revenue is inflated by Splunk’s contribution. Without Splunk, organic growth was flat to low single digits. The AI orders are real, but they’re a tiny fraction of Cisco’s total revenue—about 5% of the quarterly total. The rest comes from legacy enterprise networking, which is shrinking.
Second, the market is shifting from “AI narrative” to “AI precision.” Investors no longer reward companies just for being in the AI ecosystem. They demand proof of durable growth, high margins, and competitive moats. Cisco’s AI business has a gross margin of 65-70%, lower than its software business (80%+). The hyperscalers—Microsoft, Google, Meta—account for 80% of Cisco’s AI orders. These giants have immense bargaining power, pressuring margins.
Third, the competitive landscape is brutal. Arista Networks is the leader in AI data center networking, with a reputation for performance and reliability. Cisco is a “follower” in AI networking, not a leader. NVIDIA’s proprietary NVLink and Spectrum-X platforms are also compressing the value of traditional networking vendors.
Contrarian
Here’s the angle the market misses: the AI supercycle is not just about GPU clusters. It’s about the infrastructure that connects them. As clusters scale to 100,000 GPUs, the networking equipment’s share of total CAPEX rises from 10% to 15-20%. This is a structural upgrade cycle, not a one-time surge.
Cisco’s real opportunity is in the “enterprise AI” wave. When corporations—not just hyperscalers—start building their own AI infrastructure, Cisco’s vast customer base of 700,000 enterprise clients becomes a distribution advantage. This is a 18-36 month play, but the early signals are there: Cisco’s enterprise AI orders are still below 5% of total AI orders, but they’re growing.
Another blind spot: Splunk’s integration. Cisco+ Splunk creates a unique “observability + security + networking” platform. In the AI world, companies need to monitor their GPU clusters, detect anomalies, and secure data. No other vendor offers this combination. If Cisco can execute on cross-selling, Splunk’s revenue could double in two years.
But the market is cynical. It sees Cisco as a legacy company trying to catch up. The stock’s PE ratio of 20-25x is a discount to Arista’s 45x and NVIDIA’s 50x. This reflects the market’s skepticism about Cisco’s ability to transform.
Takeaway
So, what’s the next signal to watch? The FY2025 Q3 earnings, expected in May 2025. If AI orders accelerate to $1 billion+ per quarter, and if Splunk’s contribution to recurring revenue rises above 15%, the narrative will shift.
But for now, the market is right to be cautious. Cisco’s AI supercycle is real, but it’s a slow-moving wave, not a tsunami. The stock’s pullback is a buy signal for patient investors who believe in the enterprise AI wave. For everyone else, it’s a reminder that in the crypto world—and in tech—narrative only lasts as long as the execution.
The fork in the road where code met chaos and won.
