$CSCO

CSCO beat expectations on both EPS and revenue in Q2 and raised its FY2026 guidance. FY2026 revenue guidance was raised to the $61.2B–$61.7B range, above the Street’s $60.77B expectation. In my view, those are good numbers. In Q2, they highlighted $2.1B of hyperscaler-driven AI orders, and said hyperscaler AI orders are expected to exceed $5B. They also said hyperscaler AI infrastructure revenue in FY2026 will come in at $3B+, and that on the neocloud/sovereign/enterprise side they had $350M of AI orders in Q2 and a $2.5B+ pipeline. On the order side, product orders were up 18% YoY, while service provider & cloud segment orders jumped 65%, signaling demand is re-accelerating. The recurring revenue base is growing: RPO was $43.4B (+5% YoY) and ARR was $31B (+3% YoY). Networking is 54% of revenue, while services + security reached 37.3%. They raised the dividend and are doing strong share buybacks. As hardware sales grow, the recurring revenue base expands through maintenance/service and security subscriptions. There’s also the claim that the Splunk acquisition connects observability/security into a platform story.
Now to the key issue: they do data center networking. In the recent selloff, what the market got stuck on most was memory/component prices pressuring gross margin, and Cisco’s short-term margin guidance coming down. That caused a market that’s excited about “AI is exploding” to also haircut the stock because “costs are exploding too.” Also, their competitors in every lane are the strongest. The back end, where thousands of GPUs talk to each other and where the real money gets spent, is traditionally dominated by Nvidia (InfiniBand) and Arista Networks (Ethernet). Hyperscalers can prefer more specialized or cheaper white-box solutions instead of Cisco. To avoid missing this market, Cisco is partnering with NVIDIA and is trying to take share from Arista by pushing the thesis that Ethernet is enough for AI.
On the other hand, as GPU clusters scale, networking becomes as critical as compute. Hyperscalers will do roughly $610B–$700B of CapEx in 2026. Some analyses say a meaningful part of the 2026 CapEx increase is coming from memory price inflation; that could mean spending rises while the physical amount of equipment bought doesn’t rise as much. CSCO may not get hit by AI disruption as much as SaaS names do. In my view, while SaaS companies carry the risk of becoming unnecessary because of AI, companies like Cisco can become indispensable thanks to AI. The only risk is Cisco’s tech falling behind Arista or Nvidia, not the market disappearing. Cisco is selling the picks and shovels, but slowly.
Technical Analysis ...