Cisco targets $7.5 billion in AI infrastructure revenue despite margin pressure

Cisco is pivoting its business model to capture the capital expenditure boom in data centers, prioritizing volume in the hyperscaler segment over traditional high-margin hardware sales. This shift forces a reliance on operating efficiency to maintain profitability as product margins face headwinds from memory costs and a less favorable sales mix. For the broader streaming and cloud ecosystem, Cisco's aggressive pricing and Silicon One adoption signal a tightening race with Arista Networks for the underlying plumbing of AI-driven video workloads. Industry observers should monitor the pace of order conversion and whether operating expenses continue to decline as a share of revenue through the first half of fiscal 2027.

Cisco's push into hyperscaler AI infrastructure places it in direct competition with Arista Networks, which has been gaining share in high-speed data center switching. Arista reported record quarterly revenue of $2.2 billion in Q2 2026, driven by demand for its 400G and 800G Ethernet platforms from cloud providers building AI clusters. Arista's CEO Jayshree Ullal has consistently emphasized that the company's merchant-silicon approach gives it a cost advantage over vertically integrated competitors, a positioning that directly challenges Cisco's Silicon One strategy in the hyperscaler segment.

The broader AI infrastructure buildout is reshaping capital expenditure priorities across the networking industry. Omdia analysts estimate that telco spending on RAN products declined from $45 billion in 2022 to $35 billion in 2025, freeing up operator budgets but also signaling that data center networking is absorbing a larger share of infrastructure investment. Cisco's decision to raise hardware prices to offset memory cost inflation reflects a wider industry trend where streaming hardware price increases are squeezing margins even as demand accelerates. Ericsson CEO Börje Ekholm stated on the company's Q1 2026 earnings call that Ericsson's AI exposure would come from traffic development rather than data center expansion, underscoring how different infrastructure vendors are positioning themselves along the AI value chain.

On the technical front, Cisco's Silicon One architecture is designed to compete with merchant silicon offerings from Broadcom and custom ASICs from hyperscalers themselves. The company's ability to maintain its $7.5 billion AI infrastructure revenue target will depend on whether it can convert its order backlog into recognized revenue faster than competitors. Ericsson's June 2026 Mobility Report projects that uplink traffic will be three times higher in 2031 compared to 2025 due to AI-driven applications, a trend that will require substantial network infrastructure upgrades and could drive additional demand for high-capacity switching and routing equipment from both Cisco and Arista as service providers re-dimension their networks for workloads.