Nvidia's Networking Revenue Soars 263% to $11B as AI Data‑Center Demand Explodes

Why It Matters

The explosive growth in Nvidia's networking revenue signals that AI infrastructure is maturing beyond the GPU itself. High‑speed interconnects, Ethernet switches and power‑management hardware are emerging as essential, capital‑intensive components, expanding the AI supply chain and creating new investment opportunities. If data‑center spending slows, the interdependence of these hardware layers could amplify the impact on earnings across the ecosystem, making supply‑chain resilience a focal point for both operators and investors.

Key Takeaways

• Nvidia's networking revenue rose 263% YoY to $11 billion in the latest quarter.
• Data‑center revenue grew 75% YoY, with CEO Jensen Huang forecasting $1 trillion in GPU orders through 2027.
• Arista Networks posted $9 billion in 2025 revenue, with AI networking sales expected to hit $3.2 billion in 2026.
• Vertiv's organic orders surged 252% YoY and its backlog reached $15 billion.
• Analysts project a 21% annualized earnings growth for AI‑focused hardware firms over the next few years.

Pulse Analysis

Nvidia's networking surge is less a surprise than a logical extension of the AI arms race. Early AI deployments proved that raw compute without adequate bandwidth quickly hits a ceiling; the industry responded by building a parallel market for interconnects. Nvidia's ability to bundle NVLink and Spectrum‑X with its GPUs gives it a strategic advantage, but it also opens the door for pure‑play networking firms like Arista to capture a slice of the same spend. The competition will likely shift from price wars to differentiation through software ecosystems—Arista's EOS versus Nvidia's DGX stack—forcing both to invest heavily in integration capabilities.

Supply‑chain risk is now a multi‑layered problem. While semiconductor fab capacity has dominated headlines, the power‑delivery and cooling bottlenecks highlighted by Vertiv suggest that data‑center operators could face constraints that are harder to scale quickly. Companies that can offer end‑to‑end solutions—combining compute, networking, and thermal management—will command premium pricing and tighter customer lock‑in.

Looking ahead, the market's next inflection point will be the rollout of Blackwell and Rubin GPUs. Their higher density and performance will amplify demand for both high‑speed interconnects and robust power infrastructure. If cloud providers and enterprises maintain their current capex trajectory, we could see networking and power‑management revenues double again within two years. Conversely, any macro‑economic slowdown that curtails data‑center expansion could expose the fragility of this newly broadened AI supply chain, prompting a re‑evaluation of valuation multiples across the sector.