What happened

  • IDC forecasts $623 bn in enterprise AI infrastructure spending in 2026, a 31% rise on 2025, with networking the fastest-growing category, according to IDC analyst Leslie Rosenberg.
  • Dell’Oro Group reports worldwide enterprise networking revenue grew 15% year-over-year in the first half of 2026.
  • All five major networking segments, network security, branch routing, wireless LAN, campus switching and data-center switching, posted double-digit revenue gains, though port shipments fell 5% in Q2.
  • Average selling prices rose 19% as DDR4 memory shortages tightened component supply.
  • Wi-Fi 7 access-point shipments drove wireless LAN growth, while campus switch revenue rose despite flat port shipments.
  • Large organisations are upgrading front-end data-center switches for general-purpose servers, while smaller enterprises pull back.
  • Security tooling for AI systems is expected to grow from $200 m in 2025 to $8 bn over five years, per Dell’Oro senior director Mauricio Sanchez.

Why it matters

Enterprises that cannot match equipment demand with network capacity risk bottlenecks in AI workloads. Network architecture upgrades and enhanced north-south and east-west security are now priorities, yet many firms lack mature use cases and verified infrastructure. The mismatch between revenue growth and port shipments suggests that spending is often on higher-priced, lower-volume gear, potentially inflating budgets without delivering needed capacity. Supply-chain constraints on memory and components further limit the speed of deployment, making it harder for organisations to meet AI performance targets.

What this means for enterprise IT teams

  • Teams must audit existing network fabric to confirm it can support the higher-speed links required for AI workloads before committing to projects.
  • Budget planning should balance new-money allocations with re-prioritisation of legacy workloads, as AI and security are the least likely to face cuts.
  • Consider circularity programmes, low-percent financing or leasing options from vendors such as HPE, Juniper and Cisco to stretch limited capital.
  • Prioritise security tooling for both north-south and east-west traffic to protect expanding AI workloads.
  • Align procurement with realistic port-shipment forecasts to avoid over-paying for equipment that cannot be fully utilised.

In the words of Leslie Rosenberg, IDC analyst: "What we are primarily seeing is that ambition outpaces readiness."

The bottom line

AI networking spend is soaring, but many enterprises are not ready to turn that spend into effective infrastructure. Without clear use cases, mature network designs and realistic procurement strategies, the surge in budgets may not translate into the performance gains required for AI workloads. IT leaders need to close the readiness gap now to avoid costly overruns and under-utilised assets.

Sources

  • networkworld.com