S15.7 · Telecom & Connectivity
RAN, core and transport hardware vendors navigating flat capex cycles, geopolitically gated market access, and the emergence of AI-RAN.
Telecom network equipment providers manufacture the radio access network (RAN), core and transport hardware that carriers build and operate their networks from. The market runs an estimated $85-95B in 2024/2025 (Dell'Oro Group-tracked estimates); no single public aggregate is authoritative, because vendors report their product mix differently. Read this segment as hardware on a capex cycle, not as telecom.
Growth is flat to low-single-digit while carriers digest the capital spent on 5G build-outs; the next leg depends on 5G-Advanced, AI-RAN and open RAN adoption. Manufacturing and R&D concentrate in China, the Nordics and South Korea; deployment follows carrier capex cycles wherever they run. Concentration is extreme — Huawei and Ericsson together hold roughly two-thirds of global RAN share (Dell'Oro, 2025), with Nokia, ZTE and Samsung rounding out the field. Market access, meanwhile, runs on security grounds in opposite directions: the China market is largely closed to Western vendors, and Western markets are largely closed to Huawei and ZTE. Security and trade policy function as de facto market-access control alongside product competition, and the gate matters as much as the share.
Semiconductors and components sit upstream; mobile and fixed operators and the field-services firms that install the equipment sit downstream. The economics are the discipline here: this is low-margin hardware whose revenue tracks discrete carrier build cycles — 3G, 4G, 5G — rather than any recurring subscription stream, which makes the business considerably lumpier than most of the rest of telecom and connectivity.
AI-RAN — inference workloads running directly on radio access network infrastructure — is repricing base-station compute specifications and creating a new AI-capacity revenue line for equipment vendors, distinct from the margins on traditional radio hardware. The effect is demand-side: it expands what carriers need to buy, and automates nothing the vendors currently do.
Where the new money goes depends on who is moving. Open RAN investment aims to shift share away from the incumbent vendors above by disaggregating hardware and software; AI-RAN silicon opens the door for chip-adjacent entrants moving into telecom infrastructure. Both expand the field rather than defend it. The practical effect is a blurring line between traditional telecom equipment and AI compute infrastructure, since the same base-station hardware is increasingly expected to run general inference workloads alongside radio functions. Commercialization of AI-RAN sits on a 2-5 year horizon, and it favors vendors already selling into both telecom infrastructure and general-purpose compute over pure-play radio specialists.