AI RAN to Reach $35 B Over Next Five Years, According to Dell'Oro Group

Dell'Oro projects cumulative AI RAN revenue of $35B over the next five years, but expects AI RAN to have little or no impact on expanding the overall RAN market. GPU RAN forecasts were revised upward to a $1B+ market by the end of the period, with near-term focus on AI-for-RAN and single-purpose deployments. The outlook is supportive for incumbent RAN radio/baseband suppliers, which Dell'Oro says remain well-positioned for early AI RAN upgrades.

Analysis

The market should read this as a validation of the existing RAN refresh cycle, not a new revenue pool. If AI RAN is mostly an upgrade layer on top of incumbent radios/basebands, the economics accrue as mix and software attach, not as meaningful top-line expansion; that caps multiple expansion for the whole telecom equipment group and favors the largest installed bases over smaller challengers.

The second-order loser is the “GPU everywhere” thesis. A $1B-plus GPU RAN opportunity over five years is too small to matter for the AI supply chain in aggregate, so any read-through to NVDA-style infrastructure demand is likely overstated. More importantly, operators may use AI RAN to justify slower legacy capex while they pilot automation, which can extend the earnings trough for gear vendors by another 2-4 quarters even if the long-run tech narrative looks positive.

Contrarian view: consensus is likely overpricing the word “AI” and underpricing the phrase “won’t expand the overall market.” The first real catalyst is not more research forecasts but evidence of production deployments and order conversion; absent that, headline-driven enthusiasm should fade within days to weeks. The thesis is falsified only if carrier capex budgets reaccelerate or if AI-RAN software attach visibly lifts radio/baseband bookings over the next 2 earnings cycles.

Key Decisions for Investors

• No new long in DELL on this print; the read-through is too indirect to move FY27 fundamentals. Reassess only if management quantifies telco edge/server revenue in a way that changes segment growth.
• If ERIC or NOK gaps higher on AI-RAN enthusiasm, fade the move with 1-3 month put spreads or by selling calls into strength. Risk/reward favors downside unless order intake and backlog reaccelerate.
• Do not add NVDA/SMCI purely on telco AI-RAN optionality; the implied TAM is too small versus their existing AI drivers. Treat any telco linkage as noise unless disclosed orders become material.
• Watch for operator capex guidance and production deployment commentary over the next 2 earnings cycles. Upgrade the thesis only if AI-for-RAN moves from trial language to recurring bookings.