Nokia and Ericsson Warn That AI Integration Is Driving Up Telecom Infrastructure Costs

By Hector Herrera | June 9, 2026 | Telecom

The two largest telecom infrastructure vendors in the world have issued a warning that cuts against their own marketing narratives: AI integration in network equipment is making telecom infrastructure more expensive, not cheaper — at least for now. Both Nokia [and Ericsson](/telecom/t-mobile-ericsson-ai-native-ran-commercial-first) have publicly flagged that AI components are driving up hardware prices and extending supply chain lead times, complicating operator capital expenditure planning precisely as the industry prepares for AI-native 6G deployments.

The warning is notable for its source. Nokia and Ericsson are simultaneously selling operators on AI-native networks as their primary growth narrative — promising efficiency gains, automated optimization, and reduced operating costs. Admitting that near-term costs are rising while making that pitch takes some competitive self-awareness.

What They Said

The warnings from both companies center on two near-term cost pressures, according to Light Reading's reporting:

Component price increases. AI acceleration chips — GPUs and specialized inference processors — being embedded in next-generation radio access network (RAN) equipment and core network hardware are commanding premium prices in a supply-constrained market. Operators buying network hardware in 2026 are paying more per unit than they would have for equivalent non-AI hardware in 2024.

Supply chain lead time extensions. AI-optimized components, particularly those integrating NVIDIA or custom silicon, are subject to the same GPU supply dynamics that are affecting every sector deploying AI at scale. Lead times for AI-integrated network hardware have extended to 18-24 months in some categories, up from 6-12 months for prior-generation equipment. For operators trying to plan 12-18 month rollout schedules, that is a significant planning constraint.

The Tension in the Pitch

Ericsson and Nokia have both made AI-native networking a central part of their pitch to mobile operators over the last 18 months. The argument: AI integrated into the RAN (Radio Access Network) enables real-time optimization of spectrum allocation, traffic routing, and power consumption — producing operating expense savings that, over a deployment lifecycle, exceed the upfront hardware premium.

That argument may be correct over a 5-7 year view. But the near-term cost signal cuts the other way, and operators are running capex budgets on 12-18 month horizons, not 5-7 year ones.

The vendors are managing a classic technology adoption tension: the long-run efficiency gains require near-term cost acceptance that buyers are reluctant to absorb, particularly when the efficiency gains are modeled projections rather than demonstrated results from comparable deployments.

Operator Impact

For mobile operators globally — but particularly Tier 1 operators in the U.S., Europe, and Asia-Pacific who are the primary customers for Nokia and Ericsson — the warnings create several practical planning problems:

Capex planning uncertainty. If AI-integrated hardware costs are not stable and lead times are not predictable, operators cannot accurately forecast network upgrade program budgets. Some are responding by extending the life of existing equipment rather than accelerating AI-native upgrades — the opposite of what the vendors want.

6G timeline pressure. AI-native architecture is foundational to 6G standard design, not an optional add-on. Operators who delay AI integration in their 5G networks to avoid near-term cost increases may find themselves further behind on 6G readiness when standards finalize around 2028-2030. The cost avoidance now may compound into a larger catch-up cost later.

Vendor negotiation dynamics. When two major vendors issue similar warnings simultaneously, it can read as coordinated expectation management — a way of preparing buyers for price increases rather than a neutral market observation. Operators are acutely aware of this dynamic, and procurement teams at major carriers are likely already treating the warnings as anchoring tactics.

What the Warnings Reveal About the Market

Beyond the immediate cost signal, the Nokia and Ericsson statements reveal something important about where the AI-in-telecom narrative actually stands in mid-2026: the efficiency promise is still primarily a promise.

AI-native network equipment is being deployed, but the large-scale operating expense savings that justified the hardware premium have not yet been independently validated at the network-wide scale operators need to see before committing full deployment budgets. The vendors' own admission of near-term cost pressure suggests the ROI case depends heavily on future performance data that does not yet exist in sufficient volume.

This does not mean the technology is wrong — the engineering case for AI-optimized network operations is solid. It means the market is in the adoption phase where costs are real and benefits are projected, which is historically when adoption slows unless vendors can accelerate the evidence base.

What to Watch

Both vendors' Q2 2026 earnings calls will reveal how the cost warnings are affecting order books — specifically whether the lead time and price increases are translating into delayed purchase orders or whether large-operator deals are holding. Also watch for whether operators begin publicly pushing back on vendor pricing, which would shift negotiation leverage and potentially accelerate component price normalization. The GSMA's annual MWC Americas event in the fall will be the next major forum where operator capex plans are discussed publicly, and AI hardware cost dynamics will likely dominate the agenda.

Sources: Light Reading — AI Is Making Telecom More Expensive, Warn Ericsson and Nokia