I just saw the EU plan to force a Huawei/ZTE phase‑out—are you ready for the 2026 crunch?

Why this matters now

The European Commission is moving to make its 2020 advice to exclude “high‑risk” suppliers-mainly Huawei and ZTE-from 5G and next‑gen networks a binding EU law. If adopted, member states that don’t comply could face infringement procedures and fines. For telecom operators and AI leaders, this shifts the Huawei/ZTE question from a policy debate to an execution problem with budgets, timelines, and tooling on the line.

This matters because it accelerates supplier diversification across radio access, core, and increasingly fixed/fiber networks. Expect procurement resets, rip‑and‑replace programs, and re‑platforming of AI‑driven operations built on vendor‑specific telemetry and APIs. The likely adoption window (late 2025 to early 2026, subject to negotiations) creates a tight runway that overlaps with NIS2 and other security obligations.

Key takeaways

• Scope widens: beyond 5G core to RAN and, in many cases, fixed/fiber rollouts and upgrades.
• Enforcement shifts from peer pressure to penalties: infringement actions, funding conditions, and deadlines.
• Budget impact is material: multi‑year programs measured in billions at a country level; operators should plan a 10-15% contingency on network modernization capex.
• AI ops will break if not refactored: telemetry schemas, model features, and runbooks tied to Huawei/ZTE tooling must be re‑engineered.
• Winners: Nokia, Ericsson, Samsung, and credible Open RAN stacks; higher demand strains lead times (9-15 months for some gear).
• Geopolitical risk remains: potential patent/licensing disputes and commercial retaliation could add cost and delay.

Breaking down the announcement

In 2020, the EU’s non‑binding “5G Toolbox” urged limiting “high‑risk” vendors in critical network functions. Adoption was uneven: some countries moved quickly, others grandfathered existing deployments. The Commission now aims to convert that guidance into a regulation or directive with binding force, escalating non‑compliance to legal infringement. Drafting is underway, with political momentum reinforced by national moves (notably Germany’s decision to phase out key Huawei/ZTE components by 2026).

The working assumption among operators: a phased approach with transitional periods (often 12-36 months from adoption) and stricter controls on the 5G core and management planes first, followed by RAN and fixed access. The Commission has also signaled that EU funding for external connectivity projects may be conditioned on avoiding high‑risk vendors, extending the policy’s influence beyond the bloc’s borders.

What changes for AI‑driven network operations

Modern networks lean on AI for assurance, anomaly detection, traffic engineering, and zero‑touch provisioning. Much of that automation is wired to vendor‑specific telemetry formats, counters, config models, and fault codes. A forced supplier swap means:

• Data refactoring: feature stores and pipelines must align to new MIBs, YANG models, and event schemas; plan for model drift during cutover.
• Runbook rewrites: remediation playbooks and policy engines tied to proprietary controllers need vendor‑neutral abstractions or new integrations.
• Retraining and validation: expect 8–12 weeks of model retraining per major domain (RAN, transport, core) plus parallel A/B shadowing to avoid service regressions.
• Observability upgrades: invest in standardized telemetry (gNMI, OpenConfig) and SBOM-based firmware governance to meet NIS2 and emerging EU cyber requirements.

Operationally, the biggest risk is a performance dip during migration. Early Open RAN deployments reported higher energy draw and tuning overhead; newer releases have narrowed the gap, but achieving parity still demands careful RF optimization and tight cloud/edge placement to keep latency within SLA.

Vendor landscape and fit by use case

For brownfield 5G, Nokia and Ericsson offer the most direct swap paths, mature OSS/BSS hooks, and integrated AI assurance (e.g., AVA and Expert Analytics). Samsung is increasingly credible in RAN, with competitive mid‑band performance and vRAN traction. For disaggregated options, Open RAN stacks from Mavenir, NEC/Rakuten Symphony, and operators’ own integrations can reduce lock‑in but raise integration and tuning workload.

• Lowest execution risk: single‑vendor swaps (Nokia/Ericsson) for core and critical RAN; faster compliance, higher TCO predictability.
• Strategic flexibility: multi‑vendor/Open RAN in new builds or secondary markets; requires lab investment and SI partnerships.
• Fixed/fiber: expect tighter scrutiny of OLT/ONT supply. European vendors and established global suppliers will see uplift; ensure TR‑369/USP and open APIs for AI ops.

Costs, timelines, and risks

Country‑level rip‑and‑replace programs can total several billion euros when factoring equipment, services, and operational churn. For an individual Tier‑1 operator with significant Huawei RAN share (>30%), plan for a 24–36 month program, with 15–25% of sites touched per year to manage outage windows. Lead times for radios and transport gear have improved since 2022 but remain vulnerable to component shocks.

• Legal/compliance: unclear grace periods create planning risk; expect interim milestones (core first, then RAN).
• IP exposure: Huawei’s patent portfolio is extensive; watch for licensing shifts that could affect handset and network costs.
• Supply concentration: replacing one dependency with another is a real risk; mitigate via multi‑vendor RAN where feasible and strict API portability.
• Customer impact: churn can spike if QoE dips during migration; invest in proactive care and SLA management.

Recommendations for operators and enterprise buyers

• Stand up a board‑level program now: inventory Huawei/ZTE assets, map AI/telemetry dependencies, and model three compliance scenarios (12, 24, 36 months).
• Design for portability: normalize telemetry via OpenConfig/gNMI, replace vendor‑specific runbooks with policy-as-code, and refactor feature stores.
• Secure lab capacity: create a multi‑vendor testbed to validate performance, energy use, and AI model behavior before field rollout.
• Lock in supply and SLAs: negotiate delivery schedules, spares, and performance guarantees with at least two approved vendors; include interoperability KPIs.
• Budget realistically: add 10–15% contingency to network modernization capex; sequence upgrades to align with NIS2 and maintenance windows.
• Plan customer protection: monitor QoE in real time, pre‑empt hot spots with temporary capacity, and communicate migration timelines transparently.

Bottom line: treat the EU’s shift as a near‑certain constraint on your 2026–2028 roadmap. The operators that win will use this forced change to standardize telemetry, harden security, and reduce future switching costs-turning a compliance mandate into durable operational leverage.