6G mobile networks will be built AI-native from inception, SDxCentral reported in framing preparation requirements for operators. The publication argues intelligence must move from network management features into the radio itself.

What does "AI-native" actually mean for 6G architecture?

The term signals a departure from bolt-on automation. In 5G non-standalone and standalone deployments, AI arrived as an optimization layer atop human-defined parameters. Operators added machine learning models after standards matured. The 6G approach reverses this sequence, embedding inference capability at the protocol layer from initial specification release.

SDxCentral's analysis positions this as a fundamental redesign rather than an incremental upgrade. Radio units, baseband processing, and core network functions each require native machine learning support. The architecture treats AI as a primary interface, not an auxiliary tool.

How does this differ from current 5G automation?

Existing networks rely on closed-loop automation through service management and orchestration frameworks. 3GPP specifications define analytics functions but leave AI implementation to vendor interpretation. Operators deploy vendor-specific models against generic northbound interfaces.

The 6G model requires standardized AI interfaces across the radio access network. 3GPP Release 21 and subsequent releases are expected to incorporate AI-native interfaces as normative features, shaping vendor product roadmaps through 2028 and beyond.

What preparation steps do operators face now?

Workforce planning represents the immediate operational pressure. Radio engineers trained on 5G New Radio must acquire skills in model lifecycle management. Training data curation, inference optimization, and drift detection become core operational competencies, exposing a material deployment risk if unaddressed.

Capital expenditure cycles must align with the architectural pivot. Procurement teams evaluating 2025 and 2026 vendor roadmaps need AI-native criteria in tender documents. Legacy radio access network benchmarks will fail to capture embedded inference value. Per-cell compute budgets will rise as inference tasks shift from centralized data centers to distributed radio units.

Spectrum acquisition and regulatory timelines add further constraints. National authorities must update type approval frameworks to certify AI-native equipment. Certification procedures built for deterministic hardware require revision for probabilistic software-defined functions.

Which vendors face the largest adjustments?

Silicon suppliers bear the heaviest technical burden. Baseband chipsets must deliver inference throughput alongside traditional signal processing pipelines. Software stacks require purpose-built compilers and runtime environments for AI workloads at the cell site.

Network equipment vendors face integration challenges across radio, transport, and core domains. Vendor lock-in risk rises as AI model portability remains undefined across multi-vendor deployments.

What timeline governs commercial 6G services?

The ITU-R IMT-2030 vision framework identifies AI-native capabilities as a core design principle. Industry coordination through 3GPP working groups will determine specification maturity, with ITU-R specification finalization expected to anchor subsequent national spectrum allocation decisions. South Korea, the European Union, and Japan have each published indicative 6G launch targets through 2030, though these remain subject to standardization completion.

SDxCentral's framing asks operators to begin vendor engagement now. Procurement teams should evaluate supplier roadmaps against AI-native criteria rather than legacy benchmarks. Contract negotiations will require new service-level agreements covering model performance thresholds, retraining obligations, and inference drift detection.

The industry's preparation window narrows with each standardization milestone. Operators deferring engagement risk inheriting vendor-defined AI architectures that may not align with operational requirements.