Applied Network Technology
Every line of an operator's network agenda, from the site-build program to the newest RAN technology, is an investment decision that lands across three pillars at once. This page maps where the spend is required, and where it quietly goes missing.
The hard part is not the technology
Mobile operators do not lack a technology roadmap. They lack a way to decide which investments to fund first, and what each one actually costs once the full scope is counted.
Start with where the capital goes. Roughly 70% of network CapEx is consumed by the deployment program itself: site acquisition, construction, integration, and the field operations that follow. That is why the matrix below opens with deployment rather than the newest technology. An agenda that debates RAN technology and AI-Ops potential while treating the site-build program as a fixed cost is optimizing the smaller half of the budget.
The technologies matter too, and every one in the matrix is mature enough to deploy. The open question is rarely whether 5G Standalone core, network slicing, or generative AI works. It is whether the operator has funded the operating-model change and the data foundation each one depends on. A 5G Standalone core returns new service revenue only when charging, assurance, and the commercial process are ready to sell against it. An AI-driven assurance program returns the MTTR improvement its business case promised only when the historical ticket data is clean enough to train on.
This page frames the network agenda the way an investment committee should see it. Not a catalog of technologies but a register of funding decisions: deployment programs and modernization technologies, each drawing on three distinct pillars of spend.
Three pillars, one decision
A checkmark in the matrix marks a pillar where a technology requires material, separately-funded investment. Most technologies require two or three.
The network itself
Radios, antennas, cell sites, transport, the 5G core, the cloud platforms that run cloud-native functions, and the deployment program that builds and modifies them. The largest category of operator CapEx and OpEx by a wide margin. The pillar that is always budgeted, because it is visible and vendors compete to supply it.
How the network is run
Capital planning, site build and time-to-market, NOC and field operations, vendor integration models, governance, and the skills the workforce needs. This is where deployment turns into operations. It is also the pillar most often left out of the original business case.
The foundation under automation
Telemetry, analytics pipelines, and the AI/ML foundation, spanning RAN, transport, and core. Cross-domain data visibility is the prerequisite for assurance, automation, and every closed-loop use case. Underfund it and the intelligent technologies above it cannot perform.
Technology crossing pillars
Twenty-four items on the North American network agenda, mapped to the pillars each one draws on: four deployment programs that consume roughly 70% of capital, plus twenty modernization technologies. The pattern matters more than any single row. Every one of the twenty-four requires investment in two or three pillars at once; a program funded in only one will stall.
| Technology | Infrastructure | Business Process | Data Intelligence |
|---|---|---|---|
| A · Network Deployment & Capital Programs | |||
| Capital Planning & Site SelectionWhere capital is committed: demand modeling, coverage-gap analysis, AI-driven site scoring | ✓ | ✓ | |
| Site Construction & BuildNew build and modification programs: civil works, towers, equipment install — the largest CapEx line | ✓ | ✓ | ✓ |
| Integration & Field OperationsCommissioning, on-air verification, technician dispatch, and field service management | ✓ | ✓ | ✓ |
| Network DecommissioningLegacy technology shutdown, site decom, lease exit, and asset recovery | ✓ | ✓ | |
| B · RAN Modernization & Intelligence | |||
| O-RAN / vRAN ModernizationCloud-native virtualized RAN at scale today; open multi-vendor RU/DU/CU interfaces now converging | ✓ | ✓ | |
| SMO — Service Management & OrchestrationRAN-domain orchestration framework; hosts the Non-RT RIC and rApps | ✓ | ✓ | ✓ |
| RIC — RAN Intelligent ControllerNon-RT (rApps, >1s) and Near-RT (xApps, 10ms–1s) control loops | ✓ | ✓ | |
| AI-RANAI inference co-located with the radio; accelerated compute at the RU/DU | ✓ | ✓ | |
| Massive MIMO & Advanced Antenna SystemsMulti-antenna arrays and beamforming for spectral efficiency and capacity | ✓ | ✓ | |
| C · Core, Cloud & Connectivity | |||
| 5G Standalone (SA) CoreCloud-native 5G core on a service-based architecture (3GPP Rel-15+) | ✓ | ✓ | ✓ |
| Cloud-Native Network FunctionsContainerized network functions on a CaaS platform; CI/CD operating model | ✓ | ✓ | ✓ |
| Edge Compute (MEC)Distributed compute near the user; Multi-access Edge Computing (ETSI ISG MEC) | ✓ | ✓ | ✓ |
| Network SlicingEnd-to-end logical networks per use case (3GPP TS 28.530; GSMA GST/NEST) | ✓ | ✓ | ✓ |
| Private 5G NetworksDedicated enterprise and industrial networks; a distinct go-to-market | ✓ | ✓ | ✓ |
| D · Operations, Assurance & Automation | |||
| Network Automation & OrchestrationClosed-loop, intent-based operations (ETSI ZSM); the path toward L4 autonomy | ✓ | ✓ | ✓ |
| AI-Driven Service AssuranceFault, performance, and SLA assurance with ML; alarm correlation and RCA | ✓ | ✓ | |
| NWDAF — Network Data Analytics FunctionStandardized 5G core analytics (3GPP TS 23.288) | ✓ | ✓ | |
| Network Digital TwinSynchronized data model for simulation, what-if analysis, and change validation | ✓ | ✓ | ✓ |
| Energy Efficiency & SustainabilityAI power-saving, carrier and MIMO-symbol shutdown; ESG reporting | ✓ | ✓ | ✓ |
| E · AI, Security & Service Enablement | |||
| Generative AILLM-assisted operations, documentation, and design; RAG over network data | ✓ | ✓ | ✓ |
| Agentic AI AssistantsTool-using AI agents for NOC, field, and operational support workflows | ✓ | ✓ | |
| Network Security & Zero Trust5G security architecture (3GPP SA3) and zero-trust controls across the network | ✓ | ✓ | ✓ |
| Network APIs / Open GatewayExposed network capabilities for developers and partners (GSMA Open Gateway, CAMARA) | ✓ | ✓ | ✓ |
| RedCap & Cellular IoTReduced-capability NR for mid-tier devices (3GPP Rel-17); the cellular IoT device ladder | ✓ | ✓ | |
Every program and technology requires investment in at least two pillars, many all three.
That is the case for treating the network agenda as a portfolio decision rather than a procurement list. It holds even for Site Construction & Build, the largest CapEx line of all: the schedule slips and the cost overruns trace to permitting cycle time, contractor performance, and as-built data quality, not a shortage of steel and radios. Read the matrix by column and the pattern repeats. Infrastructure is budgeted and vendor-supplied, so it rarely stalls a program. The business-process and data columns are where initiatives fail. On network AI/ML programs, cross-domain data normalization alone consistently consumes 40–60% of total project effort, the line item most often missing from the original business case.
Where AcropolisDocs comes in
AcropolisDocs advises mobile operators on exactly this decision: which programs and technologies to fund, in what order, and what each one requires across all three pillars before it can return value.
A typical engagement opens with a pillar gap assessment, scoring the operator's current infrastructure, process, and data readiness against the deployment programs and technologies on its agenda. It produces a sequenced investment roadmap anchored to the TM Forum Autonomous Networks maturity model. Most North American Tier 1 operators sit at L1–L2 today. The route to L4, where closed-loop operations run cross-domain with human oversight on exceptions, is the roadmap this matrix is built to support.
Explore consulting servicesPillar gap assessment
Where infrastructure, process, and data readiness stand against the agenda.
Sequenced investment roadmap
What to fund first, and the dependencies that set the order.
Maturity-anchored targets
Progress measured against TM Forum Autonomous Networks L0–L5.