Industry resources

Compiled 21 September 2026 and expanded 20 October 2026 by MTN Consulting.

Standards and industry initiatives

Bodies actively defining, testing, or certifying how AI gets built into telecom networks, drawn from what the sector's own vendors and operators are engaging with, not just the largest names.

Industry associations

Trade bodies that speak for operators and equipment suppliers in their home markets.

Regulators and policy bodies

Agencies and observatories whose rules on spectrum, competition, and AI set the limits operators work within.

Labor and workforce bodies

Unions, agencies, and statistics offices that bargain over or measure how AI changes telecom jobs.

Glossary

Terms that come up often in genAItelco's coverage of AI, network operations, and telecom workforce data.

Agentic AI

AI systems that take multi-step actions toward a goal with limited human prompting at each step, as distinct from a single-turn chatbot or classifier.

AI token economics

The unit economics of AI usage measured in tokens, the chunks of text a model processes, since many AI services are priced and metered by token volume.

AI-native network

A network designed from the ground up around AI-driven operation, as opposed to AI features added onto an existing architecture.

AI-RAN

TelecomTV's report on the launch of the AI-RAN Alliance (26 February 2024) quotes Nvidia's Ronny Vasishta describing AI-RAN as AI combined with the radio access network. The report lists three focus areas. AI for RAN uses AI tools to improve network performance and efficiency. AI and RAN runs AI and RAN workloads on shared infrastructure, with RAN processes given priority when the network is busy. AI on RAN delivers AI-enabled edge applications over 5G.

AIOps (artificial intelligence for IT operations)

TM Forum describes AIOps as AI-assisted or AI-driven automation in IT and network operations. In its training material, AIOps combines big data and machine learning to automate IT operations processes.

All-Photonics Network

A network architecture using light (optical signaling) end-to-end rather than converting to electrical signals at intermediate points, aimed at lower latency and energy use for AI-era traffic. Central to the ITU-T standard above.

Attrition

Merriam-Webster defines attrition, in its workforce sense, as a reduction in the number of employees, usually through resignation, retirement, or death.

Autonomous Network Levels (ANL) / ANLAV

TM Forum's maturity model for how much of network operation runs without human intervention, and the certification scheme that assesses an operator or vendor solution against it.

Business support system (BSS)

Wikipedia (page read 20 October 2026) describes BSS as the systems a telecom provider uses to run its business operations with customers. It lists four BSS processes. Product management covers developing, selling, and pricing products, offers, and bundles. Order management covers splitting sales orders into work orders, coordinating fulfillment, and tracking order status. Revenue management covers billing, charging, settlement, and related functions such as revenue assurance. Customer management covers a unified customer view, partner management, and self-service. The page's opening definition carries no citation.

Capital intensity

BCE Inc. defines it in its filings as capital expenditures divided by operating revenues. The term has no single standard: the Vernimmen finance glossary defines it as sales generated per unit of capital, the inverse of asset turnover.

Cloud-native network function (CNF)

A network function packaged as containers. The network functions virtualization group at the European Telecommunications Standards Institute treats three names as one thing: a CNF, a container-based virtual network function, and a cloud-native virtual network function.

EBITDA and EBIT

EBIT is earnings before interest and taxes, and EBITDA is earnings before interest, taxes, depreciation, and amortization. U.S. Securities and Exchange Commission staff say a company that presents either as a performance measure must reconcile it to net income under generally accepted accounting principles (GAAP). They add that a company must not use either label for a measure calculated differently.

Full-time equivalent (FTE)

Converts employed people into a comparable unit by comparing hours worked with a full-time schedule. Eurostat defines it as a unit that makes persons comparable despite different weekly hours. In one survey glossary, the U.S. Government Accountability Office divides regular hours worked by 2,080 compensable hours (40 hours times 52 weeks) to get one FTE. That figure is a convention, not a universal rule.

Hyperscaler

A cloud provider that supplies cloud services on a massive scale. Ofcom, the UK communications regulator, applies the term to Amazon Web Services, Microsoft, and Google in its 2022 study of the UK cloud market.

Inference vs. training compute

Training builds an AI model from data, a large one-time (or periodic) compute cost. Inference is running the trained model to serve users, a smaller but continuous, volume-driven cost.

Intent-based networking

The Internet Research Task Force's informational Request for Comments (RFC 9315) defines intent as a set of operational goals a network should meet and outcomes it is supposed to deliver, expressed declaratively. It defines an intent-based network as a network that can be managed using intent.

Labor cost

MTN Consulting defines labor cost as the reported annual figure that covers salaries, benefits, and any one-time charges, such as severance packages. The figure is totaled over 12 months. Labor cost per employee is that 12-month total divided by the average number of employees over the same 12 months.

Large language model (LLM)

A language model built at large scale. The U.S. National Institute of Standards and Technology describes large language models as predicting the next token or word in a text, which lets them generate and process natural language.

Model Context Protocol (MCP)

Anthropic announced MCP on 25 November 2024 as an open standard for connecting AI assistants to the systems where data lives, such as data sources and tools.

Multi-access edge computing (MEC)

The European Telecommunications Standards Institute describes MEC as cloud computing at the edge of the network. It gives application developers and content providers cloud-style computing and an IT service environment near the network edge.

Neocloud

Ciena, an optical networking vendor, defines a neocloud in its telecom glossary as a cloud provider focused mainly on accelerated computing for artificial intelligence (AI) and high-performance computing. Such providers often offer GPU-as-a-service, meaning access to graphics processing units (GPUs) delivered as a cloud service. Ciena adds that they build large GPU clusters and need high-capacity networks within, between, and beyond their data centers. This is a vendor definition, and the glossary page carries no date.

Network APIs and CAMARA

Network application programming interfaces (APIs) expose telecom network capabilities to developers through standard interfaces. CAMARA, an open source project in the Linux Foundation launched in 2022, defines, develops, and tests telco northbound APIs.

Network digital twin

A digital representation used in the context of networking whose physical counterpart is a data network, such as a provider or enterprise network. The Internet Research Task Force's network management group defines it in an Internet-Draft (draft 11, July 2025; a later draft is active), not a finished standard.

Network slicing

3GPP defines a network slice as a logical network that provides specific network capabilities and network characteristics. A communication service can rely on one slice or a combination of slices, and slices can share parts of the underlying network.

Operations support system (OSS)

Ciena's telecom glossary defines an OSS as a software platform, or a suite of applications, that supports the operational lifecycle of communications networks and services. Typical functions include service provisioning, network inventory, fault management, performance monitoring, service assurance, and automation. This is a vendor definition, and the glossary page carries no date.

RAN / O-RAN

Radio Access Network, the part of a mobile network connecting devices to the core. "Open RAN" (O-RAN) disaggregates it into interoperable components from different vendors instead of one vendor's closed stack.

RAN Intelligent Controller (RIC), xApps, and rApps

A RIC hosts software that tunes the radio network. A paper by Hoffmann and co-authors (arXiv, November 2023) puts the near-real-time RIC on a control loop longer than 10 milliseconds and shorter than 1 second. It runs xApps, which third parties can develop. The same paper puts the non-real-time RIC above 1 second, where it handles policy optimization. It also supplies policies, enrichment information, and machine learning (ML) models to the near-real-time RIC. The O-RAN Software Community documentation describes rApps as applications hosted on the non-real-time RIC.

Reskilling, retraining, and upskilling

The International Labour Organization's glossary has no entry for reskilling. It defines upskilling as training that supplements and updates existing knowledge, skills, or competencies, and retraining as supplementary training that enables individuals to acquire new skills. Cedefop, the European Union's vocational training agency, defines retraining as training that gives access to a new occupation, according to a glossary kept by UNESCO-UNEVOC.

Retrieval-augmented generation (RAG)

Combines a language model's built-in memory with a retrieval step. The model looks up passages in an external document index and uses them when it generates an answer. The term comes from a 2020 academic paper by Lewis and co-authors.

Separations (quits, layoffs and discharges)

The U.S. Bureau of Labor Statistics (BLS) Job Openings and Labor Turnover Survey sums three types of separation into total separations. Quits are voluntary departures. Layoffs and discharges are involuntary separations that the employer initiates. Other separations cover retirements, deaths, disability, and transfers to another site of the same firm.

Sovereign AI

AI infrastructure and models a country builds or controls domestically, framed as a matter of national capability rather than relying on foreign providers.

Worker Adjustment and Retraining Notification (WARN) Act

According to the U.S. Department of Labor, this law requires covered employers, generally those with 100 or more employees, to give at least 60 calendar days of written notice. The notice applies before a plant closing or mass layoff.

Zero-touch network and Service Management (ZSM)

A framework from the European Telecommunications Standards Institute (ETSI) for automating network and service operations. Its stated target is autonomous networks capable of self-configuration, self-monitoring, self-healing, and self-optimization, on a modular, service-based architecture.

Industry events

The events calendar now has its own page: 124 telco AI events worldwide, filterable by region and organizer type, covering industry conferences, government and standards meetings, and investor events.

Data and trackers

The Chinese AI at telcos tracker lists the telcos outside Mainland China that partner with one of China's AI model developers, with a source link for every row.

Note: this list is a starting point, not exhaustive. Bodies with AI or automation responsibility exist beyond what's above; tell us who we're missing through the contact page.