Telcos Are Done With 'AI Experimentation'—They're Building AICO Empires Now

Remember when telecom companies were just the pipes? The invisible utilities that moved your calls, texts, and cat videos from Point A to Point B? That era is ending. Fast. According to NVIDIA’s latest “State of AI in Telecommunications” survey, a seismic shift is underway. The industry isn’t just using AI anymore; it’s fundamentally rebuilding itself around intelligence. The new mantra? Become an AICO—an AI Infrastructure Company—or risk becoming irrelevant.

The Great Telco Rebrand: From Pipe Dream to AICO

Sebastian Barros, Managing Director at telecom provider Circles, puts it bluntly: the goal is moving intelligence across the infrastructure, not just shuttling bits through it. This isn’t about slapping a chatbot onto a customer portal. It’s about transforming the network itself into a thinking, self-optimizing entity. The survey of 1,000 professionals (with 250 telco operators) reveals an industry at a crossroads, and the path chosen is unmistakably AI-native.

The Numbers Don't Lie: AI Is No Longer a 'Maybe'

Forget speculative promises. The data shows AI is delivering concrete, bottoms-line results today. The commitment is no longer theoretical; it’s budgetary and strategic.

• 90% of telecom operator respondents say AI is already boosting annual revenue AND slashing operational costs. That’s a dual impact few technologies achieve.
• A staggering 89% plan to increase AI spending in 2026, up from 65% the year prior. Over a third (35%) expect budget jumps exceeding 10%.
• On the productivity front, 100% of all respondents reported gains. A serious chunk (26%) cited "major to significant" improvements—more tasks, higher quality, less time.
• Open source is the secret weapon: 89% say open-source models and software are critical to their strategy, signaling a move away from monolithic, vendor-locked systems.

Top ROI Drivers: What's Getting the Checkbook Open?

When asked where AI delivers the sharpest return, telecom leaders pointed to three core areas:

1. Autonomous Networks (50%): The #1 ROI use case. Systems that self-configure, self-heal, and self-optimize.
2. Improved Customer Service (41%): From smarter chatbots to predictive issue resolution.
3. Internal Process Optimization (33%): Streamlining back-office and field operations.

Generative AI Was the Appetizer. Agentic AI Is the Main Course.

The survey highlights a clear evolution in AI deployment stages. Generative AI (GenAI) is having its moment—60% of telcos are using or assessing it, up from 49% in 2024. Its power lies in "fast productivity gains" across offices and networks.

But the real strategic shift is toward Agentic AI. As analyst Chetan Sharma notes, network automation has now overtaken customer experience as the top investment focus. Why? Because Agentic AI represents autonomous AI agents that don't just generate content—they act. They can make decisions across the network, IT systems, and customer journeys without human delay. This is what unlocks "structural ROI" and accelerates the path to full autonomy by coordinating real-time decisions across domains.

The Road to Level 5 Autonomy (and Why 6G Might Wait)

The industry-standard TM Forum autonomy scale runs from Level 1 (manual) to Level 5 (fully autonomous). The majority (88%) of organizations currently sit between Levels 1-3. But here’s the kicker: 77% of respondents expect AI-native networks to launch before the deployment of 6G.

What does an AI-native network look like? It’s built from the ground up with AI cores, not bolted on as an afterthought. The top investments driving this are:

• Using AI to enhance spectral efficiency (get more from existing airwaves).
• Supercharging RAN (Radio Access Network) performance for edge AI applications.
• Accelerating 6G R&D itself through simulation and intelligent modeling.

What's Missing? The Human Element (and Why It Matters)

Curiously, the report spends no ink on workforce reductions. The narrative isn't about replacing engineers; it's about augmenting them. The 100% productivity gain figure suggests AI is acting as a force multiplier, eliminating soul-crushing reactive work like fault prediction and configuration drift correction. The goal is to free human talent for higher-order problem-solving.

Also notably absent are specific vendor rivalries or product names (beyond NVIDIA's role as survey source). The focus is on categories and concepts—GenAI, Agentic AI, AI-native—indicating a new, platform-agnostic operational philosophy.

The New Telecom Landscape: Infrastructure as an Intelligent Service

The path forward is clear. Telcos are aggressively transitioning from connectivity providers to AI infrastructure companies. The network is becoming the platform. Agentic AI is the orchestrator that turns this vision into an operational reality, moving beyond point solutions to end-to-end autonomy.

The companies that master this shift won't just sell data plans. They’ll sell intelligent, self-healing, ultra-efficient infrastructure as a service. The 5G rollout was about speed and latency. The next decade will be defined by autonomy. And based on this survey, the race is already on, with budgets open and the endgame firmly in sight.