6G may well be the green G

6G will not be about speed. It will be about intelligence, cost, and carbon

For three decades, every new generation of mobile technology has been sold on the same promise. Faster speeds, lower latency, more capacity. It worked for 3G, it worked for 4G, and it worked again for 5G, even if many operators are still waiting for the revenues to catch up.

6G will be different. Not because the industry has suddenly become less obsessed with performance, but because the economics and the climate maths no longer allow a simple throughput arms race. The defining feature of 6G will not be peak data rates but the embedding of artificial intelligence into the fabric of the network itself, and the use of that intelligence to drive experience and efficiency at a scale the industry has never achieved before.

That shift has three big consequences:

• First, networks become AI native rather than just AI assisted.

• Second, energy efficiency becomes a core design goal rather than an afterthought.

• Third, sustainability moves from corporate slideware into the actual control loops of the network.

From smart tools to an intelligent network

Today AI in telecom mostly lives on the edges. A vendor sells an energy saving feature. An operator deploys a traffic prediction model. A data scientist builds a fault detection system. Useful, but still bolted on.

In the 6G era, AI will become the nervous system of the network, running from the physical layer up to service orchestration and into the OSS/BSS. In the 5G core, functions like the Network Data Analytics Function (NWDAF) already use machine learning to analyse traffic and optimise performance. In 6G this concept is taken further, with analytics, training and decision making becoming intrinsic to how the network behaves.

That matters because a network that understands itself can make trade offs that humans and static software never could. It can decide whether a few extra milliseconds of latency are worth a large energy saving. It can route traffic not just to the least congested node, but to the one running on the cleanest power at that moment. It can scale down entire radio sites when demand drops, and bring them back when it rises.

This is not theoretical. Telefónica already runs more than 1,400 energy efficiency projects across its networks, including radio deep sleep modes that shut down parts of base stations when traffic is low, delivering energy savings of up to 26 percent in those periods.

FarEasTone is using AI to analyse network data and cut both power use and carbon emissions without hurting service quality What 6G does is make this kind of optimisation universal, continuous, and automated.

New outcomes: why efficiency beats speed

If the industry just keeps scaling capacity in the old way, energy use and emissions will soar. That is not compatible with net zero targets or with the reality that energy already represents 20 to 40 percent of network operating costs for many operators. The winners will not be the networks that deliver the highest peak speeds, but the ones that can deliver a given level of service using the least energy.

AI is the only realistic way to do that. Machine Learning can predict traffic, optimise spectrum use, and dynamically switch off hardware that is not needed. In data centres, AI driven cooling systems can cut water use and electricity consumption dramatically. Microsoft has already shown water savings of up to 90 percent in some facilities using AI controlled cooling.

None of this is new in isolation, what is new is the idea that these mechanisms become mandatory design principles in 6G.

Sustainability as a control variable

6G introduces the idea that sustainability is not just measured, but actively managed around three concepts, observability, choice, and circular economy.

• Observability means the network knows its own environmental impact. It can measure the energy and carbon cost of training a model, running an analytics function, or serving a video stream.

• Choice means the network can make trade offs. A core network function might choose a slightly less accurate model because it uses half the energy. A radio scheduler might accept a small drop in spectral efficiency to avoid switching on a fossil fuel powered generator.

• Circular economy means the hardware itself is managed more intelligently, with AI helping to extend lifetimes, predict failures, and reduce electronic waste.

This is where the intelligence layer becomes more important than raw bandwidth. A self-driving network that can interpret high level intent, such as deliver this service while minimising carbon emissions, is far more valuable than one that can just push more bits through the air.

A more uncomfortable truth

There is a catch, of course. AI is itself energy hungry. Training and running large models consumes vast amounts of power, and in many cases inference now accounts for around 80 percent of an AI system’s operational footprint.

That is why 6G also forces the industry to get serious about efficient AI. Techniques like model pruning, quantisation, and knowledge distillation can reduce the computing load of a model by two thirds or more while keeping most of its performance.

Edge computing and federated learning reduce the need to move data across the network. Specialised chips like NPUs and TPUs do more work per watt than general CPUs.

In other words, 6G is not just about smarter networks. It is about smarter AI.

The real upgrade

None of this will make for glossy marketing. You cannot put AI native sustainability on a speed test. But it is exactly what the industry needs. 6G will be sold as a new generation of connectivity. In reality, it is a new generation of economics. Lower energy per bit, lower cost per service, and lower carbon per customer. That may not sound as exciting as a tenfold speed increase. It is far more likely to keep operators in business.