The humanoid robot has become one of the more entertaining symbols of the “Physical AI” boom. China’s recent World Humanoid Robot Games provided plenty of material: robots competing in scenarios intended to demonstrate how embodied AI might move into factories, services and eventually homes. Highs included a 100m sprint where a robot beat Usain Bolt’s world record… and lows included it crashing headlong into a barrier shortly thereafter.
There is serious money behind the spectacle. China’s government is investing heavily in humanoid robotics – for example, building training datasets of movements. Meanwhile companies such as Unitree are moving from research demonstrations towards commercial products. The timeline for any mass adoption is unclear though, to say the least.
Unitree founder and CEO Wang Xingxing, one of the most bullish people you’ll find on the topic, recently said the industry could have its “ChatGPT moment” in two to ten years. He defined that moment as the time when a robot would be able to enter an unfamiliar home or workplace and complete around 80% of tasks through voice or text instructions.
So perhaps we shouldn’t build a business case around humanoids arriving en masse to consumers any time soon. However, while less of a dopamine hit there is huge potential value lying elsewhere.
Stocking Up
The International Federation of Robotics puts the global operational stock at around 4.7 million industrial robots, with more than 500,000 new installations in 2024. Then there are burgeoning fleets of autonomous mobile robots in warehouses, delivery robots, agricultural machinery, drones and autonomous vehicles. Behind all of these is the enormous installed base of IoT devices.
Ericsson currently estimates that there were around 4.5 billion cellular IoT connections at the end of 2025. What’s more, Ericsson’s IoT forecast points out that many of these devices are already becoming more capable, with devices enabled by broadband and critical IoT representing a growing proportion of the market.
These devices don’t need the next generation to operate – they’re already doing that. But we might think about 6G as an opportunity to wrap more value or intelligence around the device.
A security camera doesn’t necessarily need more connectivity to supply a video feed, but the company might save money if intelligence at the edge can intercept and interpret the feed coming through to raise alerts, or only record particular kinds of activity (it might save congestion in the network too – double bonus). A drone doesn’t need a network to fly, but it might benefit from network-provided information about other aircraft.
This is where 6G value propositions start to look very different from previous generations. While they might leverage a broad infrastructure, the offering is going to look very different to a security firm compared to a drone delivery service. The capability to be that granular may be one of the most important differentiators from the mass-market-first approach of previous generations.
Droning on
Drones offer a particularly useful test because they combine autonomy, mobility and a requirement to operate over wide areas. 3GPP has already been working on cellular support for unmanned aerial systems. Its specifications envisage network-supported identification, tracking and traffic-management functions, as well as direct UAV-to-UAV communication for collision avoidance. ETSI’s 3GPP UAS specification explicitly describes exchanging information such as UAV identity, position, speed and flight-route information.
The drone doesn’t need 6G, or 5G for that matter, to remain airborne. However, if it opens up business opportunities for flying beyond line-of-sight in jurisdictions where it’s currently considered unsafe, then it makes a business case for dozens of different companies.
But… if 5G can provide connectivity, identification and some degree of traffic management, what exactly does 6G add?
Possibly Integrated Sensing and Communications? Samsung and LG Uplus are researching the use of cellular infrastructure to sense objects and estimate their distance, speed and direction. But a possible use case for 6G isn’t a business case.
(This is a bugbear for some of TelcoForge team. A researcher once identified over 200 uses for a piece of A4 paper, but we doubt anybody has restocked their stationary cupboard because they might need to get a spider out of the bath).
Taxonomy of Competition
Inside a factory, robots could potentially benefit from extremely predictable connectivity, low latency, local edge computing and sensing in order to collaborate between themselves and humans more safely. But that environment also has obvious alternatives: private 5G, industrial Ethernet, Wi-Fi or some combination of them, quite apart from any sensors already on the machines.
A port or airport is more interesting because machines move around a relatively large controlled environment, but the operator can still potentially deploy dedicated infrastructure.
Agriculture, drone operation or any changing environment such as logging, infrastructure repairs or public safety might be more viable markets for national-scale players. Vehicles, too, might benefit – both ‘autonomous’ ones and those with drivers.
The more a machine is mobile, geographically distributed, safety-critical and dependent on information from its surroundings, the stronger the argument becomes for a public network. That still doesn’t have to be 6G, though.
So does that mean giving up on this upcoming generation? Not necessarily, but it may mean changing our collective perspective.
Commercial Model Maker
The mobile industry has spent decades thinking in terms of coverage, capacity, spectrum efficiency and congestion. The commercial model has largely followed the engineering: build a network, connect subscribers and sell connectivity. 6G’s proposed capabilities point towards something different this time around.
The emerging 6G technology framework includes AI-native networks, integrated sensing and communications, hyper-reliable low-latency communication and ubiquitous connectivity alongside more familiar connectivity requirements. The ITU describes “ubiquitous intelligence” as one of the overarching principles for the next generation. Outside the direct 6G conversation, but certainly complementary to it, is talk of building a much more responsive, autonomous and AI-powered network capable of much smarter service enablement.
So we should probably be asking less about what feeds and speeds we provide at what cost per bit, but what capability can the network provide that somebody values?
- Perhaps an operator rents edge compute for milliseconds at a time when a machine needs more brain-power than it’s been built with. That might lengthen the time in service for that device, saving the owner money on buying a new one while getting the latest software upgrades.
- Perhaps it offers an identity, sensing and coordination platform for drones operated by different companies, reducing their risk and lowering insurance costs.
- Perhaps it becomes a resource orchestration platform across physical, cloud and AI elements, finding the most efficient ways for companies to coordinate these things in real time.
- Perhaps it simply guarantees a particular level of performance to machines moving through its coverage area.
Those are very different offerings from a mobile subscription, they depend on a much more intelligent and flexible network, and they will have some very uncomfortable ramifications. The competition is going to be very different; proposition creation and management would be almost unrecognisable; and customer management, billing and engagement operations would need to change considerably.
Question Time
TelcoForge has argued before that “build it and they will come” is unlikely to be a viable approach to the next generation of networks. The industry’s had its fingers burnt with 5G, where the biggest financial argument in favour seems to be more about the cost savings seen at the back-end rather than new revenues. That’s an argument in hindsight, though – 2015 was full of robot surgeons and connected cars.
More recently, we have argued that 6G’s biggest story may be about control rather than speed: who controls the information, intelligence and interactions taking place across increasingly autonomous systems. The growing population of autonomous machines will offer a useful way to test that proposition.
There are really three key questions around this.
- Whether the value proposition for the network can move from connecting things to augmenting them;
- Whether the resulting capability is valuable enough for somebody to pay for.
- If so, whether they’ll be paying the network operator.
These questions are some among many we hope to explore further during TelcoForge’s upcoming 6G Value Season, where we’ll be running research, public and private debates asking some of the harder questions about what 6G may be worth, to whom, and what work we need to do outside of technical standards to get there. We’ll bring everything together in our first 6G Value Report. If you’d like to be involved, please drop us a note.
Photo by Emilipothèse on Unsplash
