The UK's mobile networks are not prepared for the coming wave of AI traffic, according to one of the country's most senior network executives. Andrea Donà, networks director at VodafoneThree, said the quality of the country's mobile infrastructure is below the standard he expects. "I am embarrassed today at the level of quality of our network," he said. His warning comes as mobile operators confront an era in which AI-powered applications, autonomous agents, and machine-to-machine communication will place far greater demands on wireless networks.
Donà described the experience of trying to use a phone in a busy stadium, where the network slows to a crawl because too many people are competing for capacity. He said AI traffic would make that kind of degradation permanent. Instead of occasional congestion during a match or concert, networks would face constant pressure from software agents exchanging data in the background, devices uploading telemetry, and users relying on AI assistants that need cloud-level processing power.
UK coverage lags behind Europe and the G7
The starting point is weak. Independent analysis of mobile network performance places the UK below every member state of the European Union and every other G7 country. Globally, the UK ranks 57th for network performance. The findings, based on millions of real-world measurements, highlight a gap that has persisted for years despite repeated government promises to improve digital infrastructure.
This is not a simple question of investment. VodafoneThree has committed £11 billion to upgrading its mast estate and to modernising the technology attached to them. The company is also reducing its mast count by 30 percent, a move that suggests consolidation rather than expansion. Yet the quality problem remains, and Donà argues that the main obstacle is not money but planning law.
Almost none of the work requires a new site. Around 96 percent of the planned upgrades are on existing masts and structures. However, more than half of those upgrades still need planning permission. That means a project intended to add 5G capacity, improve latency, or deploy AI-aware network management can be delayed for months while local authorities review an application.
Planning reform as the cheapest growth policy
"Planning reform is the lowest possible cost growth policy, it doesn't require any government money," Donà said. The argument is straightforward: if the government wants better connectivity, it should make it easier for operators to upgrade the infrastructure they already own. The cost of the work is being met by the private sector. What operators need from the state is permission to proceed at speed.
The UK is not alone in wrestling with this problem. The European Union has already legislated on the matter. In 2024, the EU adopted the Gigabit Infrastructure Act, which sets deadlines for national authorities to decide permit applications for very high capacity networks. Under the act, authorities have four months to make a decision. If they miss the deadline, the permit is deemed granted. This has been in force since November of last year. Member states may opt out of the automatic approval mechanism, but only if they offer an alternative remedy such as compensation or a right to sue the authority responsible.
The EU is also going further. The proposed Digital Networks Act includes measures that would grant indefinite spectrum licences, introduce use-it-or-share-it rules, and simplify deployment conditions in exchange for tougher coverage obligations. These proposals are designed to accelerate the rollout of networks while making sure that operators meet measurable performance targets. The UK is bound by none of this legislation, and its own regulatory framework has not kept pace.
AI traffic demands new network architecture
The challenge posed by AI traffic is not simply one of volume. AI applications often need low latency because a user interacting with a voice assistant or an autonomous vehicle expects near-real-time responses. This requires compute resources to be distributed closer to the edge of the network, not concentrated in distant data centres. Mobile operators are therefore investing in edge computing, network slicing, and advanced antenna systems that can dynamically direct capacity to where it is needed.
At the same time, AI can be used to optimise the network itself. Machine learning models can predict traffic patterns, optimise handovers between cells, and detect faults before they affect users. But these benefits depend on having a network that is dense enough to generate the required data and flexible enough to act on the insights. A network that cannot handle current demand will struggle to support the control systems needed for its own improvement.
There is also a security dimension. AI-driven traffic is harder to distinguish from legitimate use. Networks will need deeper packet inspection and behavioural analysis to identify botnet activity, automated scrapers, and malicious actors. This adds processing overhead at a time when capacity is already stretched.
Critical infrastructure status and financial relief
Mobile operators also point out an inconsistency in how the state treats critical infrastructure. Data centres receive reliefs on business rates and energy costs because they are considered essential to the digital economy. Mobile networks, which carry the traffic that connects users to those data centres, do not receive the same treatment. Operators argue that this creates a distorted market in which the most visible parts of the digital economy are protected while the underlying transmission layer is left to fend for itself.
VodafoneThree's £11 billion commitment is substantial, but the industry estimates that the total cost of meeting the UK's coverage obligations will be much higher. Without a supportive regulatory environment, operators may find it difficult to justify further investment, especially in rural areas where the return on capital is lower.
Satellite 5G as an alternative path
While the UK debates planning reform, other players are attacking the connectivity gap from a different direction. A Spanish startup has raised €150 million to develop a network of satellites that can beam 5G signals directly to ordinary phones. Satellite-based connectivity has the advantage of bypassing local planning rules entirely, since the infrastructure is in orbit. This approach could provide coverage in remote areas and act as a backup for terrestrial networks during emergencies.
The startup's plan reflects a broader trend toward direct-to-device satellite services. Established satellite operators and smartphone manufacturers are exploring ways to integrate satellite connectivity into consumer devices. If these systems become commercially viable, they could change the competitive dynamics of the mobile market. However, terrestrial networks will remain essential in urban areas, where the density of demand cannot be met by satellites alone.
The shape of the problem
The debate over network readiness is ultimately about the relationship between the physical world and the digital economy. AI models are typically built and trained in facilities that enjoy planning status, tax relief, and priority access to energy. But the value of those models depends on the ability of users to reach them through a reliable network. If the transmission layer is weak, the economic potential of the entire ecosystem is diminished.
Donà's intervention is notable because it comes from an executive of the UK's largest mobile operator, not from an industry lobby group. His message is that the private sector is willing to spend, but the state must remove the obstacles that slow down deployment.
The UK's position in international rankings may continue to slide if the gap between planning rules and technological reality is not addressed. The EU has already adopted a framework that tackles the problem directly. The UK has yet to match it. Operators warn that every month of delay extends the period in which UK users experience the kind of network degradation that would be unacceptable in other advanced economies.
The comparison with a stadium is useful because stadium congestion arises from a temporary mismatch between demand and capacity. AI traffic threatens to make that mismatch permanent. Mobile networks will need not only more capacity but also smarter ways of managing it. The irony is that the technology to achieve this already exists. What is missing is the permission to put it in place.
Britain protects the building where the model runs and not the network that has to reach it.