In short: The UK is building its first generation of battery gigafactories — Agratas in Bridgwater, AESC in Sunderland, the West Midlands Gigafactory at Coventry. These are not ordinary factories. They are enormous single buildings where the whole commercial case turns on yield, intralogistics never stops, and the dry rooms tolerate no compromise. The wireless layer is part of how the cells get made, not a convenience laid over the top — which is why it belongs in the design from the first drawing.
Key Takeaways
- A gigafactory is one building the size of a small town — Agratas at Bridgwater alone runs to tens of football pitches under one roof, and a single Wi-Fi-per-zone patchwork cannot hand a moving AGV across that span without dropping it.
- Yield is the entire business, and yield is a data problem — electrode coating, calendering and formation each throw off continuous process data, and catching a drift early is the difference between a profitable ramp and scrapped cells.
- The dry room punishes the wrong network — battery assembly happens in ultra-low-humidity rooms where cabling and access-point sprawl are a contamination and maintenance liability, and a single private 5G layer cuts the in-room hardware right down.
In a nutshell
Britain finally has gigafactories to talk about
For most of the last decade the UK conversation about battery manufacturing was a conversation about absence. The cars were built here; the cells were not. That is now changing, and changing fast enough to be worth taking seriously. Agratas, Tata's battery business, is building a roughly 40GWh gigafactory at Bridgwater in Somerset — one of the largest single industrial investments the UK has seen, supplying Jaguar Land Rover and others. AESC's plant in Sunderland, alongside Nissan, is expanding to feed the next generation of electric vehicles built on Wearside. The West Midlands Gigafactory at Coventry Airport has planning consent and is hunting for its anchor manufacturer. Between them, and the supply-chain plants forming around them, they represent the start of a genuine UK cell-manufacturing base.
We have written before about UK reshoring and the case for designing factories wireless-first. Gigafactories are a sharper, more demanding version of that argument, and they deserve their own treatment. A gigafactory is not a big version of a normal factory the way a hypermarket is a big version of a corner shop. The scale changes the physics of the connectivity problem, the process changes what the network has to carry, and the economics of the ramp change how much a connectivity failure actually costs. This post sets out why we think the wireless layer belongs in the gigafactory design from the outset, and where the case is genuinely different from the general factory-floor story.
Scale that breaks the obvious network
Start with the building, because the sheer size of a gigafactory defeats the instinctive answer before any process detail comes into play. Agratas's Bridgwater site covers an enormous footprint — the production halls alone run to tens of football pitches under a single roof, with cell manufacturing, module assembly and pack lines arranged across one continuous floorplate. AESC and the other UK plants are in the same class of scale. These are some of the largest single enclosed industrial spaces in the country.
Now consider what moves across that floor. A gigafactory runs on intralogistics: automated guided vehicles and autonomous mobile robots shuttling electrode rolls, components and finished modules continuously between processes that may be hundreds of metres apart. Those vehicles cannot stop, and they cannot afford to lose their connection mid-route. The defining weakness of Wi-Fi in a building this size is handover — as a vehicle crosses from one access point's cell into the next, the roam is never quite seamless, and across a floorplate that needs hundreds of access points to cover, the cumulative probability of a dropped connection somewhere on a route is high. An AMR that pauses because it lost signal between zones is not a minor annoyance at this scale; it is a stoppage propagating through a just-in-time material flow.
Private 5G was, in effect, designed for exactly this. A single network with a handful of radio units covers the floorplate as one continuous cell, and the handover between them is managed at the network layer rather than left to the client device to scramble for. A vehicle crossing the whole building stays on one connection from end to end. For intralogistics across a gigafactory-scale floor, that is not a marginal improvement over Wi-Fi; it is the difference between a network that works and one that needs constant nursing.
Yield is the whole game
Here is the fact that reframes everything about a battery plant: the commercial success of a gigafactory is almost entirely a question of yield, and yield, in cell manufacturing, is a data problem.
When a new gigafactory comes online it does not start making good cells at full rate on day one. It ramps — a long, expensive climb during which early yield can be low and a painful share of the cells produced are scrapped. The ramp is where gigafactory business cases are won or lost; the operators who climb the yield curve fastest are the ones who make money, and the ones who stall on it burn cash on scrapped material and idle capacity. Improving yield by even a few points across a 40GWh plant is worth an enormous amount.
And yield is made and lost in the process data. Electrode coating has to lay slurry onto foil to micron tolerances; a drift in thickness or a coating defect propagates into every cell made downstream from it. Calendering presses the electrode to a precise density. Formation — the first controlled charge and discharge that brings a cell to life — generates a rich time-series for every single cell and is one of the most data-heavy steps in all of manufacturing. Across these steps a gigafactory throws off a continuous torrent of process and inspection data, and the entire art of the ramp is catching a drift, a defect or an out-of-spec trend early enough to correct it before it has scrapped a batch.
That only works if the data moves. High-resolution inspection cameras on the coating line, the dense instrumentation on the formation racks, the quality sensors threaded through every process — all of them need to stream into the edge analytics and the manufacturing-execution system in real time. A network that cannot carry that volume, or that introduces lag and loss, blinds the operator to exactly the early signals the ramp depends on. The connectivity layer in a gigafactory is not carrying email; it is carrying the data the yield curve is climbed with. That is a different order of dependency, and it is why we argue the network belongs in the process design, not bolted on after the lines are commissioned.
The dry room changes the hardware calculus
Battery cell assembly has a constraint that ordinary manufacturing does not: large parts of it happen inside dry rooms, ultra-low-humidity environments held at dew points far below anything a normal factory contends with, because moisture is poison to lithium cell chemistry. Building and maintaining dry-room conditions is expensive, and everything inside a dry room is subject to tight control — every piece of equipment, every cable penetration, every maintenance intervention is a cost and a contamination risk.
This is where the wireless-first argument stops being about convenience and becomes about the room itself. A Wi-Fi-per-zone approach inside a dry room means a proliferation of access points, each one a powered device generating heat, each one cabled back through controlled penetrations, each one an item that eventually needs maintenance access into an environment you would much rather not open up. Every one of those is friction against the dry room's whole purpose. A single private 5G layer, with radio coverage planned for the space and a minimal in-room hardware footprint, cuts that sprawl down substantially. Fewer devices in the controlled environment, fewer penetrations, fewer maintenance entries. In a setting where the cost of the environment is itself a major line item, reducing the in-room network hardware is a real and quantifiable benefit, not a tidiness preference.
Designing it in versus bolting it on
The strongest version of our argument is also the simplest: gigafactories are being built now, on greenfield sites, from a blank sheet. The cost of designing the wireless layer in from the start is a fraction of the cost of retrofitting it into a running plant — and unlike most of the UK's existing industrial estate, these buildings still have the blank sheet available.
Designing in means the radio plan is part of the building plan, so coverage follows the process layout instead of fighting it. It means the spectrum is secured early — straightforward in the UK through Ofcom's Shared Access licensing, but still a step that wants doing before commissioning, not during. It means the AGV fleet, the inspection systems and the MES are specified against a network that is known to support them, rather than discovering during the ramp that the connectivity cannot carry the load. And it means the dry-room hardware footprint is minimised by design rather than apologised for later. The ramp is hard enough on its own terms; starting it while also firefighting a connectivity layer that was treated as an afterthought is a self-inflicted wound the greenfield timing makes entirely avoidable.
Where the case is weaker
In fairness, the gigafactory connectivity case is not universal, and it is worth marking the limits. The very biggest operators may run a hybrid — private 5G for the mobile, mission-critical and wide-area traffic, with Wi-Fi 6 retained for fixed, dense, low-stakes connections such as office areas and handheld devices in non-critical zones. That mixed estate is often the right answer, and we have made the Wi-Fi-versus-5G argument at length elsewhere; the point is that the two are complementary, not that one wins outright. Private 5G also carries genuine up-front capital and a spectrum and integration step that a smaller component plant feeding the gigafactory may not need to take on day one. The case is at its strongest precisely where the building is largest, the intralogistics densest, the yield economics most punishing and the dry-room footprint biggest — which is to say, at the cell plants themselves rather than every supplier around them.
The conclusion the timing forces
Britain spent a decade worrying it had missed the battery-manufacturing window. It now has real gigafactories rising at Bridgwater, Sunderland and Coventry, with a supply chain forming around them. These plants will live or die on how fast they climb the yield curve, and that climb runs on data moving reliably across an enormous floor, through dense process instrumentation, in and out of dry rooms that punish unnecessary hardware. The wireless layer is part of how those cells get made.
The encouraging part is that the timing is, for once, on the right side. These are new buildings on blank sheets, where the network can be designed in for a fraction of what a retrofit would cost. We would simply make the case plainly: in a gigafactory the connectivity is process infrastructure, not IT overhead, and the cheapest moment to get it right is before the first cell ships.