A cell plant is a coating line, an assembly line and a very large charging warehouse, in that order. Understanding which stage dominates cost and which dominates risk explains most of what happens in gigafactory news, including why plants take so long to reach nameplate output.
Stage one: electrode manufacturing
Mixing. Active material, conductive additive and binder are dispersed into a slurry. The cathode side conventionally uses polyvinylidene fluoride binder dissolved in N-methyl-2-pyrrolidone, a solvent that is both expensive and subject to tightening regulatory attention, so it has to be recovered rather than released. The anode side generally uses a water-based system with carboxymethyl cellulose and styrene-butadiene rubber, which avoids the solvent problem entirely. Dispersion quality set here propagates through everything downstream, since agglomerates become coating defects.
Coating. The slurry is applied to metal foil, aluminium for the cathode and copper for the anode, usually through a slot die at high line speed. Coating weight uniformity is the control parameter that matters, because local variation in coating weight becomes local variation in current density in the finished cell, which becomes uneven ageing.
Drying. The solvent is evaporated in ovens that can run tens of metres long. This is one of the largest energy consumers in the plant, and for the cathode side it comes with a solvent recovery plant attached. Drying too fast migrates binder towards the surface and weakens adhesion, so line speed is bounded by physics as well as by oven length.
Calendering. The coated foil passes through rollers that compress it to a target thickness and porosity. Porosity is a design decision, not a tolerance: lower porosity raises energy density and raises tortuosity, which reduces rate capability. This single parameter is where the energy against power trade-off is physically set.
Slitting and vacuum drying. The web is cut to width and residual moisture is removed. Burrs from slitting are a defect route directly to internal short circuits, so edge quality is inspected closely.
Stage two: cell assembly
Separation into single sheets or continuous winding. Prismatic and pouch cells are usually notched and stacked; cylindrical cells are wound into a jelly roll. Stacking gives better space utilisation and thermal behaviour, winding is faster and mechanically simpler. Stacking speed has been a genuine industrial constraint and a focus of equipment development.
Tab welding. Ultrasonic or laser welding joins the current collector foils to the tabs. Weld quality determines both electrical resistance and mechanical durability under vibration, and it is a routine source of field failures when it is wrong.
Housing. The stack or roll goes into a hard case or a pouch, and the housing is sealed except for the fill port.
Electrolyte filling and wetting. Electrolyte is dosed under vacuum, then the cell rests so liquid can penetrate the pore structure of the electrodes and separator. Wetting is slow and cannot easily be rushed. Incomplete wetting leaves dry regions that never participate, showing up later as lower capacity and localised ageing.
Final sealing. The cell is closed. From here on, defects are internal and largely invisible.
Assembly happens in a dry room. Water reacts with the fluorinated electrolyte salt to produce hydrofluoric acid, which attacks both electrodes, so dew points are held far below ambient: typically around -30 to -40 °C for general assembly areas, and -40 °C or lower where electrolyte is handled, with some plants specifying -50 °C or below at the fill step. Dry rooms are expensive to construct and to run, and their cost scales with volume, which is one reason plants are built large.
Stage three: formation, ageing and grading
Formation. The finished cell is charged for the first time under a controlled profile. During this charge, electrolyte reduces at the anode surface and forms the solid electrolyte interphase. That film is what stops the electrolyte reducing indefinitely, so the cell only becomes a working battery at this step. The profile used, the temperature and any applied pressure influence the properties of the film and therefore the cell's whole life.
Degassing. Gas generated during formation is removed, and for pouch cells the gas pocket is cut off and the cell resealed.
Ageing. Cells rest, typically for days and in some processes for up to three weeks when wetting, formation and ageing are counted together (Wood et al., 2015), while open circuit voltage is monitored. A cell with an internal defect self-discharges faster than its peers, so this period is the screen that catches defects before shipment. Shortening it saves working capital and floor space and raises the escape rate, which is a genuine trade rather than an efficiency to be found.
End of line testing and grading. Capacity, resistance and self-discharge are measured, and cells are sorted so that a pack is built from closely matched cells. Matching quality directly affects how hard the BMS has to work over the pack's life.
Why formation and yield decide the economics
| Stage | Dominant cost driver | Typical bottleneck |
|---|---|---|
| Electrode manufacturing | Materials, drying energy, solvent recovery | Coating and drying line speed |
| Assembly | Equipment capital, dry room operation | Stacking rate, wetting time |
| Formation and ageing | Floor space, working capital, electricity | Cycler capacity and calendar days |
Materials dominate the bill of materials for a mature plant, but they are not what separates a competitive plant from a struggling one, because everyone buys in roughly the same market. Conversion cost, everything except materials, runs at roughly 20 to 30 per cent of total production cost, and Boston Consulting Group's 2025 analysis put it at about 13 dollars per kWh for an NMC pouch cell made in China against 17 in the United States and 22 in Germany. Two other items do the competitive work.
Formation and ageing tie up an enormous amount of capital in cells that are finished but not sellable, occupy a large share of the building, and need thousands of channels of cycling equipment. Cutting formation time is one of the highest-value process improvements available, which is why it attracts so much research effort.
Yield is the other. A cell scrapped after formation carries the full cost of every prior step, including its materials, its energy and its share of depreciation. Scrap rates of 15 to 30 per cent are common in the first years of a new plant, and Fraunhofer FFB's ramp-up analysis found reject rates still around 10 per cent after five years. The gap between a plant at high yield and a plant still climbing is larger than any plausible difference in materials procurement. This is why ramp timelines rather than nameplate capacity are the number to watch in any new factory announcement.
Contamination as a safety system
Metal particles introduced during production are a principal root cause of internal short circuits, and internal short circuits are a principal initiation route for thermal runaway. A particle of the wrong size and composition in the wrong place can penetrate the separator or dissolve and redeposit as a bridge.
Consequently, cleanliness in a cell plant is not housekeeping. Filtration, tool wear management, material handling discipline and inline inspection, increasingly with automated optical and X-ray methods, sit in the same category as any other safety control. Defect rates per million are already low, but a plant producing hundreds of millions of cells a year makes very small probabilities into certainties across a fleet, which is the reasoning behind propagation-resistant pack design.
What is changing
Dry electrode processing removes the solvent, the drying oven and the recovery plant, which are among the largest energy and capital items in the conventional route. Industrialisation at scale is underway rather than complete, and the process constraints differ enough that it is a plant redesign rather than an equipment swap.
Faster formation protocols, higher temperature or pressure-assisted formation, and better inline quality prediction all aim at the same target of reducing time spent in the formation and ageing area.
Inline inspection with machine learning models trained to predict end-of-line quality from earlier process signals is being deployed to catch defects sooner, when the scrap is cheaper.
Chemistry shifts change the process too. Sodium-ion can use aluminium current collectors on both sides, which alters materials handling. Solid-state programmes replace the wetting step with entirely different assembly problems, which is one of several reasons their manufacturing readiness lags their laboratory results.
Informational and educational content only. Not professional, financial, legal, or engineering advice.