The Emerging Industrial Stack

9 layers · 14 constraints mapped

Materials, Energy and Components hold all 14 Bottleneck Atlas constraints. They are also the slowest layers to build, so the whole stack moves at their pace.

The nine-layer industrial stack. All 14 constraints in the Bottleneck Atlas sit in the bottom three layers: Materials (6), Energy (3) and Components (5). None sit in Compute, Processes, Infrastructure, Nodes, Networks or Systems. Time to add capacity runs from about 16 years for a new mine to weeks for a software release. LAYERATLAS CONSTRAINTSTIME TO ADD CAPACITY (LOG) 9Systems none weeks 8Networks none months 7Nodes none 3-4 yr 6Infrastructure none 2-3 yr 5Processes none ~1 yr 4Compute none 3-5 yr 3Components 2-3 yr 2Energy ~5 yr 1Materials ~16 yr Binding Tight Concentrated Surplus Not at scale
Layers
9
Constraints
14
In layers 1-3
14 of 14
Mine lead time
~16 yr
Hub
Supply Chain

Finding

Every constraint sits in the bottom three layers

Map the 14 constraints from the Bottleneck Atlas onto the stack and all of them land in Materials, Energy or Components. The binding ones (transformers, interconnection, rare earth magnets, strain-wave reducers) and the surpluses (lithium, cells, polysilicon, SiC) live in the same three layers.

Those are also the slowest layers to build. A new mine averages about 16 years from discovery to output. A grid connection waits about five years in the queue. A cell plant takes two to three. At the top of the stack, a fleet adds vehicles in months and software ships in weeks.

So the stack runs at the speed of its bottom layers. Capital and attention go to compute, autonomy and services, where change is fast; the pace is set by mines, steel, transformers and interconnection, where it is not. When the bottom layers overbuild, as China did with lithium, cells and polysilicon, the glut also starts there and moves up as falling prices.

The nine layers

From materials to systems

Each layer depends on the one below it. The table lists what sits in each, which Atlas constraints fall there, and a typical time to add capacity.

The nine stack layers: examples, constraints and build time
LayerExamplesAtlas constraintsTime to add capacity
9 · Systems
Capabilities delivered to users		Robotaxi service, electrified freight, AI inference, DER managementNone in the AtlasSoftware release: weeks
8 · Networks
Nodes coordinated as one		Fleet dispatch, virtual power plants, plant telemetry, V2GNone in the AtlasAdd units to a fleet: months
7 · Nodes
Individual machines that do work		EVs, autonomous trucks, humanoids, BESS units, chargers, serversNone in the AtlasNew vehicle program: 3-4 yr
6 · Infrastructure
Fixed sites		Gigafactories, fabs, depots, data centers, substations, portsNone in the AtlasGigafactory: 2-3 yr
5 · Processes
Methods that turn inputs into outputs		Electrode coating, formation, lithography, gigacasting, EAF steelNone in the AtlasNew line ramp: ~1 yr
4 · Compute
Embedded control and inference		Autonomy SoCs, AI accelerators, MCUs, PLCs, edge processorsNone in the AtlasNew fab: 3-5 yr
3 · Components
Engineered building blocks		Cells, chips, motors, inverters, reducers, LiDAR, sensorsStrain-wave reducers Mature-node chips SiC wafers Battery cells Tactile sensorsCell plant: 2-3 yr
2 · Energy
Generation, storage, delivery		Solar, wind, nuclear, BESS, transmission, transformers, switchgearTransformers and GOES Grid interconnection SwitchgearGrid interconnection: ~5 yr
1 · Materials
Physical inputs		Lithium, nickel, graphite, copper, rare earths, polysilicon, steelRare earth magnets Copper Graphite Cathode material Lithium PolysiliconNew mine: ~16 yr

Two examples

A gigafactory and a robotaxi service share the bottom of the stack

Different at the top, the same at the bottom. Both draw on the same materials, the same grid and many of the same components, so they compete for the same constrained inputs.

Stack layers in a battery gigafactory and a robotaxi service
LayerBattery gigafactoryRobotaxi service
MaterialsLithium, nickel, graphite, copper foil, aluminumMetals, polymers, silicon, rare earth magnets
EnergyGrid supply, onsite solar, BESS, microgridCharging network, depot BESS, grid supply
ComponentsCells, modules, packs, power electronicsPacks, motors, cameras, radar, LiDAR, compute modules
ComputeMES, machine vision, AI quality controlAutonomy SoC, perception and planning
ProcessesCoating, calendering, assembly, filling, formationVehicle build, OTA pipeline, maintenance
InfrastructureCampus, utility interconnection, logisticsDepots, service centers, charging hubs, mapped roads
NodesRobots, AGVs, pack lines, formation racksIndividual robotaxis
NetworksPlant telemetry, supply data flowsDispatch, telemetry, OTA, V2X
SystemsBattery manufacturing and supply orchestrationAutonomous ride-hailing

Above the stack

Three scale layers

  • Civilization. Systems at the scale of cities and regions: industrial clusters, logistics corridors, electrified ports. Where investment and policy meet.
  • Planetary. Coordination across countries: energy and mineral trade, satellite coverage, standards. Where systemic risk sits, including export controls.
  • Space. Launch, orbital communications and sensing, and early orbital compute. Small today, but it already feeds the planetary layer.

Sources

Sources and scope

Scope
Nine-layer model of the electrified industrial economy, with the Bottleneck Atlas constraints mapped to layers
Sources
ElectronsX Bottleneck Atlas; IEA mine lead-time analysis; LBNL interconnection queue data; build-time ranges are indicative
Updated
October 2026