Deployment is the constraint

The technology to rebuild the physical economy mostly exists. The reactors work. The robots work. The solar cell and the battery are cheap and getting cheaper. What limits the pace now is a different thing. The binding constraint has moved from the lab to the field.

For two decades the scarce input was invention. Could the thing be made to work. That question is largely answered across the industrial stack. The scarce input now is deployment: getting the working thing built, connected, permitted, and running at scale. This is a less glamorous problem and a harder one. It lives in physics, queues, and order books, and no amount of code moves it.

Four mechanisms show where it binds.

Grid queues

Start with the clearest bottleneck. In the United States, the amount of power generation and storage waiting to connect to the grid runs past 2,000 gigawatts, per Lawrence Berkeley National Laboratory. That is more than the entire installed capacity of the country, sitting in a queue.

The wait has stretched with the pile. A project reaching commercial operation in 2025 spent a median of about five years in the interconnection queue. In 2008 the same step took under two years. The generators exist. The demand exists. The connection is the constraint, and it is measured in years.

The queue is not a paperwork delay. Connecting a large generator can force upgrades miles away on the network, and the study process that assigns those costs is slow, sequential, and easy to game by withdrawing. A data centre that wants firm power in 2027 is bidding against that clock. The power it needs may be built and still stranded behind a connection it cannot get.

Permitting

Permitting is the same problem wearing a different uniform. A transmission line crosses jurisdictions, and each one holds a veto. A large project accumulates environmental review, land rights, and local consent, and the calendar runs on the slowest of them.

The pattern repeats across the stack. A nuclear restart is an engineering job of months and a licensing job of years. A new mine or fab clears its technical hurdles long before it clears its legal ones. The kit is ready. The permission is not.

Order books

Where the queue is not regulatory, it is industrial. The equipment that builds the physical economy is made by a short list of firms with finite capacity, and their order books have lengthened. Grid transformers now carry lead times measured in years. High-voltage switchgear, turbines, and the specialist tools that outfit a semiconductor fab all run long.

An order book is a deployment constraint you cannot code your way past. You can want a transformer, fund a transformer, and design a plant around a transformer, and still wait two years for the transformer. The supplier base was sized for a slower world, and demand arrived faster than capacity.

Capacity is slow to add for the same reason the demand is hard to meet. A transformer plant is itself a capital-intensive, multi-year, physics-bound build. The firms that could expand are wary of committing to a spike that might fade, so they add capacity in careful steps. The bottleneck feeds itself, and the lead time is the price of that caution.

Learning curves

The fourth mechanism cuts the other way, and it is why the constraint rewards the effort spent on it. Deployment carries a cost. It also drives the cost down.

Wright’s law holds that for high-volume, short-cycle technologies, unit cost falls a fixed percentage with every doubling of cumulative production. Solar modules have fallen about 20% per doubling. Lithium-ion cells went from over 1,000 dollars per kilowatt-hour in 2010 to under 150 today. The learning came from building, not from a breakthrough. Each doubling of deployed volume bought the next cost reduction. Robots that ship in volume will ride the same curve, which means deployment is both the constraint and the engine that relaxes it.

The pilot-to-plant gap

One deployment problem deserves its own name. Call it the pilot-to-plant gap. A technology proves itself in a single demonstration, then stalls on the way to a hundred plants running it as routine. The first unit is an engineering achievement. The hundredth is an industrial one, and the two are different disciplines. The demo clears the science. The rollout clears the supply chain, the workforce, the financing, and the integration into a working operation. Most industrial technologies die in that gap, and the ones that cross it define the decade. It gets its own post.

Put the four mechanisms together and the shape is clear. Grid queues, permitting delays, stretched order books, and the slow grind of learning curves are the terrain the physical rebuild has to cross. The invention is done. The deployment is the work, and it is where the pace, the cost, and the winners will be decided.

The weekly read on where deployment binds is Industrial Sector Insights.