Saturday, September 5th, 2026
The constraint on the AI buildout has been described as electricity for two years. The nearer one is a procurement queue for hardware whose price and availability are now set by trade policy rather than by fabrication.
The pitch for putting data centres in orbit opens with free power and ends there. Power was never the part that could not be solved, and the constraint that replaces it is worse.
A ship out of service is capital earning nothing, and the number of places large enough to lift one out of the water has barely changed in decades. The queue, not the repair, is what costs the money.
Almost every byte that crosses an ocean crosses a cable, and the cables break constantly. The vessels that fix them are a few dozen ageing ships held on shared retainers, and the replacement pipeline is close to empty.
Freight moves by road across Europe not because the track is missing but because the places to load and unload it are. The investment now going in is unglamorous, terminal-shaped and long overdue.
Commercial interest in northern shipping routes is running well ahead of the capacity to escort, chart and rescue along them. The constraint is a shipyard problem that takes a decade to solve and was deferred for thirty years.
Organisations are sending each request to whichever model suits it, which lowers cost and quietly turns model selection into a piece of infrastructure somebody has to own.
Engineering organisations are building the same abstraction over the same infrastructure, independently, because the alternative is asking every team to learn all of it.
Authentication requirements that were optional for years are enforced, and organisations that never thought about their sending domain are discovering their mail has quietly stopped arriving.
Thousands of rural bridges are being downrated rather than replaced, and a load limit reroutes freight, school buses and fire apparatus onto roads that were never meant to carry them.
Billions went into berths, cranes and channel depth after the congestion of the early decade. The constraint moved twenty miles inland, to the road and rail that carry a container away from the water.
Water law across much of the country was written on the assumption that an aquifer was effectively bottomless and that a landowner's right to pump was a property right. Both assumptions are being retired, unevenly and expensively.
Governments across the country are rewriting approval processes after concluding that the gap between funded projects and finished ones is largely administrative.
A per-gallon levy funds American highways while fuel consumption decouples from road use. The replacement everyone points to is a mileage fee, and almost nobody has been willing to build one.
Training got the headlines and the capital budget. The recurring cost of running models in production is what finance departments are now trying to forecast, mostly badly.
Corporate AI budgets are shifting out of innovation labs and into core operating lines, a change that is quietly redrawing how companies account for technology itself.
The easy miles are wired. What remains are the households that made the economics impossible in the first place, and the programs now confronting them.
More generation capacity is waiting for permission to connect to the grid than is currently operating on it. The bottleneck is not construction or capital but a study process nobody designed for this volume.
Repatriation was dismissed for years as a vendor talking point. It is now a line item, driven less by ideology than by workloads whose usage stopped being unpredictable.
Tens of thousands of small systems have spent a generation postponing pipe replacement. The postponement is ending on its own schedule, and the bill is arriving in towns least able to bond for it.
The scramble of a thousand developer-tool startups is resolving into something duller and more durable: infrastructure priced, regulated and consolidated like the utility it always was.