Interconnection is the new oil (unfortunately)

The scarcest resource in American energy is not lithium, and it is not land. It is a place in line.

About two terawatts of proposed generation and storage are waiting for permission to connect to the grid. The median wait is roughly five years. Most of the projects in that line are finished on paper and idle in practice. This is a software problem wearing a hard hat.

How the line works

A developer files a request. The utility runs a study to see what the new load or generator will do to its network. The study takes months. Its answer often says that three other projects, filed earlier, must be restudied first. The line moves backwards a little each time it moves forward.

None of this is malicious. The studies are hard, the staff are few, and the models were built when a feeder carried one kind of load. But a five-year queue turns a good project into a bad one, because costs move while it waits.

A place in line is worth more than the project standing in it.

What founders do about it

We met Priya Raman in 2024. Her first plan was a better study model, and we said no in nine days, with two pages on why. She rebuilt the plan around those pages. Phaseyard now runs interconnection studies in weeks, using the utility’s own data, with the engineer’s judgment left where it was. We said yes eleven months later, and it took us six days.

Clampwise comes at the same problem from the other end. A transformer that reports its own health installs in eleven minutes and gives the planner a real number instead of a guess. Kilnworks is a quieter case: an industrial customer who can say exactly when it will draw power is a customer a utility can say yes to.

Why this is a business

A queue is a market. Everyone in it wants to move up, and a few will pay for it. But the honest product is not a shortcut. It is a faster, cheaper, more checkable study, with the same engineers signing it. The utility gets an answer in weeks. The developer gets a number it can plan around. The engineer gets to spend a Friday on a judgment instead of a spreadsheet.

We do not expect these companies to look large in year one. A study tool sells to a handful of engineers at a time. But each engineer who can finish a study in a week tells the next one, and the next one has a queue of their own.

What we are watching

We expect the queue to shrink less by reform than by arithmetic. When enough studies are fast and checkable, the slow ones look odd. We are watching three things: study turnaround at the utilities that have adopted software, the share of requests withdrawn before study, and the number of engineers who say they would rather not go back.

We will keep writing about this until the line is shorter. It may take a while. We are comfortable with that, and we have a seat.

— Teo Okafor