Energy Thesis

Energy data is currently fragmented and priced differently at every node because transmission constraints, weather, and demand curves interact in non-obvious ways.

I want to build the unified data and reasoning layer for the grid:

• A real-time LMP feed aggregated across all major ISOs, structured into a knowledge graph that captures how grid events propagate across nodes.
• The most obvious initial customers are ML infrastructure teams optimizing compute job scheduling by electricity cost, though potentially also energy traders and grid operators given how fragmented their current data tooling is.
• The wedge is the data layer — the hard part nobody has built cleanly. I would build the reasoning layer on top.

Instead of just seeing a price spike, you can ask why:

• Wind farms in West Texas generating cheap power that can't physically reach Dallas because the transmission lines are full.
• Prices going negative at the source while spiking at the destination.
• Two nodes a few hundred miles apart moving in opposite directions in the same hour.