The grid's demand shock is showing up as nodal basis — and batteries are the natural arbitrageur

The load story everyone repeats — data centers, EVs, electrification — is usually told as a capacity problem: not enough generation, not enough wires. It is also, and more immediately for anyone trading or owning storage, a price problem. The demand shock is already showing up as basis.

The demand shock has nowhere to go but the queue

The numbers are large and one-directional. Data centers are set to drive roughly 46% of US load growth through 2030 (NREL); US data-center capacity is heading toward 51–62 GW by the end of 2026, up from about 31 GW three years earlier; ERCOT alone logged 198 GW of new large-load interconnection requests in the first quarter of 2026. Meanwhile the grid that would serve all of it cannot be built fast enough — about 2.3 TW sit in US interconnection queues, with a 4+ year median wait and only ~13% of projects queued between 2000 and 2019 ever built. New load and the new supply meant to serve it stall in the same line.

That mismatch prices out as nodal basis

When load lands where the network is constrained and generation can’t interconnect fast enough to relieve it, the price separates — at specific nodes, not evenly across a zone. Those nodal basis spreads are widening and persistent, which makes them a structural feature to trade around rather than noise to average out. Battery storage is the natural arbitrageur of that spread: energy arbitrage already drives 91% of CAISO and 76% of ERCOT battery revenue, and installed BESS in ERCOT went from about 1 GW in 2020 to 16.3 GW by the middle of 2026. The asset class and the opportunity are growing into each other.

The tools mostly look the wrong way

Most of the desk’s tooling was built for gas and reports what already cleared, at the hub. That is backward-looking and zonal — the two properties a battery can least afford, because it is paid at a single node for what the basis is about to do. Benchmarking platforms tell you what value was realized last quarter; static indices give a daily snapshot. Useful context, both — but neither ranks nodes by forward opportunity or ties the signal to a dispatch decision.

The demand shock isn’t only a reason to build more storage — it’s a reason to run the storage you already have differently. The spread is structural, it lives at the node, and it is legible before it’s obvious if you’re looking forward instead of back. That is the edge worth building for.