About
Everyone is building. Nobody is checking.
Hundreds of billions of dollars are being committed right now to announced AI data centers. Most won't get built. Historically, most of what enters U.S. interconnection queues is never built — LBNL's Queued Up series finds that, of the capacity that requested interconnection in 2000–2020, 13% of that capacity had reached commercial operation by the end of 2025, and 75% of that capacity had been withdrawn — and no authoritative source tells a regulator, an investor, or a utility whether a specific project is real or speculative before the capital moves — or before the cost lands on the public's power bill. LOADSTAR exists to answer that one question, with evidence attached to every answer.
What we are
LOADSTAR is the reality index for AI power: a 0–100 score, built from public signals, that measures how much of an announced project has independent corroboration — starting with the power supply behind it, the buildout's #1 bottleneck. Like the indices that track market volatility or credit risk, it doesn't forecast the future or offer a take — it measures a present condition, transparently, so the number means the same thing to whoever reads it. And unlike any of them, you can watch this one build itself, live — every figure traceable to the official filing behind it.
Why now
In June 2026, the Federal Energy Regulatory Commission ordered six grid operators to filter speculative large-load requests out of their interconnection queues — the first time a federal regulator formally named the problem this index was built to measure. We didn't create that problem, and we don't take credit for the order. We started measuring it before it had a name.
What we won't do
Today, LOADSTAR covers the US at county and state level — not yet node or feeder. The score is descriptive: it measures corroboration now, not what will happen. The first sealed calls begin resolving around December 2026 — published either way, hit or miss — and until then we don't sell accuracy we haven't earned yet. About 93% of the sources behind today's scores are news reporting, not official filings — we mark which is which on every item, always. None of this is an apology. It's the boundary of what we can verify right now, stated plainly instead of blurred.