Texas's Batch Zero process turned data-center demand into an eligibility screen [1][2][3]

The prior file at ngtimes.org/2026/06/21/ferc-large-load-orders-start-ai-power-filing-clock asked for a public receipt before the frame hardened. Today's record supplies one, but it does not settle every claim.

Scale explains why Texas invented a new process instead of working the old one. The Utility Dive figure behind this week's coverage puts roughly 438 gigawatts of potential large load in ERCOT's connection pipeline, a number so far beyond historical demand growth that treating it as real capacity would break planning arithmetic. Grid operators have seen this pattern before in generation interconnection, where speculative projects clog queues for years and crowd out builders who intend to construct. Batch Zero is ERCOT's admission that the same theater has arrived on the consumption side: hundreds of applicants, many holding optioned land and little else, each asking for megawatts nobody can yet prove they will use. [1][2]

The screen's logic is deliberately unromantic. Applicants advance by documenting what speculative projects lack: control of the site rather than an option, financing that survives scrutiny, completed system studies, willingness to accept curtailment during scarcity, or self-supply arrangements that reduce dependence on the grid at peak. Each criterion converts a press release into evidence. A developer with land, capital, and engineering already done passes; a land flipper with a logo does not. That is the entire design, and it is why both hyperscalers and speculators watched the first batch list so closely. [2][3]

The MSM frame is straightforward: Texas is sorting a massive large-load queue. The X frame is sharper and less patient: data centers are squatting on power they may never use. Both frames understate what is actually being arbitrated. Squatting is one failure mode; the opposite is equally dangerous. If screening gets too aggressive, legitimate flexible load gets priced out of a grid that could have used its off-peak appetite. Batch Zero's terms decide which error Texas prefers, and those terms are public documents worth reading line by line. [1][2]

The ratepayer angle is where X's suspicion earns keep. New large loads trigger transmission upgrades whose costs traditionally socialize across all customers. Whether Batch Zero forces data centers to fund their own network impact, post security, or accept curtailment determines whether Texans' bills subsidize compute buildouts for global customers. The accepted batch's cost-protection terms will answer that question before any legislator does. [1][3]

The stakes extend nationally. ERCOT connects to almost nothing outside Texas, which makes it the cleanest laboratory in America for large-load policy: no interstate complications, one regulator, and demand growth measured against its own history rather than imported electrons. Whatever eligibility template survives here travels to PJM and other RTOs through consultants and copycat tariffs within a year. [1][2]

There is also a quieter signal in the process itself. An operator willing to tell 438 gigawatts of aspirational demand "prove it" is betting that genuine AI-driven load is strong enough to survive screening. That bet is testable in the only currency that matters: how many applicants clear Batch Zero with binding commitments attached. [2]

That matters because the public decision is no longer about whether the topic feels important. It is about which document controls the next claim. Here the controlling documents are the Batch Zero tariff provisions and the accepted-project list, not the announcement count. [1][2][3]

The remaining gap is practical. The first batch's accepted projects and their cost-protection terms remain the next receipts. Until ERCOT publishes them, the responsible headline is a receipt check, not a victory lap. Count who clears, then read what they agreed to pay for.

-- THEO KAPLAN, San Francisco