Greg Abbott’s 3 August letter put capital allocation behind an administrative gate. Addressed to Public Utility Commission Chairman Thomas Gleeson and ERCOT chief executive Pablo Vegas, it directed examination of data centres “before any additional data centers are approved to move forward.” [E1] The immediate object is an interconnection process in which computing projects are competing for permission to become large new loads. Texas is therefore attaching scrutiny, security and potential loss of capital to a place in that queue.
The commission already has the machinery on the docket. Project No. 58481 is titled “RULEMAKING TO IMPLEMENT LARGE LOAD INTERCONNECTION STANDARDS UNDER PURA §37.0561,” and its proposal-for-publication filing was lodged on 12 March before publication in the Texas Register on 27 March. [E2][E3] Proposed §25.194(d)(10) sets security at “$50,000 per MW of the requested peak demand.” [E3] A 1,000-MW request therefore carries $50 million of security at that stated rate.
The harder filter arrives when a project withdraws or leaves capacity unused. Proposed §25.194(g)(4) and (h)(4) provide for “refunding 20% of the balance, the remaining 80% of the balance must be paid” after the specified costs and residual-balance calculation. [E3] That published language makes the 80% treatment part of the proposed rule itself, not merely a workshop position or staff comment. An applicant can still leave the queue, but under that proposal a large part of the residual commitment leaves with it.
Scale explains why the money matters. ERCOT said on 29 July that it was tracking approximately 474 GW of large loads seeking interconnection, with about 90% attributed to data centres. [E4] Its current all-time peak-demand record is 91,089 MW on 22 July, a figure the grid operator still labels unofficial until final settlement, so the tracked large-load total is about 5.20 times that peak. [E5] “Seeking interconnection” is the operative description: the 474 GW should not be described as 474 GW already under study, financed, permitted or being built.
The filed record also contains a drafting discrepancy that matters to anyone pricing an application. The Texas Register preamble describes a $100,000-per-MW figure, while proposed §25.194(d)(10) itself states $50,000 per MW of requested peak demand. [E3] The discrepancy is in the published record; it does not by itself establish which figure a final commission rule will retain.
One obvious counter-case is queue hygiene. A large unrecoverable deposit gives applicants a reason to abandon implausible requests before planners spend scarce study capacity on them. No published ERCOT or commission aggregate estimate in this record quantifies how much of the 474 GW is duplicate or speculative, so a capital screen could remove weak requests while also leaving the best-capitalised developers better able to hold multiple credible positions.
The call: p=0.46 that by 30 September 2026 the commission adopts a final rule keeping the nonrefundable share at 80% or more for the residual security treatment described in the proposal. YES requires a final commission order or adopted rule by that date that leaves at least 80% of the applicable residual balance nonrefundable; a staff draft, workshop document or governor’s statement does not settle it. NO settles if a qualifying final order or adopted rule sets that share below 80%, removes that treatment, or no qualifying final artifact exists by the deadline. The proposed rule has already put a price on persistence: in Texas, a place in the power queue is becoming a claim applicants must back with cash they may not get back. [E3]