Data Center Feasibility Study in Texas

Texas near the top for this code

Texas ranks 2 of 52 among states for SBA loans under the data center class, with 103 approvals worth $61.7 million over FY2017 to FY2026, and at 1.47x its location quotient shows the class weighing more in Texas than nationally. Inside Texas it is a small group, 24 of 39 among classes. The loans are modest: the median is $215,000, or 1.05x the national figure for this code, and Texas approvals in the middle half run from $111,000 to $490,000. No metric clears our distinctness screen. Those sizes tell the reader what this class really holds in Texas, which at these sizes looks like hosting and data processing businesses more than data center buildings.

Who lends to Texas hosting firms

SBA 7(a) carries 97 of the Texas approvals and SBA 504 only 6, and the franchise share is 0.00%. Newtek Bank, National Association holds 16 loans worth $8.6 million, BayFirst National Bank 11 worth $1.7 million, The Huntington National Bank 6 worth $1.6 million and Northeast Bank 5 worth $545,600. Dallas County leads with 22 approvals, Travis follows with 18 and Harris with 16, then Collin with 10. Activity has grown, from 4 approvals in each of FY2020 and FY2021 to 14 worth $13.0 million in FY2023 and 18 worth $10.4 million in FY2026.

Cancellations high, charge-offs withheld

At least 17.4% of the FY2017 to FY2024 Texas approvals in this class were cancelled after approval. The ceiling is the same 17.4%, as nothing approved in those years remains undrawn. That is a high share of deals that never closed as approved. We withhold the Texas charge-off rate for this class: fewer than 30 of its loans funded in FY2017 to FY2021, too few for us to publish a rate, and the missing figure should not be read as an absence of losses.

For a borrower, the practical point is that a Texas loan in this class typically finances a technology service business, so a feasibility study has to test that business. Demand means contracted or recurring revenue: the customers, contract lengths and churn. Cost covers servers, power, cooling and any space build-out, priced from quotes, along with how quickly the equipment must be replaced. Debt service coverage should rest on revenue already under contract more than on pipeline, given how many approvals here were cancelled. Where a project does involve a building, the appraisal of a specialised facility depends on power capacity and its value to another tenant, and those are the facts the study should put in front of the appraiser.

A coding limit shapes the cell: colocation buildings share their code with hosting and data processing services and cannot be separated, so these Texas figures mix property-heavy projects with service firms that own little real estate.