SBA 7(a) Data Center Feasibility Study in New Jersey

Small is the word for New Jersey data center loans: a $167,500 median at 0.84x the nationwide 7(a) benchmark, and an interquartile band from $100,000 up to only $250,000. That tight band, with a top quartile boundary at $250,000, is the cell's most unusual feature, because it means few if any large facilities are financed under 7(a) here. The 30 loans total $14.8 million for FY2017 to FY2026. New Jersey is ninth on loans counted, yet data centers hold just the twenty-second place among its asset types. No loan was franchised.

Two banks and the central counties

TD Bank, National Association holds 7 loans and $735,000, and Northeast Bank 5 loans and $701,500, so average loans are near $100,000. That the lender tally simply tells which institutions hold notes, and says nothing about their approach. Middlesex County leads with 8 loans, Somerset follows with 6, Bergen, Burlington and Monmouth have 4, and Cumberland, Essex, Hudson and Hunterdon 1 each.

The pattern by year

Counts began at 3, 1 and 2, then reached 3 in FY2020 at $2.3 million. FY2021 and FY2022 had a single loan each at $250,000. FY2023 had 6 loans for $3.6 million, FY2024 8 loans but only $1.5 million, FY2025 4 for $835,000, and FY2026 a lone loan of $4.9 million, the largest of any year in the table. One outsized loan can thus stand out among many small ones.

What the code mixes in

The code does not distinguish colocation space from hosting services and also covers data processing firms. With loans this small, most borrowers are probably small hosting and IT firms. New Jersey publishes no cancellation or charge-off percentage for this cell, and silence there is not a zero.

What a New Jersey study has to prove

The study should state the business: whether the borrower leases racks, runs a managed-hosting service, or operates its own equipment room. It should test customer contracts and churn, power and cooling cost, connectivity, backup power and the lease of the space. At loans of this size, working capital and equipment dominate, so the study should show the useful life of servers and how they are replaced. Run the coverage test with the biggest customer gone, and show how the business would carry on if the premises were lost.