Data Center Feasibility Study and SBA Loan Data
The SBA loan record for data center projects nationally and by state, and what a feasibility study for one has to test.
Daniel Smith, MAI
Data center feasibility study consultant
Loan Analytics is a data center feasibility study consultant. We prepare the independent study a lender files when it finances a hosting company, a colocation operator or an owner building its own facility, and the SBA's loan files show what that lender has already seen in this class. From FY2017 to FY2026 the agency approved 911 SBA 7(a) loans to data center borrowers for $476.9 million and 50 SBA 504 loans for $41.6 million, at medians of $200,000 and $591,000. Those medians describe operating hosting and processing firms more than buildings, yet the same record holds 66 7(a) approvals in the largest size band the table fills, together carrying $225.4 million, more than any other band. A request at that scale goes before a credit officer whose first questions are about power.
Our scope follows the points where these loans have gone wrong. Before funding, at least 11.2% of 7(a) data center approvals and 10.9% of 504 approvals were cancelled, and in Texas, second among the states by loans, the cancellation floor reaches 17.4%. Once loans fund, 7(a) data center loans that disbursed from FY2017 through FY2021 show a charge-off floor of 6.9%, against 3.3% charged off within five years, so losses in this class kept arriving after the early years. The study therefore tests two things: that the project can close as approved, and that it still carries its debt once servers age and contracts come up for renewal. It covers the power the utility can deliver to the site and when, redundancy, each customer contract and its churn, the cloud and colocation alternatives those customers could move to, the electrical, cooling, fire suppression and fiber budget, and debt service coverage measured on contracted revenue.
We write each study for the lender that will read it, and the data center lenders in the record differ sharply. Newtek Bank, National Association approved 97 loans for $60.3 million, Live Oak Banking Company 26 loans for $40.0 million, and TD Bank, National Association 33 loans for $2.6 million. A study that supports a small equipment request is not the study a lender needs when a hosting firm buys its own building with a 504 loan, where the program's upper quartile is $1.1 million. Each study is fitted to the program the lender files under, to that lender and to the site, and Daniel Smith, MAI signs it; the program may be 7(a) or 504, USDA's Business and Industry guarantee, or plain conventional credit.
Request Scope & Timeline
Hosting businesses more than buildings
Data center borrowers took 911 loans under 7(a) over FY2017 to FY2026, totaling $476.9 million at a median of $200,000, while the 504 program made just 50 loans for $41.6 million, at a median of $591,000. Those medians are small for a property type sized by its power supply, and the limitation explains why: the industry code holds data processing and hosting businesses together with any colocation building, and the two cannot be separated. Most of what this record shows is lending to companies that host, process or manage data, with equipment and working capital the likely uses.
Floors for hosting borrowers
The 7(a) charge-off floor is 6.9% for loans that funded in fiscal 2017 to 2021, a loss record that reflects operating businesses whose main assets are servers and customer contracts rather than land. The 504 side has too few data center loans in that cohort, under 30, for us to print a loss rate. Cancellation after approval runs at 11.2% for 7(a) and 10.9% for 504. Annual 7(a) approvals jumped to 151 in FY2024 and 145 in FY2025, then eased to 87 in FY2026.
California leads with 113 loans and $64.9 million, and Texas follows with 103 loans and $61.7 million. New York's 66 loans carried only $17.5 million, against $42.6 million for Florida's 65, so the typical New York borrower is a much smaller hosting or processing firm. Virginia, with 26 loans and $8.9 million, closes the table.
Power first, then customers
Any study for a data center, or for a hosting business that wants its own building, has to start with power: the utility capacity available at the site, the timeline and cost to bring it in, and redundancy. Demand should be shown through signed or letter-of-intent customers, their contract terms and their churn, and through the competing colocation and cloud options they could switch to. Cost has to include electrical and cooling plant, fire suppression, fiber connectivity and physical security, which can exceed the cost of the shell. Debt service coverage should be tested on contracted revenue, with power cost passed through or hedged and equipment replacement cycles reserved for. The appraisal must separate the building from its mechanical and electrical systems and value the shell for an alternative user, because specialized systems lose value quickly.
Because the code is the borrower's line of business, a company that owns a server room inside its own office is counted in its own industry, and a large wholesale data hall financed outside SBA leaves no trace here.
SBA loan record for data center, United States, FY2017 to FY2026
| Program | Loans | Dollars | Median loan | Cancelled (floor) | Charged off (floor) |
|---|---|---|---|---|---|
| SBA 7(a) | 911 | $476.9 million | $200,000 | 11.2% | 6.9% |
| SBA 504 | 50 | $41.6 million | $591,000 | 10.9% | not reported: fewer than 30 loans in the cohort |
Download this table (CSV). Cite as: Loan Analytics, SBA loan record for data center, United States, FY2017 to FY2026, https://www.analytics.loan/data-center-feasibility-study#record-C31.
Cancellation counts FY2017 to FY2024 approvals; charge-off counts FY2017 to FY2021 disbursed loans to date. Both are floors, and a rate needs at least 30 loans in its cohort.
SBA 7(a) data center loans approved nationally, by fiscal year
| Fiscal year | Loans | Dollars |
|---|---|---|
| FY2017 | 98 | $46.7 million |
| FY2018 | 80 | $31.1 million |
| FY2019 | 69 | $29.8 million |
| FY2020 | 55 | $29.0 million |
| FY2021 | 63 | $54.1 million |
| FY2022 | 86 | $43.1 million |
| FY2023 | 77 | $46.2 million |
| FY2024 | 151 | $72.8 million |
| FY2025 | 145 | $73.6 million |
| FY2026 | 87 | $50.5 million |
Download this table (CSV). Cite as: Loan Analytics, SBA loan record for data center, United States, FY2017 to FY2026, https://www.analytics.loan/data-center-feasibility-study#record-C31.
SBA 504 data center loans approved nationally, by fiscal year
| Fiscal year | Loans | Dollars |
|---|---|---|
| FY2017 | 8 | $7.4 million |
| FY2018 | 3 | $2.5 million |
| FY2019 | 11 | $7.2 million |
| FY2020 | 2 | $917,000 |
| FY2021 | 6 | $3.9 million |
| FY2022 | 7 | $4.5 million |
| FY2023 | 4 | $4.3 million |
| FY2024 | 5 | $2.4 million |
| FY2025 | 3 | $8.4 million |
| FY2026 | 1 | $217,000 |
Download this table (CSV). Cite as: Loan Analytics, SBA loan record for data center, United States, FY2017 to FY2026, https://www.analytics.loan/data-center-feasibility-study#record-C31.
States with the most data center loans
| State | Loans | Dollars |
|---|---|---|
| California | 113 | $64.9 million |
| Texas | 103 | $61.7 million |
| New York | 66 | $17.5 million |
| Florida | 65 | $42.6 million |
| Illinois | 39 | $21.2 million |
| Ohio | 39 | $17.3 million |
| Colorado | 32 | $23.4 million |
| Maryland | 31 | $22.5 million |
| New Jersey | 31 | $15.6 million |
| Georgia | 28 | $15.4 million |
| Minnesota | 28 | $12.9 million |
| North Carolina | 28 | $10.4 million |
| Michigan | 27 | $21.8 million |
| Virginia | 26 | $8.9 million |
Download this table (CSV). Cite as: Loan Analytics, SBA loan record for data center, United States, FY2017 to FY2026, https://www.analytics.loan/data-center-feasibility-study#record-C31.
Data center feasibility study data by state
- Data center feasibility study in California
- Data center feasibility study in Texas
- Data center feasibility study in New York
- Data center feasibility study in Florida
- Data center feasibility study in Illinois
- Data center feasibility study in Ohio
- Data center feasibility study in Colorado
- Data center feasibility study in Maryland
- Data center feasibility study in New Jersey
- Data center feasibility study in Georgia
- Data center feasibility study in Minnesota
- Data center feasibility study in North Carolina
- Data center feasibility study in Michigan
- Data center feasibility study in Virginia
Data center feasibility study data by metro area
- Data center feasibility study in the New York-Newark-Jersey City, NY-NJ Metro Area
- Data center feasibility study in the Dallas-Fort Worth-Arlington, TX Metro Area
- Data center feasibility study in the Chicago-Naperville-Elgin, IL-IN Metro Area
- Data center feasibility study in the Washington-Arlington-Alexandria, DC-VA-MD-WV Metro Area
- Data center feasibility study in the Los Angeles-Long Beach-Anaheim, CA Metro Area
- Data center feasibility study in the Miami-Fort Lauderdale-West Palm Beach, FL Metro Area
- Data center feasibility study in the Austin-Round Rock-San Marcos, TX Metro Area
Colocation buildings cannot be separated from hosting services. Shared code: 518210 also holds data processing and hosting businesses.
Source: U.S. Small Business Administration, 7(a) and 504 loan-level FOIA files (data.sba.gov), as of 2026-09-30.
Related feasibility study pages
- Data center feasibility study and SBA loan data
- SBA feasibility study company
- Data Center Feasibility Study in California
- Data Center Feasibility Study in Texas
- Data Center Feasibility Study in New York
- Data Center Feasibility Study in Florida
- Data Center Feasibility Study in Illinois
- Data Center Feasibility Study in Ohio
- Data Center Feasibility Study in Colorado
- Data Center Feasibility Study in Maryland
- Data Center Feasibility Study in New Jersey
- Data Center Feasibility Study in Georgia
- Data Center Feasibility Study in Minnesota
- Data Center Feasibility Study in North Carolina
SBA lending to data center projects, FY1991 to FY2026
Every SBA 7(a) and 504 loan approved for a data center project since FY1991, by fiscal year, nationally.
A record that begins in FY2003
The table holds no count for FY1991 to FY2002. In those years too few loans in SBA's files carry a code that places them in this class, so the years are marked not measured rather than shown as empty. The series opens in FY2003 with 29 7(a) approvals and 2 under 504, and the 7(a) count then climbed fast: 54 in FY2004, 80 in FY2005, 120 in FY2006, 125 in FY2007 and 129 in FY2008, a count the class did not pass again until FY2024. The money did not climb with it. FY2006's 120 loans carried $12.8 million and FY2008's 129 loans carried $20.9 million, less than the $54.1 million that FY2021 carried on 63 loans. The first wave of SBA lending in this class was a wave of small approvals to processing and hosting firms, and a lender who reads the counts of those years as a building boom misreads them.
Larger loans on flatter counts, FY2009 to FY2015
After FY2008 the count fell to 82 in both FY2009 and FY2010, yet FY2009's $20.8 million nearly matched the $20.9 million of FY2008 on far fewer loans. From there the dollars outran the counts: 105 loans for $30.6 million in FY2011, 67 for $24.9 million in FY2012, 77 for $31.9 million in FY2013, 84 for $25.0 million in FY2014 and 115 for $46.7 million in FY2015. FY2013 shows the change most clearly, because its 77 loans carried more money than the 129 of FY2008. A comparison drawn from the whole record would blend the small approvals of the mid-2000s with the larger ones that followed, so our studies read a request against the FY2017 to FY2026 window and against the size bands further down this page, not against the long-run count.
FY2017 to FY2026: a dip, then the busiest years in the record
The window opens with 98 loans and $46.7 million in FY2017, the same dollar total FY2015 had reached on 115 loans. The count then slid to 69 in FY2019 and 55 in FY2020, the fewest since FY2004, before FY2021 turned it with 63 loans for $54.1 million, the highest dollar total the table had printed to that point. FY2022 added 86 loans for $43.1 million. Then came the surge: 151 loans for $72.8 million in FY2024, the most loans and the largest printed dollar total in the class record, followed by 145 loans in FY2025. FY2026 fell back to 87. The dollar totals for FY2018, FY2020, FY2023, FY2025 and FY2026 are withheld in this table, and this page puts no estimate in their place.
What the surge means for a request filed now
The 6.9% charge-off floor on this page counts loans disbursed in FY2017 to FY2021. None of the FY2024 or FY2025 approvals, the largest years of lending the class has produced, sits inside that cohort, and their loss record has not been written yet. A data center loan put before a credit committee now is read by people who know the class grew quickly, who cannot yet see how that growth performed, and who saw approvals drop to 87 in FY2026. That committee leans on the borrower's own evidence rather than on the class history, so the study has to carry it: contracted revenue by customer and term, the churn those contracts have shown, the power cost and how it passes through to customers, and the reserve that replaces equipment before the loan matures.
The 504 side: thin in every year
The 504 record never passed 18 approvals in a year. That peak came in FY2010, when 18 loans carried $13.2 million against $17.3 million for 7(a) in the class, the only year in the printed record when the two programs came near each other in dollars. FY2013, with 10 loans and $10.8 million, and FY2019, with 11 loans and $7.2 million, are the only other years at 10 or more, and no year after FY2019 passed 7. Across the whole record the program made 148 data center loans for $113.6 million, against 2,158 7(a) loans for $780.9 million; in the FY2017 to FY2026 window it made 50, and in FY2026 just 1. Because 504 loans in this class run larger, at a median of $591,000, a borrower who brings a 504 data center project brings a file type lenders have rarely seen and one with no loss rate to read it against. The study has to supply the market and cost evidence that the program's history cannot.
Where the money sits across the record
The FY2017 to FY2026 window holds 911 of the record's 2,158 7(a) loans, and $476.9 million of its $780.9 million. Measured by money rather than by count, SBA data center lending is a recent business: FY2006 to FY2008 produced counts not far below the recent peak on far smaller totals, and the largest printed dollar totals all fall in FY2015 or later. For a borrower this matters in a practical way. The loans a lender will set a new request beside are the recent ones, made at today's sizes and under power and contract conditions closer to the ones the project faces, and a study that cites the mid-2000s counts as precedent compares the request with loans of a different kind.
| Fiscal year | SBA 7(a) loans | SBA 7(a) dollars | SBA 504 loans | SBA 504 dollars |
|---|---|---|---|---|
| FY1991 | not measured | not measured | not measured | not measured |
| FY1992 | not measured | not measured | not measured | not measured |
| FY1993 | not measured | not measured | not measured | not measured |
| FY1994 | not measured | not measured | not measured | not measured |
| FY1995 | not measured | not measured | not measured | not measured |
| FY1996 | not measured | not measured | not measured | not measured |
| FY1997 | not measured | not measured | not measured | not measured |
| FY1998 | not measured | not measured | not measured | not measured |
| FY1999 | not measured | not measured | not measured | not measured |
| FY2000 | not measured | not measured | not measured | not measured |
| FY2001 | not measured | not measured | not measured | not measured |
| FY2002 | not measured | not measured | not measured | not measured |
| FY2003 | 29 | withheld | 2 | fewer than 5 loans |
| FY2004 | 54 | withheld | 3 | fewer than 5 loans |
| FY2005 | 80 | $13.9 million | 5 | $2.4 million |
| FY2006 | 120 | $12.8 million | 9 | $3.3 million |
| FY2007 | 125 | $20.5 million | 7 | $4.4 million |
| FY2008 | 129 | $20.9 million | 7 | $2.3 million |
| FY2009 | 82 | $20.8 million | 7 | $7.6 million |
| FY2010 | 82 | $17.3 million | 18 | $13.2 million |
| FY2011 | 105 | $30.6 million | 5 | $1.8 million |
| FY2012 | 67 | $24.9 million | 8 | $8.1 million |
| FY2013 | 77 | $31.9 million | 10 | $10.8 million |
| FY2014 | 84 | $25.0 million | 6 | $3.8 million |
| FY2015 | 115 | $46.7 million | 7 | $8.6 million |
| FY2016 | 98 | withheld | 4 | fewer than 5 loans |
| FY2017 | 98 | $46.7 million | 8 | $7.4 million |
| FY2018 | 80 | withheld | 3 | fewer than 5 loans |
| FY2019 | 69 | $29.8 million | 11 | $7.2 million |
| FY2020 | 55 | withheld | 2 | fewer than 5 loans |
| FY2021 | 63 | $54.1 million | 6 | $3.9 million |
| FY2022 | 86 | $43.1 million | 7 | $4.5 million |
| FY2023 | 77 | withheld | 4 | fewer than 5 loans |
| FY2024 | 151 | $72.8 million | 5 | $2.4 million |
| FY2025 | 145 | withheld | 3 | fewer than 5 loans |
| FY2026 | 87 | withheld | 1 | fewer than 5 loans |
| FY1991 to FY2026 | 2,158 | $780.9 million | 148 | $113.6 million |
| FY2017 to FY2026 | 911 | $476.9 million | 50 | $41.6 million |
Source: SBA 7(a) and 504 loan-level FOIA files as of 2026-09-30 (data.sba.gov; gross approvals, cancelled approvals included), aggregated by Loan Analytics; each figure is kept with its source file and retrieval date. A class figure counts the loans whose NAICS code places them in the class; in years when SBA's files leave the code missing or unplaceable for more than one loan in a hundred, the figure is a floor, and where too few loans are coded the year is not measured. Download this table (CSV). Cite as: Loan Analytics, "SBA 7(a) and 504 loans to data center projects by fiscal year", https://www.analytics.loan/data-center-feasibility-study#sba-record-sba-by-year.
Every state ranked for data center lending
SBA loans to data center projects in each state from FY2017 to FY2026, ranked, with loan size, program split and the cohort floors where at least 30 loans support them.
California and Texas at the top
California's 113 loans for $64.9 million and Texas's 103 for $61.7 million stand well clear of the rest; New York follows at 66 and Florida at 65. The two leaders borrow differently. California's median data center loan is $255,800 against $215,000 in Texas, and California holds 10 of the class's 50 SBA 504 loans, more than any other state, while Texas holds 6. A study for a California project can point to a state record deep enough to carry a median, a cancellation floor and a charge-off floor. Most states cannot, and the study for them has to borrow its benchmarks from the metro row or from the national class.
Cancellation runs high in Texas and Florida
The cancellation floors separate the leading states sharply. Texas stands at 17.4% and Florida at 14.9%, both above the national floors of 11.2% for 7(a) and 10.9% for 504, while New York sits at 8.2%, California at 5.7% and Ohio at 3.3%, the lowest the table prints. A cancelled approval is a loan the lender agreed to make that never funded. In Texas and Florida, then, a large part of the risk in this class shows up between approval and closing, before any loss could be counted. For a site in either state the study has to test the conditions a data center loan must meet to close: a power delivery date the utility will put in writing, equipment quotes that still hold at closing, and customer commitments that survive until the money moves.
California's loss floor, the only one a state prints
California is the only state with enough FY2017 to FY2021 loans for a charge-off floor, and its 6.0% counts both programs together. That floor sits under the 6.9% national 7(a) floor, and California's cancellation floor of 5.7% is the second lowest the table prints, so on the printed floors California's data center loans have closed and held up better than the class as a whole. A California lender can read a request against a state record of its own. Everywhere else the charge-off column reads not reported, because no other state reaches 30 loans in the cohort. Those cells are not zeros, and a study for a project outside California must not present them as a clean record; it measures the downside against the national floor and against the borrower's own contracts.
Small tickets in New York, Minnesota and Virginia
New York's 66 loans sit at a median of $132,500, while Florida's 65 sit at $275,000, the highest state median in the table. Missouri, far below New York by count with 24 loans, carried $17.7 million, slightly more than New York's $17.5 million. Minnesota's median of $102,800 is the lowest printed, and Virginia's $118,000 is next. In these states the SBA borrowers in this class are mostly small hosting and processing firms, and the state record offers few large data center loans to compare with. A building-scale project in New York or Minnesota cannot borrow scale from its state row; the study builds the case from the site's power, the signed customers and, where one exists, the metro row, which in New York's case is larger than the state's.
Virginia, West Virginia and the halls this record cannot see
Virginia, the last state in the table with enough loans for a median, has 26 loans for $8.9 million. West Virginia's 5 loans carried $9.1 million, more than Virginia's 26, and Connecticut's 10 loans carried $9.6 million, also more. The pattern follows from what this class counts, namely each borrower's own trade: a hosting firm appears, and a large wholesale hall financed outside SBA does not. A lender who reads this table for Virginia sees small operating firms, not the building projects some sponsors bring, and the study has to say plainly which kind of project it underwrites and draw its comparisons from evidence of that kind. From Missouri down, the table gives counts and dollars only, since every one of those states falls short of the loan count a median or floor needs, and the study for a site there leans on the national quartiles and its own fieldwork.
Mid-table states with larger loans, and states without a 504 loan
Below the leaders, counts and dollars part ways. Colorado's 32 loans carried $23.4 million, Maryland's 31 carried $22.5 million and Michigan's 27 carried $21.8 million, each more than Illinois carried on 39 loans ($21.2 million) and more than Ohio's $17.3 million, also on 39. North Carolina shares Colorado's $250,000 median, but its 28 loans carried only $10.4 million. Maryland, Georgia and North Carolina made no 504 data center loan in the window, and Ohio, Colorado, New Jersey, Minnesota and Michigan made 1 each. A 504 data center project in those states reaches a lender with almost no state precedent in that program, which raises the weight the study's market, power and cost evidence must bear.
| State | Loans | Dollars | Median loan | SBA 7(a) loans | SBA 504 loans | Cancelled (floor) | Charged off (floor) |
|---|---|---|---|---|---|---|---|
| California | 113 | $64.9 million | $255,800 | 103 | 10 | 5.7% | 6.0% |
| Texas | 103 | $61.7 million | $215,000 | 97 | 6 | 17.4% | not reported: fewer than 30 loans in the cohort |
| New York | 66 | $17.5 million | $132,500 | 64 | 2 | 8.2% | not reported: fewer than 30 loans in the cohort |
| Florida | 65 | $42.6 million | $275,000 | 60 | 5 | 14.9% | not reported: fewer than 30 loans in the cohort |
| Illinois | 39 | $21.2 million | $200,000 | 34 | 5 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Ohio | 39 | $17.3 million | $200,000 | 38 | 1 | 3.3% | not reported: fewer than 30 loans in the cohort |
| Colorado | 32 | $23.4 million | $250,000 | 31 | 1 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Maryland | 31 | $22.5 million | $170,000 | 31 | 0 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| New Jersey | 31 | $15.6 million | $185,000 | 30 | 1 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Georgia | 28 | $15.4 million | $180,000 | 28 | 0 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Minnesota | 28 | $12.9 million | $102,800 | 27 | 1 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| North Carolina | 28 | $10.4 million | $250,000 | 28 | 0 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Michigan | 27 | $21.8 million | $185,000 | 26 | 1 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Virginia | 26 | $8.9 million | $118,000 | 24 | 2 | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
| Missouri | 24 | $17.7 million | too few loans for these measures | ||||
| Massachusetts | 22 | $5.1 million | too few loans for these measures | ||||
| Utah | 21 | $8.1 million | too few loans for these measures | ||||
| Washington | 21 | $10.1 million | too few loans for these measures | ||||
| Arizona | 16 | $10.0 million | too few loans for these measures | ||||
| Pennsylvania | 16 | $11.4 million | too few loans for these measures | ||||
| Wisconsin | 16 | $10.3 million | too few loans for these measures | ||||
| Oregon | 15 | $6.4 million | too few loans for these measures | ||||
| Nevada | 13 | $7.3 million | too few loans for these measures | ||||
| Indiana | 11 | $3.5 million | too few loans for these measures | ||||
| Connecticut | 10 | $9.6 million | too few loans for these measures | ||||
| Delaware | 9 | $5.5 million | too few loans for these measures | ||||
| Kentucky | 9 | $5.6 million | too few loans for these measures | ||||
| New Hampshire | 9 | $2.2 million | too few loans for these measures | ||||
| Idaho | 8 | $3.9 million | too few loans for these measures | ||||
| Kansas | 8 | $5.2 million | too few loans for these measures | ||||
| South Carolina | 8 | $2.4 million | too few loans for these measures | ||||
| Tennessee | 7 | $7.1 million | too few loans for these measures | ||||
| Iowa | 6 | $535,000 | too few loans for these measures | ||||
| Oklahoma | 6 | $2.8 million | too few loans for these measures | ||||
| West Virginia | 5 | $9.1 million | too few loans for these measures | ||||
| Wyoming | 5 | $1.6 million | too few loans for these measures | ||||
| Hawaii | 4 | fewer than five loans | too few loans for these measures | ||||
| Louisiana | 4 | fewer than five loans | too few loans for these measures | ||||
| Alaska | 3 | fewer than five loans | too few loans for these measures | ||||
| Alabama | 3 | fewer than five loans | too few loans for these measures | ||||
| Rhode Island | 3 | fewer than five loans | too few loans for these measures | ||||
| Arkansas | 2 | fewer than five loans | too few loans for these measures | ||||
| Mississippi | 2 | fewer than five loans | too few loans for these measures | ||||
| Montana | 2 | fewer than five loans | too few loans for these measures | ||||
| North Dakota | 2 | fewer than five loans | too few loans for these measures | ||||
| Nebraska | 2 | fewer than five loans | too few loans for these measures | ||||
| New Mexico | 2 | fewer than five loans | too few loans for these measures | ||||
| Vermont | 2 | fewer than five loans | too few loans for these measures | ||||
| Maine | 1 | fewer than five loans | too few loans for these measures | ||||
| South Dakota | 1 | fewer than five loans | too few loans for these measures |
Source: Loan Analytics' SBA loan record from SBA's 7(a) and 504 loan-level files, approvals FY2017 to FY2026 as of 2026-09-30; rates are floors and need at least 30 loans in their cohort; dollars are not shown where fewer than five loans would reveal a single loan. Download this table (CSV). Cite as: Loan Analytics, "SBA data center loans by state, FY2017 to FY2026", https://www.analytics.loan/data-center-feasibility-study#states-states.
Metro areas ranked for data center lending
The metropolitan and micropolitan areas with a data center loan record, ranked by SBA loans.
New York: the metro row carries more than the state
The New York-Newark-Jersey City metro leads the metro table with 67 loans and $24.1 million, against 66 loans and $17.5 million for New York State, because the metro reaches into New Jersey, whose statewide record is 31 loans and $15.6 million. The metro median, $150,000, sits between New York's $132,500 and New Jersey's $185,000. For a site anywhere in that area, the lender's natural comparison is the metro row rather than either state, and the study uses it: a request far above the $150,000 median sits in the upper reach of the area's data center record and needs power and customer evidence to show why it is that large.
Texas places Dallas and Austin, California only Los Angeles
Texas is the only state that places more than a single metro on the table. Dallas-Fort Worth-Arlington has 45 loans for $23.4 million at a $200,000 median, and Austin-Round Rock-San Marcos has 27 loans for $21.2 million at $350,000, the highest median of any metro. Austin's median sits above the Texas median of $215,000 and Dallas's below it. California places only Los Angeles-Long Beach-Anaheim, with 37 of the state's 113 loans and $27.5 million, the most dollars of any metro, at a median of $327,500; the rest of California's record is spread across areas that do not reach the table. A study in Austin or Los Angeles is read against the large-ticket end of the class, and one elsewhere in Texas or California against a state median that Austin and Los Angeles lift.
Washington, Chicago and Miami against their states
The Washington-Arlington-Alexandria metro has 39 loans for $23.9 million at a median of $249,100, above both the Virginia median of $118,000 and the Maryland median of $170,000. Data center borrowers in that corridor take larger loans than either state's statewide record suggests, so a study there compares the request with the metro row. Chicago-Naperville-Elgin runs the other way: 42 loans, more than Illinois's 39 because the metro reaches into Indiana, at a $187,500 median below the Illinois $200,000. Miami-Fort Lauderdale-West Palm Beach holds 30 of Florida's 65 loans at a $210,000 median, well under Florida's $275,000, which suggests Florida's larger data center loans were made outside the Miami metro.
No loss rate at metro level
No metro on the table prints a charge-off floor. New York-Newark-Jersey City shows not reported, its cohort being under the 30 loans a rate needs, and every other metro shows withheld. Neither reading is a zero, and neither is evidence that the loans performed well. A lender financing a data center in one of these metros has no local loss history in this class, so it falls back on the national 7(a) floor of 6.9%, or on California's 6.0% for a Los Angeles site, and on the borrower's own record. The study's downside case therefore has to stand on the project: what coverage looks like when the largest customer leaves, when power cost rises without a pass-through, and when equipment has to be replaced before the debt is repaid.
| Metro or micro area | Loans | Dollars | Median loan | Charged off (floor) |
|---|---|---|---|---|
| New York-Newark-Jersey City, NY-NJ Metro Area | 67 | $24.1 million | $150,000 | not reported: fewer than 30 loans in the cohort |
| Dallas-Fort Worth-Arlington, TX Metro Area | 45 | $23.4 million | $200,000 | withheld |
| Chicago-Naperville-Elgin, IL-IN Metro Area | 42 | $17.1 million | $187,500 | withheld |
| Washington-Arlington-Alexandria, DC-VA-MD-WV Metro Area | 39 | $23.9 million | $249,100 | withheld |
| Los Angeles-Long Beach-Anaheim, CA Metro Area | 37 | $27.5 million | $327,500 | withheld |
| Miami-Fort Lauderdale-West Palm Beach, FL Metro Area | 30 | $14.9 million | $210,000 | withheld |
| Austin-Round Rock-San Marcos, TX Metro Area | 27 | $21.2 million | $350,000 | withheld |
Source: Loan Analytics' SBA loan record from SBA's 7(a) and 504 loan-level files, approvals FY2017 to FY2026 as of 2026-09-30; rates are floors and need at least 30 loans in their cohort; dollars are not shown where fewer than five loans would reveal a single loan. Download this table (CSV). Cite as: Loan Analytics, "SBA data center loans by metro area, FY2017 to FY2026", https://www.analytics.loan/data-center-feasibility-study#metros-metros.
SBA lenders to data center projects
The lenders with a recorded book of data center loans, ranked; no lender with fewer than 5 loans is listed.
Newtek Bank, National Association leads by count and by dollars
Newtek Bank, National Association heads the lender table on both measures, with 97 data center loans for $60.3 million. The Huntington National Bank is second by count with 67 loans, but its $22.6 million sits well below the leader's dollars and below the $40.0 million that Live Oak Banking Company lent on 26 loans. A borrower whose file goes to either of the two leaders reaches a credit team that has approved data center loans many times and will compare the study with a book of its own.
Live Oak Banking Company: fewest loans, second in dollars
Live Oak Banking Company closes the table by count, with 26 loans, yet its $40.0 million is the second-largest data center book by dollars, more than The Huntington National Bank lent on 67 loans. Its data center loans run large, toward the end of the class where the size bands place 66 7(a) loans and $225.4 million. A study prepared for a lender with that kind of book meets a credit team used to larger data center requests, and it has to carry the full power case, contract-by-contract revenue, a cost budget with the electrical and mechanical plant broken out, and a value for the shell apart from its systems.
The small-ticket books
Several lenders built their data center books from small approvals. TD Bank, National Association made 33 loans for $2.6 million, the smallest dollar book on the table; Northeast Bank (ME) made 44 for $6.3 million; BayFirst National Bank 39 for $7.1 million; and U.S. Bank, National Association 43 for $7.5 million. These books fit what the archived passage on this page reads into the class's small medians: hosting and processing companies borrowing for equipment and working capital. A data center request at one of these lenders that runs far above its usual ticket is likely to be read as an exception, and the study has to justify the size from contracted revenue rather than from the lender's history with the class.
Between the two ends
JPMorgan Chase Bank, National Association (29 loans, $8.4 million), Wells Fargo Bank National Association (34 loans, $8.7 million) and Manufacturers and Traders Trust Company (43 loans, $13.4 million) sit between the small-ticket books and the large ones. No lender on the table prints a charge-off floor: each shows withheld or not reported, the second where a cohort falls short of 30 loans, and neither reading is a zero. The table leaves out any lender under 5 data center loans and shows only the ranked leaders, so a bank absent from it may still hold a small data center book.
Fitting the study to the lender
Read together, the table shows that the lender, as much as the class, decides what a data center study must carry. The two largest dollar books belong to Newtek Bank, National Association, at $60.3 million, and to Live Oak Banking Company, at $40.0 million, so a request in the upper size bands can easily end up with a lender that has seen comparable files and will test the study against them. A small-ticket lender asked to finance a hosting company's own building will look for evidence it has rarely needed in this class. We ask for the lender's name in the scope request for that reason: the depth of the power, contract and valuation work is set by who will read it.
| Lender | Loans | Dollars | Charged off (floor) |
|---|---|---|---|
| Newtek Bank, National Association | 97 | $60.3 million | not reported: fewer than 30 loans in the cohort |
| The Huntington National Bank | 67 | $22.6 million | withheld |
| Northeast Bank (ME) | 44 | $6.3 million | withheld |
| Manufacturers and Traders Trust Company | 43 | $13.4 million | not reported: fewer than 30 loans in the cohort |
| U.S. Bank, National Association | 43 | $7.5 million | not reported: fewer than 30 loans in the cohort |
| BayFirst National Bank | 39 | $7.1 million | not reported: fewer than 30 loans in the cohort |
| Wells Fargo Bank National Association | 34 | $8.7 million | withheld |
| TD Bank, National Association | 33 | $2.6 million | not reported: fewer than 30 loans in the cohort |
| JPMorgan Chase Bank, National Association | 29 | $8.4 million | withheld |
| Live Oak Banking Company | 26 | $40.0 million | not reported: fewer than 30 loans in the cohort |
Source: Loan Analytics' SBA loan record from SBA's 7(a) and 504 loan-level files, approvals FY2017 to FY2026 as of 2026-09-30; rates are floors and need at least 30 loans in their cohort; dollars are not shown where fewer than five loans would reveal a single loan. Download this table (CSV). Cite as: Loan Analytics, "SBA lenders by data center loans approved, FY2017 to FY2026", https://www.analytics.loan/data-center-feasibility-study#lenders-lenders.
Loan sizes, program split and cohort floors for data center loans
How large data center loans run under each SBA program, how they fall into size bands, and how often they were cancelled or charged off.
The quartiles: 7(a) small, 504 larger
The 7(a) quartiles run from $84,400 at the lower quartile through a $200,000 median to $440,500 at the upper quartile. The 504 lower quartile, $323,250, is above the 7(a) median, the 504 median of $591,000 clears even the 7(a) upper quartile, and at its own upper quartile the 504 reaches $1.1 million. The two programs fund different kinds of data center credit, and a lender compares a request with the program it is filed under. A 7(a) request of $591,000, ordinary in 504, sits above the 7(a) upper quartile and will be read as a large loan for the class.
The top band carries the dollars
Under 7(a), 419 loans sit at $150,000 or less, yet together they come to only $32.7 million. The band above it, to $350,000, holds 248 loans and $66.7 million, and the band starting above $350,000 holds 125 loans and $77.2 million. Then the shape turns. The next band holds 53 loans and $74.9 million, and the highest band the table fills holds 66 loans, more than the band below it, carrying $225.4 million, the most of any band. The class has a distinct group of large approvals sitting on a broad base of small ones, and that group, not the median loan, is where a lender's data center exposure concentrates.
Why the large approvals decide the study
A data center or hosting request in the highest band the 7(a) table fills is one of 66 such loans in the FY2017 to FY2026 window. Requests of that size are where a lender most needs evidence that only a study supplies: whether the utility can deliver the power on the date the business plan assumes, whether contracted revenue covers debt service with power cost passed through, and what the shell alone would fetch from a different user once its specialized systems lose value. A request in the smallest band, where 419 loans sit, is more likely an equipment or working capital loan to a firm with a trading history, which a lender can often judge from the firm's own statements.
No approval in the open-ended top band
Neither program made a data center loan in the table's last, open-ended band: the row reads 0 loans and $0 for 7(a) and for 504. Every SBA data center approval in the window fits below that line. A project whose capital need runs above it sits outside this record entirely, as the archived passage on this page notes for a large wholesale hall financed outside SBA. A study for such a project cannot cite this page's class figures as precedent; it rests on the project's own power, contracts and cost.
504 sizes and the floors
The 504 loans gather in the band starting above $350,000, with 23 loans for $14.0 million; 11 sit in the band below it for $2.8 million, 7 in the band above it for $9.8 million, and 6 in the highest band the program fills, where the dollars are withheld. Only 3 504 loans were $150,000 or less, and their dollars are not shown because they rest on fewer than 5 loans. The floors complete the picture: cancellation of 11.2% for 7(a) and 10.9% for 504, close in both programs; a 7(a) charge-off floor of 6.9% against 3.3% charged off within five years; and no 504 loss rate at all, the 504 cohort being too small to print one. Losses arriving after the fifth year point the study at the later years of the loan: the server refresh cycle, contract renewals and the shell's value to another user.
| Program | Loans | Dollars | Lower quartile | Median loan | Upper quartile | Cancelled (floor) | Charged off (floor) | Charged off within 5 years |
|---|---|---|---|---|---|---|---|---|
| SBA 7(a) | 911 | $476.9 million | $84,400 | $200,000 | $440,500 | 11.2% | 6.9% | 3.3% |
| SBA 504 | 50 | $41.6 million | $323,250 | $591,000 | $1.1 million | 10.9% | not reported: fewer than 30 loans in the cohort | not reported: fewer than 30 loans in the cohort |
Source: Loan Analytics' SBA loan record from SBA's 7(a) and 504 loan-level files, approvals FY2017 to FY2026 as of 2026-09-30; rates are floors and need at least 30 loans in their cohort; dollars are not shown where fewer than five loans would reveal a single loan. Download this table (CSV). Cite as: Loan Analytics, "SBA data center loans by program, FY2017 to FY2026", https://www.analytics.loan/data-center-feasibility-study#loan-sizes-programs.
| Loan size | SBA 7(a) loans | SBA 7(a) dollars | SBA 504 loans | SBA 504 dollars |
|---|---|---|---|---|
| $150,000 or less | 419 | $32.7 million | 3 | fewer than 5 loans |
| Over $150,000 to $350,000 | 248 | $66.7 million | 11 | $2.8 million |
| Over $350,000 to $1 million | 125 | $77.2 million | 23 | $14.0 million |
| Over $1 million to $2 million | 53 | $74.9 million | 7 | $9.8 million |
| Over $2 million to $5 million | 66 | $225.4 million | 6 | withheld |
| Over $5 million | 0 | $0 | 0 | $0 |
Source: SBA 7(a) and 504 loan-level FOIA files as of 2026-09-30 (data.sba.gov; gross approvals, cancelled approvals included), aggregated by Loan Analytics; each figure is kept with its source file and retrieval date. Download this table (CSV). Cite as: Loan Analytics, "SBA data center loans by size band, FY2017 to FY2026 approvals", https://www.analytics.loan/data-center-feasibility-study#loan-sizes-size-bands.
Questions buyers ask
Who prepares a data center feasibility study for a bank or SBA lender?
A consultant independent of the borrower, the seller and the builder, so that the lender can rely on the report. Loan Analytics prepares data center and hosting business studies for lenders working under either SBA program, under a USDA Business and Industry guarantee or on conventional terms, and Daniel Smith, MAI signs every one. The work runs from the site's power outward to customers, competition, cost, coverage and value, and places the request in the SBA data center record on this page: 911 7(a) loans for $476.9 million and 50 504 loans for $41.6 million from FY2017 to FY2026.
What will my lender want to see in a data center feasibility study?
Evidence on the points where this class goes wrong. Power comes first: the capacity the utility can deliver to the site, when, at what cost, and the redundancy behind it. Demand comes next: signed contracts or letters of intent, their terms and churn, and the choices those customers have elsewhere. Cost must cover electrical and cooling plant, fire suppression, fiber and security, which together can cost more than the building shell. Debt service coverage is measured against signed revenue only, with the power bill either recharged to customers or hedged, and with money set aside to replace equipment. The appraisal separates the shell from its systems. Against a 6.9% charge-off floor for 7(a) loans in this class, a lender reads each part closely.
How much does a data center feasibility study cost?
The fee figures come from our own 2026 survey, published on this site, of what feasibility providers post. Its first band covers an SBA study of one site where the property is a standard special-purpose type, at $5,000 to $12,500. Its second covers complex assignments, such as phased construction, several sites or an unusual concept, at $10,000 to $25,000. For USDA lenders, a study done to 7 CFR Part 5001 sits in a band of its own, $7,500 to $18,750, with some studies above that. Data centers are not named in the survey. Studying an operating hosting firm at one location fits the first band; a ground-up colocation building whose power arrives in phases fits the survey's description of the second. Each study is quoted after scoping.
What moves the fee for a data center study up or down?
Our 2026 fee survey lists three things that move a study within its band: scope, asset class and program. For a data center, scope turns mostly on whether the business already operates. An operating hosting company with its own statements and customer list is shorter work than a ground-up building with no revenue, where the ramp from first power to filled racks has to be built from contracts and letters of intent. Phased power delivery adds work, because each phase needs its own capacity, cost and occupancy case. A study that must serve USDA as well as SBA is written to the longer USDA standard, which the survey lists at $7,500 to $18,750 or more.
How long does a data center feasibility study take?
We name the timeline when we answer the scope request, measured back from the date the lender needs the study, which the request form asks for. In a data center file the schedule depends less on the writing than on evidence from third parties: the utility's statement of available capacity and its delivery date, signed customer contracts or letters of intent, and current quotes for electrical, cooling and fire suppression equipment. A request that arrives with those documents moves faster than one that has to wait for them; the utility's answer on power is the piece no consultant controls.
Why are SBA data center loans so small when data centers cost so much to build?
Because the class is set by what the borrower does, not by the building it occupies. The code that places a loan here groups data processing and hosting companies with colocation buildings, and SBA's files do not let the two be told apart. Most of the record is therefore lending to operating firms: the 7(a) median is $200,000, its lower quartile $84,400, and 419 of the 911 7(a) loans were $150,000 or less. Larger, building-type borrowing is more likely among the 504 loans, with a $591,000 median, and among the 66 7(a) loans in the highest band the size table fills.
Which SBA program fits a data center or hosting loan better, 7(a) or 504?
This page shows how each program has been used, not which one suits a given project; that depends on the lender and the uses of the loan. From FY2017 to FY2026, 7(a) made 911 data center loans and 504 made 50, at medians of $200,000 and $591,000, and the 504 lower quartile of $323,250 sits above the 7(a) median. Cancellation was close in the two programs, at floors of 11.2% and 10.9%. Only 7(a) has a printed loss floor, 6.9%; 504 has too few loans in its cohort. Our study follows whichever of the two the lender chooses, and while the structure is undecided it can address both.
Does the SBA record include large wholesale or hyperscale data halls?
No. The size table's last, open-ended band holds 0 loans and $0 in both programs, so every data center approval from FY2017 to FY2026 fits below it. Because the class is keyed to what each borrower does for a living, a large wholesale hall financed outside SBA leaves no mark here, and a firm whose server room sits in its own offices is classed under its main business. The largest SBA data center loans sit in the highest band the table fills: 66 7(a) loans carrying $225.4 million, and 6 504 loans whose dollars are withheld. A study for a larger project stands on its own evidence.
How many SBA data center loans have been charged off?
Of the 7(a) data center loans that disbursed from FY2017 through FY2021, 6.9% is the charge-off floor to date, and 3.3% is the part of it that came in the first five years after funding, so losses in this class kept arriving after the fifth year. These are floors: they count losses recorded so far, and later losses can only raise them. For 504 no floor is printed, because fewer than 30 loans of that program fall in the cohort; a blank there means unknown, not zero. California, at 6.0% across both programs, is the only state with a printed floor. The study answers this record with a downside case on the borrower's contracts: the loss of the largest customer, a rise in power cost and an early equipment refresh.
How often is an approved data center loan cancelled before it funds?
At least 11.2% of 7(a) data center approvals and 10.9% of 504 approvals made in FY2017 to FY2024 were cancelled. The state floors spread widely: 17.4% in Texas, 14.9% in Florida, 8.2% in New York, 5.7% in California and 3.3% in Ohio. A cancellation means the lender approved the loan and it never funded. The study tests in advance the conditions a data center loan has to meet to close: a power delivery date the utility will confirm, equipment prices that hold, customer commitments that last until funding, and an equity contribution that matches the cost budget.
Which states see the most SBA data center lending?
California ranks first, 113 loans for $64.9 million; Texas second, 103 for $61.7 million; then New York, 66 for $17.5 million, and Florida, 65 for $42.6 million. Illinois and Ohio follow at 39 each, then Colorado at 32 and Maryland and New Jersey at 31. State medians range from $275,000 in Florida down to $102,800 in Minnesota. From Missouri, with 24 loans, down the table, states lack the loans to support a median or a floor, and for a project in one of them the study measures the request against the national quartiles and the project's own evidence.
Which banks make SBA loans to data center and hosting businesses?
From FY2017 to FY2026 the most active were Newtek Bank, National Association with 97 loans and $60.3 million; The Huntington National Bank with 67 and $22.6 million; Northeast Bank (ME) with 44 and $6.3 million; U.S. Bank, National Association, 43; Manufacturers and Traders Trust Company, also 43; BayFirst National Bank with 39; Wells Fargo Bank National Association with 34; TD Bank, National Association with 33; JPMorgan Chase Bank, National Association with 29; and Live Oak Banking Company with 26 loans and $40.0 million. No lender with fewer than 5 loans is listed. The spread in dollars shows which lenders take the larger requests.
Which metro areas see the most SBA data center loans?
New York-Newark-Jersey City leads with 67 loans and $24.1 million, then Dallas-Fort Worth-Arlington with 45, Chicago-Naperville-Elgin with 42, Washington-Arlington-Alexandria with 39, Los Angeles-Long Beach-Anaheim with 37, Miami-Fort Lauderdale-West Palm Beach with 30 and Austin-Round Rock-San Marcos with 27. Los Angeles carries the most dollars, $27.5 million, and Austin the highest median, $350,000, against $150,000 in the New York metro. No metro prints a charge-off floor, so a study in any of them uses the metro's loan sizes as the local comparison and the national floor for the downside.
Is SBA lending to data center businesses rising or falling?
It surged, then eased. 7(a) approvals reached 151 loans and $72.8 million in FY2024, the most loans and the largest printed dollar total in the class record, and 145 loans in FY2025, before falling to 87 in FY2026. The 504 side stayed thin throughout, with 1 approval in FY2026. Because the 6.9% charge-off floor covers only loans disbursed in FY2017 to FY2021, the surge years carry no loss record yet, and a lender reading a request now has no class history for that vintage to lean on, which places more weight on the borrower's own contracts and on the study that tests them.
How does the study show demand for a new colocation or hosting facility?
From customers, not from the size of the industry. The study lists the contracts and letters of intent in hand, with term, power draw, price and renewal dates, and the churn the operator has seen on contracts it already holds. It maps the other places those customers could go, both colocation space and cloud services, and what moving would cost them. It then builds the ramp from first power to full occupancy out of those commitments, and tests coverage at each stage. A lender reading a request in a class whose 7(a) charge-off floor is 6.9% looks for contracted revenue first.
How is a data center building valued for a loan?
The lender needs the shell valued on its own, for an owner who would not need the power and cooling plant, with the electrical, mechanical and cooling systems treated apart, since equipment of that kind loses value fast. The study's cost and market work is built to line up with that split: what the site and shell are worth to another user, and what the systems add while they remain current. For a 504 data center loan, where the median is $591,000 and the upper quartile $1.1 million, the shell value is the part of the collateral that outlasts the equipment.
Can a hosting company use a feasibility study to buy or build its own facility?
That is the rarer request in this class, and the one a lender examines most closely. The SBA record here is mostly lending to operating hosting and processing firms, at a 7(a) median of $200,000, while a company moving its racks into a building it owns brings a real estate case on top of an operating one, closer in kind to the larger 504 loans, whose median is $591,000. The study shows whether the building pays for itself under the firm's contracts: power capacity and redundancy at the new site, the cost of moving customers without losing them, the electrical and cooling fit-out, and coverage on contracted revenue, with the shell valued for an owner who would not run a data center.
My state shows no rate on this page. What does the study rely on instead?
This page prints a rate only for a cohort of 30 loans or more. California alone prints a state charge-off floor, and California, Texas, New York, Florida and Ohio print cancellation floors; from Missouri down, no state carries a median or any floor. A blank or not reported cell is never a zero and never a clean record. Where the state is thin, the study uses the national 7(a) floors of 11.2% and 6.9%, the metro row where one exists, and evidence gathered for the site: the utility's capacity statement, the colocation and cloud alternatives in the market, and the customers' own contracts.
Can the study be prepared for a USDA Business and Industry lender?
Yes. Loan Analytics prepares data center studies for USDA Business and Industry lenders as well as for SBA and conventional ones. The tables on this page are SBA records only and carry no USDA loans, so a study for a B&I lender draws its comparisons from the site, the power and the customers rather than from a USDA class history. For a USDA study, our 2026 fee survey records $7,500 to $18,750 and higher where the work is done to 7 CFR Part 5001. Tell us in the scope request which program the lender will use.
Loan Analytics prepares the feasibility study a lender reads before it commits: for a data center project anywhere in the United States, financed under SBA 7(a), SBA 504, USDA Business and Industry or a conventional bank loan. Tell us the project and the lender's deadline, and we reply with scope, timeline and fee.
Daniel Smith, MAI
Request Scope & Timeline
Start with the data.
Contact Us