A cheaper parcel can become the more expensive site surprisingly quickly.
For an illustrative data center with 20 MW of IT capacity, a one-cent difference per kilowatt-hour adds $1.971 million to annual energy expense at 1.25 PUE and 90% utilization. That is before demand charges, connection costs and other tariff terms. These are model assumptions, not a Google or utility quote.
I have been researching what Google’s confirmed Texas developments tell us about buying data-center land. The useful detail is in the arrangements behind the announcement: land control, interconnection, power procurement and the conditions under which a facility can actually operate.
Google’s February 24 announcement with AES in Wilbarger County describes land and interconnection agreements and 20-year power purchase agreements. Its June 4 Meitner announcement describes a separate development in Gray and Roberts counties with Intersect. These are distinct projects. They do not establish a Google commitment in Lea County, New Mexico. The Lea County headline sent to me remains unconfirmed in the source material I reviewed.
The land calculation deserves the same care. A $3 million property with 100 usable acres costs $30,000 per usable acre. If environmental constraints, access and setbacks leave only 60 usable acres, that becomes $50,000. The purchase price has not changed. The development footprint has.
Timing matters too. Six months of additional simple interest on $10 million of constant drawn debt at 10% annually is $500,000. That excludes fees, taxes and lost revenue. It is an illustration of carrying cost, not a prediction about any specific project.
I own land in Louisiana, which is part of why I am interested in these questions. My next research piece will examine Meta’s Hyperion project and distinguish the financing arrangements, approved power infrastructure and proposed expansions. A nearby project is a reason to investigate a parcel, not proof that the parcel can obtain service.
I am building Liquid Labor to make this research easier to inspect. You can compare projects, follow source links and see where the evidence is still missing:
https://www.liquid-labor.com/tracker/
If you are evaluating a site now, what has changed your decision most: the delivery date for power, the cost of upgrades, or the usable acreage? I would be interested in an example you can share without disclosing confidential terms.
Sources
Google and AES, Wilbarger County, February 24, 2026:
https://storage.googleapis.com/gweb-uniblog-publish-prod/documents/wilbarger_county_PR.pdf
Google, Meitner Energy Center, June 4, 2026:
https://blog.google/innovation-and-ai/infrastructure-and-cloud/global-network/meitner-energy-center/
Research and calculations: Uwe Cerron, Liquid Labor. Illustrative comparisons do not value a specific property or establish available utility capacity.