the farmland paradox

published: December 22, 2025 โ€ข

overview

america has over 1 billion square feet of vacant industrial and commercial space. yet hyperscale data center operators are converting productive farmland into data centers at an accelerating rate. this appears economically inefficient and politically controversial.

the explanation is simple: it is not about land. it is about power infrastructure.

hyperscale data centers require 500-1,000mw of electrical capacity. this level of demand cannot be served by urban distribution grids designed for residential and light commercial loads. the topology of the electrical grid, combined with interconnection queue delays and brownfield remediation complexity, makes rural greenfield development faster and more economical than urban brownfield redevelopment.

the core question

why build on farmland when vacant industrial sites exist?

the united states has extensive brownfield inventory:

  • 1+ billion sq ft of vacant commercial/industrial space
  • 450,000+ brownfield sites identified by epa
  • existing electrical service, water, sewer, transportation infrastructure
  • locations in or near demand centers
  • no community opposition to industrial use

productive farmland conversion generates:

  • community opposition
  • environmental review delays
  • political controversy
  • loss of agricultural production

yet greenfield farmland remains the dominant site type for hyperscale development.

the answer in brief

power infrastructure determines location. land cost, availability, and logistics are secondary.

urban and suburban electrical grids are distribution networks designed to deliver 120-480v power to buildings. they cannot deliver 500-1,000mw at transmission voltages (115-765kv) required by hyperscale data centers.

rural locations near transmission substations or generation facilities can interconnect at transmission voltage with shorter approval timelines. brownfield sites require distribution-to-transmission upgrades that add 2-4 years and $30-80m in costs.

power interconnection is the binding constraint. everything else follows from this constraint.

summary comparison table

factorgreenfield (farmland)brownfield (industrial)
land cost$2,000-$10,000/acre$50,000-$200,000/acre
power interconnection18-36 months48-72+ months
build timeline18-24 months36-48 months
remediation cost$0$12-30m+
total capex (500mw)$3.5-4.5b$3.8-5.0b
time to revenue24-36 months48-60 months
unlevered irr18-19%16-17%

the three structural forces

1. physics: grid topology

electrical grids have hierarchical architecture:

  • generation (power plants)
  • transmission (high voltage, 115-765kv, long distance)
  • distribution (medium voltage, 12-35kv, local delivery)
  • service (low voltage, 120-480v, end users)

hyperscale data centers require transmission-level interconnection. urban brownfield sites are served by distribution networks. upgrading distribution to transmission requires:

  • new transmission line construction
  • substation construction or expansion
  • utility service territory coordination
  • multi-year regulatory approval

rural greenfield sites near existing transmission infrastructure bypass these requirements.

2. economics: development costs

brownfield sites carry additional costs:

  • environmental remediation: $12-30m+
  • demolition and clearance: $5-15m
  • title and lien resolution: $2-8m
  • extended permitting and legal: $3-10m
  • power infrastructure upgrades: $30-80m

total brownfield premium: $52-143m

greenfield sites avoid these costs. land cost differential ($8-15m for 150-200 acres) is negligible relative to total project capex ($3.5-4.5b for 500mw).

3. policy: incentive structure

state and local economic development incentives do not offset brownfield costs:

  • property tax abatements: available for both greenfield and brownfield
  • sales tax exemptions: available for both
  • job creation credits: available for both
  • energy efficiency incentives: available for both

no major incentive programs specifically favor brownfield redevelopment for data centers. epa brownfield grants are limited to $500k-$1m, insufficient for hyperscale remediation costs.

policy framework is location-agnostic, removing any corrective mechanism for brownfield preference.

detailed analysis

the bottom line

land is cheap. power is expensive. time is priceless.

a hyperscale data center operator evaluating sites compares:

  • greenfield: 24-36 months to revenue, 18-19% unlevered irr
  • brownfield: 48-60 months to revenue, 16-17% unlevered irr

the decision is rational given current infrastructure topology and policy framework.

changing this outcome requires changing the constraints:

  • accelerated interconnection for brownfield sites
  • transmission infrastructure investment in urban areas
  • targeted incentives that offset brownfield premiums

absent these changes, farmland conversion will continue.

on this page