dadams/data-centers
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add data center demographic, RUCA, and energy capacity analysis

Browse Source 
Adds three coordinated changes:

- Request nameplate, summer, and winter capacity from the EIA
  operating-generator-capacity endpoint and project them as typed columns
  on energy_eia_operating_generator_capacity_flat. The original ingest
  only pulled latitude and longitude, leaving the flat table with no MW
  values despite its name.
- New cluster_analysis.ipynb joins master_data_centers to ACS-2024
  demographics, USDA RUCA-2020 codes (loaded from new/), and EIA
  generation capacity within 50 km of each site.
- Summary doc consolidates the headline findings: DC tracts skew higher
  income / more educated / more racially diverse than US average, the
  metro over-index is only 1.11x, the non-metro tail is dominated by
  hyperscalers in the Columbia River corridor (OR+WA = 66% of non-metro
  DCs), and Microsoft co-locates with Palo Verde Nuclear in Goodyear AZ.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
David Adams
2026-05-18 08:14:57 -07:00
parent e7f23a87b2
commit eccfbdbad9
3 changed files with 1073 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

841
cluster_analysis.ipynb Normal file
View File

@@ -0,0 +1,841 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "0",
"metadata": {},
"source": [
"# Clustering Analysis"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"from pathlib import Path\n",
"\n",
"import pandas as pd\n",
"import psycopg2\n",
"\n",
"\n",
"def load_env_file(env_path: str = '.env') -> None:\n",
" p = Path(env_path)\n",
" if not p.exists():\n",
" print(f'No {env_path} file found in {Path.cwd()}')\n",
" return\n",
" loaded = 0\n",
" for raw_line in p.read_text(encoding='utf-8').splitlines():\n",
" line = raw_line.strip()\n",
" if not line or line.startswith('#') or '=' not in line:\n",
" continue\n",
" key, value = line.split('=', 1)\n",
" key = key.strip()\n",
" value = value.strip().strip('\"').strip(\"'\")\n",
" if key and key not in os.environ:\n",
" os.environ[key] = value\n",
" loaded += 1\n",
" print(f'Loaded {loaded} env var(s) from {env_path}')\n",
"\n",
"\n",
"def require_env(keys):\n",
" missing = [k for k in keys if not os.getenv(k)]\n",
" if missing:\n",
" raise EnvironmentError(\n",
" 'Missing required env vars: ' + ', '.join(missing) +\n",
" '.\\nSet them in this notebook, or add them to a .env file.'\n",
" )\n",
"\n",
"\n",
"load_env_file('.env')\n",
"\n",
"required_keys = ['PGWEB_HOST', 'PGWEB_PORT', 'PGWEB_USER', 'PGWEB_PASSWORD']\n",
"require_env(required_keys)\n",
"\n",
"DB_NAME = os.getenv('PGDATABASE', 'data_centers')\n",
"\n",
"\n",
"def get_conn():\n",
" return psycopg2.connect(\n",
" host=os.environ['PGWEB_HOST'],\n",
" port=os.environ['PGWEB_PORT'],\n",
" user=os.environ['PGWEB_USER'],\n",
" password=os.environ['PGWEB_PASSWORD'],\n",
" dbname='data_centers',\n",
" )\n",
"\n",
"\n",
"with get_conn() as conn:\n",
" with conn.cursor() as cur:\n",
" cur.execute('select current_database(), current_user, version()')\n",
" db, usr, ver = cur.fetchone()\n",
" print('Connected to DB:', db)\n",
" print('As user:', usr)\n",
" print('Postgres:', ver.split(',')[0])\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "2",
"metadata": {},
"outputs": [],
"source": [
"# List tables in the database (user schemas only, excluding system + PostGIS internals).\n",
"TABLES_SQL = \"\"\"\n",
"select\n",
" table_schema,\n",
" table_name,\n",
" table_type\n",
"from information_schema.tables\n",
"where table_schema not in ('pg_catalog', 'information_schema', 'tiger', 'tiger_data', 'topology')\n",
"order by table_schema, table_name\n",
"\"\"\"\n",
"\n",
"with get_conn() as conn:\n",
" tables_df = pd.read_sql(TABLES_SQL, conn)\n",
"\n",
"print(f'{len(tables_df):,} tables/views found')\n",
"tables_df"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3",
"metadata": {},
"outputs": [],
"source": [
"# Inspect columns for the tables we want to join.\n",
"INSPECT_TABLES = [\n",
" 'master_data_centers',\n",
" 'master_data_center_spatial_clusters',\n",
" 'data_center_census_tracts_2024',\n",
" '_dc_census_tract_acs_2024',\n",
" 'energy_eia_operating_generator_capacity_flat',\n",
"]\n",
"\n",
"COLS_SQL = \"\"\"\n",
"select table_name, column_name, data_type\n",
"from information_schema.columns\n",
"where table_schema = 'public' and table_name = any(%s)\n",
"order by table_name, ordinal_position\n",
"\"\"\"\n",
"\n",
"with get_conn() as conn:\n",
" cols_df = pd.read_sql(COLS_SQL, conn, params=(INSPECT_TABLES,))\n",
"\n",
"for t in INSPECT_TABLES:\n",
" sub = cols_df[cols_df['table_name'] == t]\n",
" print(f'\\n=== {t} ({len(sub)} cols) ===')\n",
" print(sub[['column_name', 'data_type']].to_string(index=False))\n"
]
},
{
"cell_type": "markdown",
"id": "4",
"metadata": {},
"source": [
"## Ingest RUCA codes\n",
"\n",
"USDA Rural-Urban Commuting Area (RUCA) codes classify each census tract on a 1–10 scale from \"Metropolitan area core\" (1) to \"Rural area\" (10), based on population density and commuting flows. Source file: `new/RUCA-codes-2020-tract.csv` (~85K tracts).\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "5",
"metadata": {},
"outputs": [],
"source": [
"# Push RUCA codes CSV -> public.ruca_codes_2020_tract (idempotent: drops + recreates).\n",
"from psycopg2.extras import execute_values\n",
"\n",
"RUCA_CSV = Path('new/RUCA-codes-2020-tract.csv')\n",
"RUCA_TABLE = 'public.ruca_codes_2020_tract'\n",
"\n",
"# Map source CSV columns -> snake_case DB columns.\n",
"COL_MAP = {\n",
" 'TractFIPS23': 'tract_fips_23',\n",
" 'CountyFIPS23': 'county_fips_23',\n",
" 'CountyCode23': 'county_code_23',\n",
" 'CountyName23': 'county_name_23',\n",
" 'TractFIPS20': 'tract_fips_20',\n",
" 'TractCode20': 'tract_code_20',\n",
" 'TractName20': 'tract_name_20',\n",
" 'CountyFIPS20': 'county_fips_20',\n",
" 'CountyCode20': 'county_code_20',\n",
" 'CountyName20': 'county_name_20',\n",
" 'StateFIPS20': 'state_fips_20',\n",
" 'StateName20': 'state_name_20',\n",
" 'UrbanAreaCode20': 'urban_area_code_20',\n",
" 'UrbanAreaName20': 'urban_area_name_20',\n",
" 'UrbanCore': 'urban_core',\n",
" 'UrbanCoreType': 'urban_core_type',\n",
" 'PrimaryRUCA': 'primary_ruca',\n",
" 'PrimaryRUCADescription': 'primary_ruca_description',\n",
" 'PrimaryDestinationCode': 'primary_destination_code',\n",
" 'PrimaryDestinationName': 'primary_destination_name',\n",
" 'SecondaryRUCA': 'secondary_ruca',\n",
" 'SecondaryRUCADescription': 'secondary_ruca_description',\n",
" 'SecondaryDestinationCode': 'secondary_destination_code',\n",
" 'SecondaryDestinationName': 'secondary_destination_name',\n",
" 'Population': 'population',\n",
" 'LandArea': 'land_area',\n",
" 'PopDensity': 'pop_density',\n",
"}\n",
"\n",
"# File is Latin-1 (has bytes like 0xf1 = ñ from Spanish place names).\n",
"fips_str_cols = [c for c in COL_MAP if 'FIPS' in c or 'Code' in c]\n",
"ruca_df = pd.read_csv(\n",
" RUCA_CSV,\n",
" dtype={c: str for c in fips_str_cols},\n",
" encoding='latin-1',\n",
")\n",
"ruca_df = ruca_df.rename(columns=COL_MAP)\n",
"print(f'CSV rows: {len(ruca_df):,} cols: {ruca_df.shape[1]}')\n",
"\n",
"# PK is tract_fips_20 (always populated). Some tracts that existed in 2020 are gone\n",
"# in 2023 (water-only tracts, dissolves), so tract_fips_23 can be null.\n",
"DDL = f\"\"\"\n",
"drop table if exists {RUCA_TABLE};\n",
"create table {RUCA_TABLE} (\n",
" tract_fips_23 text,\n",
" county_fips_23 text,\n",
" county_code_23 text,\n",
" county_name_23 text,\n",
" tract_fips_20 text primary key,\n",
" tract_code_20 text,\n",
" tract_name_20 text,\n",
" county_fips_20 text,\n",
" county_code_20 text,\n",
" county_name_20 text,\n",
" state_fips_20 text,\n",
" state_name_20 text,\n",
" urban_area_code_20 text,\n",
" urban_area_name_20 text,\n",
" urban_core smallint,\n",
" urban_core_type text,\n",
" primary_ruca smallint,\n",
" primary_ruca_description text,\n",
" primary_destination_code text,\n",
" primary_destination_name text,\n",
" secondary_ruca text,\n",
" secondary_ruca_description text,\n",
" secondary_destination_code text,\n",
" secondary_destination_name text,\n",
" population integer,\n",
" land_area double precision,\n",
" pop_density double precision\n",
");\n",
"create index ruca_codes_2020_tract_state_idx on {RUCA_TABLE} (state_fips_20);\n",
"create index ruca_codes_2020_tract_primary_ruca_idx on {RUCA_TABLE} (primary_ruca);\n",
"create index ruca_codes_2020_tract_fips_23_idx on {RUCA_TABLE} (tract_fips_23);\n",
"\"\"\"\n",
"\n",
"cols = list(COL_MAP.values())\n",
"records = [tuple(None if pd.isna(v) else v for v in row) for row in ruca_df[cols].itertuples(index=False, name=None)]\n",
"\n",
"with get_conn() as conn:\n",
" with conn.cursor() as cur:\n",
" cur.execute(DDL)\n",
" execute_values(\n",
" cur,\n",
" f\"insert into {RUCA_TABLE} ({', '.join(cols)}) values %s\",\n",
" records,\n",
" page_size=2000,\n",
" )\n",
" cur.execute(f'select count(*) from {RUCA_TABLE}')\n",
" print(f'{RUCA_TABLE}: {cur.fetchone()[0]:,} rows')\n",
" cur.execute(f\"\"\"\n",
" select primary_ruca, count(*) as n\n",
" from {RUCA_TABLE}\n",
" group by 1 order by 1\n",
" \"\"\")\n",
" print('\\nRUCA distribution (all US tracts):')\n",
" for ruca, n in cur.fetchall():\n",
" print(f' {ruca}: {n:>6,}')\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "6",
"metadata": {},
"outputs": [],
"source": [
"# Sanity-check EIA generator coordinate ranges.\n",
"# Prior note: EIA source data had a longitude sign error. Verify before spatial joining.\n",
"EIA_COORD_SQL = \"\"\"\n",
"select\n",
" min(longitude) as lon_min, max(longitude) as lon_max,\n",
" min(latitude) as lat_min, max(latitude) as lat_max,\n",
" count(*) filter (where longitude > 0) as pos_lon_rows,\n",
" count(*) filter (where longitude < 0) as neg_lon_rows,\n",
" count(*) as total_rows\n",
"from public.energy_eia_operating_generator_capacity_flat\n",
"where longitude is not null and latitude is not null\n",
"\"\"\"\n",
"\n",
"with get_conn() as conn:\n",
" eia_coord_df = pd.read_sql(EIA_COORD_SQL, conn)\n",
"\n",
"print(eia_coord_df.T)\n",
"# For US plants we expect longitude in roughly [-180, -65]. If pos_lon_rows is large,\n",
"# the sign-flip correction is still needed when spatial-joining.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "7",
"metadata": {},
"outputs": [],
"source": [
"# Build the joined analysis dataset.\n",
"#\n",
"# Joins:\n",
"# master_data_centers (m)\n",
"# LEFT JOIN master_data_center_spatial_clusters (c) ON master_id\n",
"# LEFT JOIN _dc_census_tract_acs_2024 (acs) ON m.geoid = acs.geoid\n",
"# LEFT JOIN ruca_codes_2020_tract (ruca) ON m.geoid = ruca.tract_fips_20\n",
"# LEFT JOIN (EIA operating generators within RADIUS_KM, latest period) aggregated per DC\n",
"#\n",
"# Energy aggregation: latest period, status='OP', sum of nameplate_capacity_mw\n",
"# (and counts) within RADIUS_KM, broken out by fuel.\n",
"\n",
"RADIUS_KM = 50\n",
"\n",
"JOIN_SQL = f\"\"\"\n",
"with latest_period as (\n",
" select max(period) as period\n",
" from public.energy_eia_operating_generator_capacity_flat\n",
"),\n",
"eia_latest as (\n",
" select e.plant_id, e.generator_id, e.energy_source_code,\n",
" e.nameplate_capacity_mw, e.geom\n",
" from public.energy_eia_operating_generator_capacity_flat e\n",
" join latest_period lp on e.period = lp.period\n",
" where e.status = 'OP' and e.geom is not null\n",
"),\n",
"energy_nearby as (\n",
" select\n",
" m.master_id,\n",
" count(*) as eia_gen_count,\n",
" count(distinct plant_id) as eia_plant_count,\n",
" sum(nameplate_capacity_mw) as eia_capacity_mw,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'NG') as eia_capacity_ng,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code in ('BIT','SUB','LIG','RC','ANT')) as eia_capacity_coal,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'NUC') as eia_capacity_nuclear,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'SUN') as eia_capacity_solar,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'WND') as eia_capacity_wind,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'WAT') as eia_capacity_hydro,\n",
" sum(nameplate_capacity_mw) filter (where energy_source_code = 'GEO') as eia_capacity_geothermal\n",
" from public.master_data_centers m\n",
" join eia_latest e\n",
" on st_dwithin(m.geom::geography, e.geom::geography, {RADIUS_KM} * 1000)\n",
" where m.geom is not null\n",
" group by m.master_id\n",
")\n",
"select\n",
" m.master_id, m.name, m.operator, m.city, m.state, m.country,\n",
" m.power_mw, m.area_sqft, m.longitude, m.latitude, m.geoid,\n",
" c.cluster_id, c.is_noise, c.nearest_neighbor_km,\n",
" acs.population, acs.median_age, acs.households, acs.avg_household_size,\n",
" acs.median_household_income, acs.per_capita_income,\n",
" acs.poverty_rate, acs.unemployment_rate,\n",
" acs.bachelor_or_higher_pct, acs.broadband_subscription_pct,\n",
" acs.hispanic_latino_pct, acs.hispanic_latino_population,\n",
" acs.non_hispanic_white_pct, acs.non_hispanic_white_population,\n",
" acs.non_hispanic_black_pct, acs.non_hispanic_black_population,\n",
" acs.non_hispanic_asian_pct, acs.non_hispanic_asian_population,\n",
" acs.primary_industry, acs.primary_industry_pct,\n",
" ruca.primary_ruca, ruca.primary_ruca_description,\n",
" ruca.urban_core, ruca.urban_core_type,\n",
" ruca.pop_density as tract_pop_density,\n",
" ruca.land_area as tract_land_area_sqmi,\n",
" coalesce(en.eia_gen_count, 0) as eia_gen_count,\n",
" coalesce(en.eia_plant_count, 0) as eia_plant_count,\n",
" coalesce(en.eia_capacity_mw, 0) as eia_capacity_mw,\n",
" coalesce(en.eia_capacity_ng, 0) as eia_capacity_ng,\n",
" coalesce(en.eia_capacity_coal, 0) as eia_capacity_coal,\n",
" coalesce(en.eia_capacity_nuclear, 0) as eia_capacity_nuclear,\n",
" coalesce(en.eia_capacity_solar, 0) as eia_capacity_solar,\n",
" coalesce(en.eia_capacity_wind, 0) as eia_capacity_wind,\n",
" coalesce(en.eia_capacity_hydro, 0) as eia_capacity_hydro,\n",
" coalesce(en.eia_capacity_geothermal, 0) as eia_capacity_geothermal\n",
"from public.master_data_centers m\n",
"left join public.master_data_center_spatial_clusters c on c.master_id = m.master_id\n",
"left join public._dc_census_tract_acs_2024 acs on acs.geoid = m.geoid\n",
"left join public.ruca_codes_2020_tract ruca on ruca.tract_fips_20 = m.geoid\n",
"left join energy_nearby en on en.master_id = m.master_id\n",
"\"\"\"\n",
"\n",
"with get_conn() as conn:\n",
" joined_df = pd.read_sql(JOIN_SQL, conn)\n",
"\n",
"print(f'rows: {len(joined_df):,} cols: {joined_df.shape[1]}')\n",
"print('non-null geoid: ', joined_df['geoid'].notna().sum())\n",
"print('non-null cluster_id: ', joined_df['cluster_id'].notna().sum())\n",
"print('non-null primary_ruca: ', joined_df['primary_ruca'].notna().sum())\n",
"print('DCs with >=1 nearby gen: ', (joined_df['eia_gen_count'] > 0).sum())\n",
"print(f\"median nearby capacity: {joined_df['eia_capacity_mw'].median():,.0f} MW\")\n",
"print(f\" 90th percentile: {joined_df['eia_capacity_mw'].quantile(0.9):,.0f} MW\")\n",
"print(f\" max: {joined_df['eia_capacity_mw'].max():,.0f} MW\")\n",
"joined_df.head()\n"
]
},
{
"cell_type": "markdown",
"id": "8",
"metadata": {},
"source": [
"## Quick demographic analysis\n",
"\n",
"The joined dataset has one row per data center, enriched with the demographics of its containing census tract. Note that multiple DCs can share a tract, so tract-level stats are weighted by DC count in these summaries (i.e. \"the average DC sits in a tract with...\" rather than \"the average DC-hosting tract has...\").\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9",
"metadata": {},
"outputs": [],
"source": [
"# Top-line demographic profile of the average DC's containing tract.\n",
"demo_cols = [\n",
" 'population', 'median_age', 'avg_household_size',\n",
" 'median_household_income', 'per_capita_income',\n",
" 'poverty_rate', 'unemployment_rate',\n",
" 'bachelor_or_higher_pct', 'broadband_subscription_pct',\n",
" 'hispanic_latino_pct', 'non_hispanic_white_pct',\n",
" 'non_hispanic_black_pct', 'non_hispanic_asian_pct',\n",
"]\n",
"demo_cols = [c for c in demo_cols if c in joined_df.columns]\n",
"\n",
"summary = joined_df[demo_cols].agg(['count', 'mean', 'median', 'std', 'min', 'max']).round(2).T\n",
"summary.columns = ['n', 'mean', 'median', 'std', 'min', 'max']\n",
"\n",
"# US national benchmarks (ACS 5-yr ~2024) for context\n",
"benchmarks = {\n",
" 'median_household_income': 78_538,\n",
" 'per_capita_income': 43_313,\n",
" 'poverty_rate': 12.4,\n",
" 'unemployment_rate': 5.4,\n",
" 'bachelor_or_higher_pct': 35.0,\n",
" 'broadband_subscription_pct': 89.0,\n",
" 'hispanic_latino_pct': 19.5,\n",
" 'non_hispanic_white_pct': 58.4,\n",
" 'non_hispanic_black_pct': 12.1,\n",
" 'non_hispanic_asian_pct': 6.4,\n",
"}\n",
"summary['us_avg'] = pd.Series(benchmarks).reindex(summary.index)\n",
"summary['vs_us'] = (summary['mean'] - summary['us_avg']).round(2)\n",
"summary\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "10",
"metadata": {},
"outputs": [],
"source": [
"# Geographic concentration: where are the data centers, and what do those places look like?\n",
"state_summary = (\n",
" joined_df.groupby('state', dropna=False)\n",
" .agg(\n",
" dc_count=('master_id', 'count'),\n",
" avg_power_mw=('power_mw', 'mean'),\n",
" total_power_mw=('power_mw', 'sum'),\n",
" median_hh_income=('median_household_income', 'median'),\n",
" median_poverty=('poverty_rate', 'median'),\n",
" median_bachelor_pct=('bachelor_or_higher_pct', 'median'),\n",
" median_broadband_pct=('broadband_subscription_pct', 'median'),\n",
" median_pct_white=('non_hispanic_white_pct', 'median'),\n",
" median_pct_hispanic=('hispanic_latino_pct', 'median'),\n",
" median_pct_black=('non_hispanic_black_pct', 'median'),\n",
" )\n",
" .sort_values('dc_count', ascending=False)\n",
" .round(1)\n",
")\n",
"print(f'{joined_df[\"state\"].nunique()} states/territories represented')\n",
"state_summary.head(15)\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "11",
"metadata": {},
"outputs": [],
"source": [
"# Cluster vs. non-cluster: do DCs in spatial clusters sit in different demographic settings\n",
"# than isolated ones? (cluster_id is null/is_noise=True for unclustered DCs.)\n",
"joined_df['in_cluster'] = joined_df['cluster_id'].notna() & (joined_df['is_noise'] != True)\n",
"\n",
"compare_cols = [\n",
" 'median_household_income', 'per_capita_income',\n",
" 'poverty_rate', 'bachelor_or_higher_pct', 'broadband_subscription_pct',\n",
" 'non_hispanic_white_pct', 'hispanic_latino_pct', 'non_hispanic_black_pct',\n",
" 'population', 'eia_gen_count',\n",
"]\n",
"compare_cols = [c for c in compare_cols if c in joined_df.columns]\n",
"\n",
"cluster_compare = (\n",
" joined_df.groupby('in_cluster')[compare_cols]\n",
" .median()\n",
" .round(1)\n",
" .T\n",
" .rename(columns={False: 'isolated', True: 'in_cluster'})\n",
")\n",
"cluster_compare['delta'] = (cluster_compare['in_cluster'] - cluster_compare['isolated']).round(1)\n",
"print(f\"DCs in a cluster: {joined_df['in_cluster'].sum():,} isolated: {(~joined_df['in_cluster']).sum():,}\")\n",
"cluster_compare\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "12",
"metadata": {},
"outputs": [],
"source": [
"# Quick visual sweep: distribution of key demographic features for DC tracts,\n",
"# with US-average reference lines for context.\n",
"import matplotlib.pyplot as plt\n",
"\n",
"panels = [\n",
" ('median_household_income', 78_538, 'Median household income (USD)'),\n",
" ('poverty_rate', 12.4, 'Poverty rate (%)'),\n",
" ('bachelor_or_higher_pct', 35.0, \"Bachelor's degree or higher (%)\"),\n",
" ('broadband_subscription_pct', 89.0, 'Broadband subscription (%)'),\n",
" ('non_hispanic_white_pct', 58.4, 'Non-Hispanic white (%)'),\n",
" ('hispanic_latino_pct', 19.5, 'Hispanic/Latino (%)'),\n",
"]\n",
"panels = [(c, b, lab) for c, b, lab in panels if c in joined_df.columns]\n",
"\n",
"fig, axes = plt.subplots(2, 3, figsize=(15, 8))\n",
"for ax, (col, bench, label) in zip(axes.ravel(), panels):\n",
" s = joined_df[col].dropna()\n",
" ax.hist(s, bins=40, color='steelblue', edgecolor='white', alpha=0.85)\n",
" ax.axvline(s.median(), color='darkorange', linestyle='-', lw=2, label=f'DC median = {s.median():.1f}')\n",
" ax.axvline(bench, color='firebrick', linestyle='--', lw=2, label=f'US avg = {bench}')\n",
" ax.set_title(label)\n",
" ax.legend(fontsize=8)\n",
"\n",
"plt.tight_layout()\n",
"plt.show()\n"
]
},
{
"cell_type": "markdown",
"id": "13",
"metadata": {},
"source": [
"## RUCA (urban / rural) analysis\n",
"\n",
"RUCA primary code key:\n",
"- **1**: Metropolitan area core\n",
"- **2**: Metropolitan area high commuting\n",
"- **3**: Metropolitan area low commuting\n",
"- **4–6**: Micropolitan area (small city + commuting tracts)\n",
"- **7–9**: Small town (core + commuting tracts)\n",
"- **10**: Rural area\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "14",
"metadata": {},
"outputs": [],
"source": [
"# DC distribution across RUCA codes vs. the national baseline of all US tracts.\n",
"ruca_buckets = {\n",
" 1: 'Metro core', 2: 'Metro high-commute', 3: 'Metro low-commute',\n",
" 4: 'Micro core', 5: 'Micro high-commute', 6: 'Micro low-commute',\n",
" 7: 'Small town core', 8: 'Small town high-commute', 9: 'Small town low-commute',\n",
" 10: 'Rural',\n",
"}\n",
"\n",
"def ruca_band(r):\n",
" if pd.isna(r): return 'Unknown'\n",
" r = int(r)\n",
" if r <= 3: return 'Metropolitan'\n",
" if r <= 6: return 'Micropolitan'\n",
" if r <= 9: return 'Small town'\n",
" return 'Rural'\n",
"\n",
"dc_ruca = joined_df.copy()\n",
"dc_ruca['ruca_label'] = dc_ruca['primary_ruca'].map(ruca_buckets)\n",
"dc_ruca['ruca_band'] = dc_ruca['primary_ruca'].apply(ruca_band)\n",
"\n",
"# National baseline (share of US tracts in each band).\n",
"NATIONAL_SQL = \"\"\"\n",
"select\n",
" case\n",
" when primary_ruca between 1 and 3 then 'Metropolitan'\n",
" when primary_ruca between 4 and 6 then 'Micropolitan'\n",
" when primary_ruca between 7 and 9 then 'Small town'\n",
" when primary_ruca = 10 then 'Rural'\n",
" else 'Unknown'\n",
" end as ruca_band,\n",
" count(*) as tracts\n",
"from public.ruca_codes_2020_tract\n",
"group by 1\n",
"\"\"\"\n",
"with get_conn() as conn:\n",
" national_df = pd.read_sql(NATIONAL_SQL, conn)\n",
"national_df['tracts_pct'] = (100 * national_df['tracts'] / national_df['tracts'].sum()).round(1)\n",
"\n",
"dc_by_band = (\n",
" dc_ruca.groupby('ruca_band').size().rename('dcs').to_frame()\n",
" .assign(dcs_pct=lambda d: (100 * d['dcs'] / d['dcs'].sum()).round(1))\n",
")\n",
"band_compare = dc_by_band.join(national_df.set_index('ruca_band')[['tracts', 'tracts_pct']])\n",
"band_compare['over_index'] = (band_compare['dcs_pct'] / band_compare['tracts_pct']).round(2)\n",
"print('Data centers vs. all US tracts, by RUCA band:')\n",
"print(band_compare.reindex(['Metropolitan', 'Micropolitan', 'Small town', 'Rural', 'Unknown']).fillna(0))\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "15",
"metadata": {},
"outputs": [],
"source": [
"# Fine-grained RUCA breakdown: DC count, median power, demographics, energy infra\n",
"# at each of the 10 RUCA codes.\n",
"ruca_profile = (\n",
" dc_ruca.groupby('primary_ruca', dropna=False)\n",
" .agg(\n",
" dcs=('master_id', 'count'),\n",
" median_power_mw=('power_mw', 'median'),\n",
" total_power_mw=('power_mw', 'sum'),\n",
" med_hh_income=('median_household_income', 'median'),\n",
" med_poverty=('poverty_rate', 'median'),\n",
" med_bachelor_pct=('bachelor_or_higher_pct', 'median'),\n",
" med_pct_white=('non_hispanic_white_pct', 'median'),\n",
" med_pct_black=('non_hispanic_black_pct', 'median'),\n",
" med_pct_hispanic=('hispanic_latino_pct', 'median'),\n",
" med_pop_density=('tract_pop_density', 'median'),\n",
" med_eia_gens_50km=('eia_gen_count', 'median'),\n",
" )\n",
" .round(1)\n",
")\n",
"ruca_profile.insert(0, 'description', ruca_profile.index.map(ruca_buckets))\n",
"print('Per-RUCA-code profile of data centers:')\n",
"ruca_profile\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "16",
"metadata": {},
"outputs": [],
"source": [
"# Plot: DC count by RUCA band vs. national tract share.\n",
"fig, axes = plt.subplots(1, 2, figsize=(14, 5))\n",
"\n",
"order = ['Metropolitan', 'Micropolitan', 'Small town', 'Rural']\n",
"plot_df = band_compare.reindex(order).fillna(0)\n",
"\n",
"ax = axes[0]\n",
"x = range(len(plot_df))\n",
"width = 0.38\n",
"ax.bar([i - width/2 for i in x], plot_df['dcs_pct'], width, label='Data centers', color='steelblue')\n",
"ax.bar([i + width/2 for i in x], plot_df['tracts_pct'], width, label='All US tracts', color='lightgray', edgecolor='gray')\n",
"ax.set_xticks(list(x))\n",
"ax.set_xticklabels(plot_df.index, rotation=15)\n",
"ax.set_ylabel('% of total')\n",
"ax.set_title('DC share vs. national tract share, by RUCA band')\n",
"ax.legend()\n",
"\n",
"ax = axes[1]\n",
"colors = ['firebrick' if v > 1 else 'steelblue' for v in plot_df['over_index']]\n",
"ax.barh(plot_df.index, plot_df['over_index'], color=colors)\n",
"ax.axvline(1.0, color='black', linestyle='--', lw=1)\n",
"ax.set_xlabel('Over-index (1.0 = at parity with national)')\n",
"ax.set_title('How much DCs over- or under-represent each RUCA band')\n",
"for i, v in enumerate(plot_df['over_index']):\n",
" ax.text(v, i, f' {v:.2f}x', va='center')\n",
"\n",
"plt.tight_layout()\n",
"plt.show()\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "17",
"metadata": {},
"outputs": [],
"source": [
"# The non-metro tail: who's building in rural / small-town / micropolitan tracts?\n",
"# These are often the most interesting builds (hyperscale greenfield, low-cost power).\n",
"nonmetro = dc_ruca[dc_ruca['ruca_band'].isin(['Rural', 'Small town', 'Micropolitan'])].copy()\n",
"\n",
"print(f'Non-metro DCs: {len(nonmetro):,}\\n')\n",
"\n",
"# Top operators in non-metro tracts.\n",
"print('Top operators in non-metro tracts:')\n",
"top_ops = (\n",
" nonmetro.groupby('operator', dropna=False)\n",
" .agg(dcs=('master_id', 'count'),\n",
" total_power_mw=('power_mw', 'sum'),\n",
" median_power_mw=('power_mw', 'median'))\n",
" .sort_values('dcs', ascending=False)\n",
" .head(15)\n",
" .round(1)\n",
")\n",
"print(top_ops, '\\n')\n",
"\n",
"# Top states in non-metro tracts.\n",
"print('Top states for non-metro DCs:')\n",
"top_states = (\n",
" nonmetro.groupby('state', dropna=False)\n",
" .agg(dcs=('master_id', 'count'),\n",
" total_power_mw=('power_mw', 'sum'),\n",
" med_pop_density=('tract_pop_density', 'median'))\n",
" .sort_values('dcs', ascending=False)\n",
" .head(10)\n",
" .round(1)\n",
")\n",
"print(top_states, '\\n')\n",
"\n",
"# The biggest non-metro builds by power.\n",
"print('Largest non-metro DCs by stated power_mw:')\n",
"big_nonmetro = (\n",
" nonmetro.dropna(subset=['power_mw'])\n",
" .nlargest(15, 'power_mw')\n",
" [['name', 'operator', 'city', 'state', 'power_mw',\n",
" 'primary_ruca', 'primary_ruca_description', 'tract_pop_density']]\n",
" .reset_index(drop=True)\n",
")\n",
"big_nonmetro\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "18",
"metadata": {},
"outputs": [],
"source": [
"# power_mw coverage across the DC universe — most rows are null, which is why\n",
"# \"biggest non-metro by power\" surfaced only a handful.\n",
"coverage = (\n",
" dc_ruca.assign(has_power=dc_ruca['power_mw'].notna())\n",
" .groupby('ruca_band', dropna=False)\n",
" .agg(dcs=('master_id', 'count'),\n",
" with_power_mw=('has_power', 'sum'))\n",
")\n",
"coverage['pct_with_power'] = (100 * coverage['with_power_mw'] / coverage['dcs']).round(1)\n",
"print('power_mw coverage by RUCA band:')\n",
"print(coverage)\n",
"print(f\"\\nOverall: {dc_ruca['power_mw'].notna().sum():,} / {len(dc_ruca):,} DCs have power_mw \"\n",
" f\"({100*dc_ruca['power_mw'].notna().mean():.1f}%)\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "19",
"metadata": {},
"outputs": [],
"source": [
"# Now that EIA nameplate_capacity_mw is loaded, size non-metro DCs by the\n",
"# generation capacity within 50 km of each site (instead of the sparse power_mw).\n",
"# Re-derive non-metro slice from the updated joined_df.\n",
"nonmetro = joined_df[joined_df['primary_ruca'].isin([2,3,4,5,6,7,8,9,10])].copy()\n",
"nonmetro['ruca_band'] = nonmetro['primary_ruca'].apply(ruca_band)\n",
"\n",
"print(f'Non-metro DCs: {len(nonmetro):,}\\n')\n",
"\n",
"# Largest non-metro DCs ranked by nearby grid capacity.\n",
"big_by_grid = (\n",
" nonmetro.sort_values('eia_capacity_mw', ascending=False)\n",
" .head(20)\n",
" [['name', 'operator', 'city', 'state',\n",
" 'primary_ruca', 'primary_ruca_description',\n",
" 'eia_capacity_mw', 'eia_capacity_nuclear', 'eia_capacity_hydro',\n",
" 'eia_capacity_ng', 'eia_capacity_coal',\n",
" 'eia_capacity_solar', 'eia_capacity_wind']]\n",
" .round(0)\n",
" .reset_index(drop=True)\n",
")\n",
"print('Largest non-metro DCs by nearby grid capacity (50 km):')\n",
"big_by_grid\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "20",
"metadata": {},
"outputs": [],
"source": [
"# Hyperscalers' non-metro footprint, sized by surrounding grid capacity + fuel mix.\n",
"hyperscaler_map = {\n",
" 'Amazon Web Services': 'AWS', 'Amazon AWS': 'AWS', 'Amazon': 'AWS',\n",
" 'Microsoft': 'Microsoft',\n",
" 'Meta': 'Meta', 'Meta, Inc.': 'Meta', 'Facebook': 'Meta',\n",
" 'Google': 'Google', 'Alphabet': 'Google',\n",
" 'Apple': 'Apple',\n",
" 'Oracle': 'Oracle',\n",
" 'Yahoo': 'Yahoo',\n",
"}\n",
"nonmetro['op_group'] = nonmetro['operator'].map(hyperscaler_map).fillna(\n",
" nonmetro['operator'].where(nonmetro['operator'].notna(), 'Unknown')\n",
")\n",
"\n",
"hyperscaler_view = (\n",
" nonmetro[nonmetro['op_group'].isin(['AWS','Microsoft','Meta','Google','Apple','Oracle','Yahoo','Unknown'])]\n",
" .groupby('op_group')\n",
" .agg(\n",
" dcs=('master_id', 'count'),\n",
" states=('state', 'nunique'),\n",
" sum_nearby_capacity_mw=('eia_capacity_mw', 'sum'),\n",
" median_nearby_capacity_mw=('eia_capacity_mw', 'median'),\n",
" sum_nearby_hydro_mw=('eia_capacity_hydro', 'sum'),\n",
" sum_nearby_nuclear_mw=('eia_capacity_nuclear', 'sum'),\n",
" sum_nearby_ng_mw=('eia_capacity_ng', 'sum'),\n",
" sum_nearby_solar_mw=('eia_capacity_solar', 'sum'),\n",
" sum_nearby_wind_mw=('eia_capacity_wind', 'sum'),\n",
" )\n",
" .sort_values('dcs', ascending=False)\n",
" .round(0)\n",
")\n",
"print(\"Non-metro DCs by operator group, sized by aggregate nearby grid capacity:\")\n",
"hyperscaler_view\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": ".venv",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.14.5"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

11
ingest_eia_energy_layers.py
View File

@@ -78,7 +78,13 @@ EIA_DATASETS = {
# values must be requested explicitly. seds returns only id columns; the numeric
# value column must be requested explicitly.
EIA_DATASET_DATA_FIELDS = {
"electricity/operating-generator-capacity": ["latitude", "longitude"],
"electricity/operating-generator-capacity": [
"latitude",
"longitude",
"nameplate-capacity-mw",
"net-summer-capacity-mw",
"net-winter-capacity-mw",
],
"electricity/facility-fuel": ["generation", "gross-generation"],
"seds": ["value"],
}
@@ -897,6 +903,9 @@ def build_flat_tables(conn):
properties->>'balancing-authority-name' as balancing_authority_name,
latitude_raw as latitude,
longitude_fixed as longitude,
nullif(properties->>'nameplate-capacity-mw', '')::double precision as nameplate_capacity_mw,
nullif(properties->>'net-summer-capacity-mw', '')::double precision as net_summer_capacity_mw,
nullif(properties->>'net-winter-capacity-mw', '')::double precision as net_winter_capacity_mw,
properties as raw_properties
from fixed
"""

222
output/data_center_demographic_ruca_energy_summary.md Normal file
View File

@@ -0,0 +1,222 @@
# US Data Centers — Demographic, Urban-Rural & Energy Context Analysis
**Date:** 2026-05-18
**Notebook:** [cluster_analysis.ipynb](../cluster_analysis.ipynb)
**Universe:** 1,833 data centers in `public.master_data_centers`, joined to ACS-2024 demographics, USDA RUCA-2020 codes, and EIA operating-generator capacity (50 km radius, latest period 2026-02, status=OP).
> **Update 2026-05-18**: 196 previously-null `state` values were backfilled from `geoid` (first 2 chars = state FIPS). All 1,833 DCs now have a state; all state-level numbers below reflect the corrected attribution.
---
## Headline findings
1. **DC tracts are richer, more educated, and more diverse than the US average.** Median household income $103,623 vs. national $78,538 (+32%); 49% bachelor's+ vs. 35% (+14 pp); poverty rate 7.2% vs. 12.4%. Non-Hispanic white share is *below* national (50% vs. 58%), driven by Asian-heavy (mean 13% vs. 6%) and Hispanic-significant tracts.
2. **The metro skew is more modest than expected: 1.11×.** 89% of DCs sit in metropolitan tracts, but 80% of *all* US tracts are metropolitan — so DCs are only slightly more concentrated than the underlying population distribution would predict.
3. **The non-metro tail is overwhelmingly hyperscale and Pacific Northwest.** Of 190 DCs outside metropolitan tracts (RUCA 4–10), AWS owns 67, Meta 22, Microsoft 10, Google 4, Yahoo 2 — combined 55% of the non-metro footprint. Oregon (86) and Washington (40) alone hold 66% of non-metro DCs, anchored to the Columbia River hydropower corridor.
4. **Clustered DCs are demographically distinct from isolated ones.** DCs in DBSCAN clusters (n=1,583) sit in tracts with $108K median income vs. $73K for isolated DCs (n=250) — a $35K gap. Clustered DCs are more educated (+18 pp bachelor's), more diverse (–25 pp non-Hispanic white), and embedded in much denser energy infrastructure (89 vs. 40 generators within 50 km).
5. **Microsoft co-locates with the largest US nuclear plant.** Microsoft's Goodyear, AZ campus has 14.6 GW of generation within 50 km — including 4.2 GW from Palo Verde Nuclear, the largest in the US. Despite the campus being in a RUCA-2 "Metro high-commute" tract (not strictly metro core), the surrounding grid is the densest by capacity in our analysis.
---
## Dataset coverage and joins
| Source table | Rows | Join key | Coverage |
|---|---|---|---|
| `master_data_centers` | 1,833 | base | — |
| `master_data_center_spatial_clusters` | 1,831 | `master_id` | 99.9% |
| `_dc_census_tract_acs_2024` | ~73,000 tracts | `geoid` | 1,807 matched (98.6%) |
| `ruca_codes_2020_tract` | 85,528 tracts | `tract_fips_20 = geoid` | 1,826 matched (99.6%) |
| `energy_eia_operating_generator_capacity_flat` | 4.7M rows | `ST_DWithin(geom, 50km)` | 1,831 DCs have ≥1 nearby gen |
Energy aggregation uses period `2026-02` only with `status='OP'`, summing `nameplate_capacity_mw` for operating generators within 50 km of each DC. Note: EIA capacity columns were added to this table on 2026-05-17 — prior to that the `_flat` table had no MW values despite its name.
---
## 1. Demographic profile of DC tracts (n=1,807 with non-null ACS)
| Metric | DC tract (median) | DC tract (mean) | US avg | Δ mean vs. US |
|---|---:|---:|---:|---:|
| Median household income | $103,623 | $114,543 | $78,538 | **+$36,005** |
| Per-capita income | $51,283 | $55,725 | $43,313 | +$12,412 |
| Poverty rate | 7.2% | 10.1% | 12.4% | −2.3 pp |
| Unemployment rate | 3.5% | 4.4% | 5.4% | −1.0 pp |
| Bachelor's+ % | 49.3% | 46.2% | 35.0% | **+11.2 pp** |
| Broadband subscription % | 94.9% | 93.5% | 89.0% | +4.5 pp |
| Non-Hispanic white % | 50.2% | 51.0% | 58.4% | **−7.4 pp** |
| Hispanic / Latino % | 12.8% | 19.5% | 19.5% | 0.0 pp |
| Non-Hispanic Black % | 5.9% | 10.6% | 12.1% | −1.5 pp |
| Non-Hispanic Asian % | 6.4% | 13.4% | 6.4% | **+7.0 pp** |
**Interpretation.** DC tracts skew toward high-income, highly-educated, technically connected, and racially diverse (specifically Asian-heavy). The race composition is interesting: DC tracts are *less* non-Hispanic white than national average, not more. This reflects DC siting in mixed-race coastal/exurban tech corridors (Bay Area, Northern Virginia, Seattle) rather than in homogeneous suburbs.
**Data quality note.** `avg_household_size` contains sentinel-value pollution (min: −666,666,666), so the mean is unusable; the median (2.55) is sensible.
---
## 2. Geographic concentration (top 15 states)
| State | DC count | Total power_mw (where known) | Median HH income | Median bachelor's % | Median % white | Notes |
|---|---:|---:|---:|---:|---:|---|
| **VA** | **378** | 255 | $141,250 | 62.6% | 42.5% | Loudoun / DC-Alley dominance (20.6% of all US DCs) |
| TX | 162 | 597 | $88,228 | 46.2% | 32.0% | DFW + Austin + San Antonio |
| CA | 147 | 130 | $164,928 | 56.4% | 22.4% | Bay Area + LA basin |
| OR | 145 | 125 | $72,719 | 20.0% | 63.2% | Columbia River hydro corridor (rural) |
| OH | 103 | 135 | $128,875 | 47.0% | 74.5% | Columbus boom — fastest-rising market |
| WA | 93 | 70 | $91,623 | 21.9% | 40.3% | Quincy/Wenatchee + Seattle |
| AZ | 69 | 54 | $85,335 | 35.2% | 51.6% | Phoenix/Goodyear hyperscale |
| IA | 65 | 0 | $93,393 | 34.3% | 88.1% | 88% white (rural Midwest) |
| NJ | 62 | 98 | $147,321 | 59.4% | 32.9% | NYC-metro carrier hotels |
| IL | 61 | 128 | $96,191 | 52.9% | 52.0% | Chicago metro |
| GA | 50 | 241 | $101,176 | 51.4% | 31.6% | Atlanta + high-power rural builds |
| NY | 48 | 47 | $77,465 | 47.6% | 74.8% | NYC + upstate |
| NV | 41 | 0 | $93,409 | 31.2% | 34.6% | Reno + Las Vegas |
| TN | 32 | 0 | — | — | 54.8% | Nashville + Memphis (newly visible after state backfill) |
| NC | 31 | 56 | $82,708 | 44.7% | 59.6% | Charlotte + Catawba (nuclear-adjacent) |
**Virginia alone holds 20.6% of all US DCs** (378 of 1,833), with the most affluent tract profile in the top 15 — a Loudoun County effect. The state backfill substantially elevated **Ohio (76 → 103)** and **Texas (135 → 162)**, pushing TX into the #2 slot. The previously-uncounted **Tennessee (32) now appears in the top 15**.
Oregon and Washington tracts look notably different from the urban-heavy states (lower income, lower education, lower broadband, higher Hispanic share), reflecting their rural Columbia River siting.
---
## 3. Spatially clustered DCs vs. isolated DCs
DBSCAN cluster assignment from `master_data_center_spatial_clusters` (1,583 clustered, 250 isolated):
| Metric (median) | Isolated | In cluster | Δ |
|---|---:|---:|---:|
| Median household income | $73,500 | $108,359 | **+$34,859** |
| Bachelor's+ % | 33.2 | 51.2 | **+18.0 pp** |
| Poverty rate | 11.6 | 6.9 | −4.7 pp |
| Non-Hispanic white % | 71.0 | 45.9 | **−25.1 pp** |
| EIA generators within 50 km | 40 | 89 | +49 |
| EIA capacity within 50 km (MW) | 2,176 | 3,300 | +1,125 |
**Reading.** A clustered data center sits, at the median, in a tract that is ~$35K richer, 18 pp more educated, and 25 pp less non-Hispanic white than an isolated one — and is surrounded by twice as much energy infrastructure (and 50% more generation capacity). The isolated set looks like rural / small-town America (whiter, poorer, less educated); the clustered set looks like coastal exurban tech corridors.
---
## 4. RUCA (urban-rural) distribution
National baseline of all US tracts: 80% Metropolitan, 9% Micropolitan, 3% Small town, 8% Rural.
| RUCA band | DCs | DC % | US tract % | Over-index |
|---|---:|---:|---:|---:|
| Metropolitan (1–3) | 1,636 | 89.3% | 80.1% | **1.11×** |
| Micropolitan (4–6) | 98 | 5.3% | 9.0% | 0.59× |
| Small town (7–9) | 15 | 0.8% | 2.9% | 0.28× |
| Rural (10) | 77 | 4.2% | 7.6% | 0.55× |
| Unknown / missed match | 7 | 0.4% | — | — |
**Reading.** The metro skew is real but only mild — 1.11×. The eye-catching pattern is that **rural tracts (RUCA 10) hold more DCs than micropolitan or small-town combined**, because the hyperscale greenfield model deliberately bypasses small-city economies in favor of remote, cheap-power, low-population sites.
### Per-RUCA-code drilldown
| RUCA | Description | DCs | Median HH income | Median pop density | Median EIA gens (50km) |
|---:|---|---:|---:|---:|---:|
| 1 | Metro core | 1,425 | $110,333 | 1,859 / sq mi | 97 |
| 2 | Metro high-commute | 206 | $105,404 | 96 | 49 |
| 3 | Metro low-commute | 5 | $119,495 | 22 | 23 |
| 4 | Micropolitan core | 54 | $63,698 | 312 | 53 |
| 5 | Micropolitan high-commute | 22 | $72,465 | 191 | 51 |
| 6 | Micropolitan low-commute | 22 | $72,719 | 69 | 59 |
| 7 | Small town core | 14 | $87,522 | 2,336 | 40 |
| 8 | Small town high-commute | 1 | $69,074 | 36 | 41 |
| 10 | Rural area | 77 | $93,820 | 12 | 42 |
**Two surprises:**
- Rural DCs (RUCA 10) sit in tracts with $93.8K median income — *higher* than micropolitan DCs ($63.7K–$72.7K). The rural DC sites are not poor rural America; they are wealthy-by-rural-standards counties chosen for power and water access.
- Micropolitan-core DCs (RUCA 4) have the *lowest* median income at $63.7K — the closest thing to "economic-development DC siting" in the dataset.
---
## 5. Non-metro deep dive (190 DCs, RUCA 4–10)
### Operators
| Operator | Non-metro DCs |
|---|---:|
| Amazon Web Services | 67 |
| *(null operator)* | 50 |
| Meta | 20 (+ 2 as "Meta, Inc.") |
| Microsoft | 10 |
| Google | 4 |
| Rowan Green Data | 4 |
| NTT | 2 |
| Yahoo | 2 |
| Amazon AWS *(dupe)* | 2 |
**The five hyperscalers (AWS, Meta, Microsoft, Google, Yahoo) account for 105 of 190 non-metro DCs (55%).** If the 50 null-operator rows skew similarly hyperscale (likely — they're disproportionately in OR/WA), the share is probably closer to 75%.
### States (post-backfill)
| State | Non-metro DCs |
|---|---:|
| Oregon | 86 |
| Washington | 40 |
| Texas | 9 |
| New Mexico | 7 |
| North Carolina | 6 |
| Pennsylvania | 5 |
| Wisconsin | 4 |
| New York | 3 |
| Tennessee | 3 |
| Georgia | 3 |
**Oregon + Washington = 126 (66%) of all non-metro DCs.** This is the Columbia River basin: Prineville / Hermiston / Boardman / The Dalles (OR) and Quincy / East Wenatchee / Moses Lake (WA). The pull is hydroelectric power (cheap, low-carbon, abundant) and cool dry climate (free-cooling).
The state backfill clarified the rest of the non-metro tail: **Texas (9)** and **Pennsylvania (5)** were previously hidden in the null bucket. These likely represent shale-gas-adjacent builds (Permian and Marcellus respectively).
---
## 6. Energy footprint by operator (using EIA capacity within 50 km)
Aggregated across DCs in RUCA 2–10 (i.e. anything outside dense metro core, n=401):
| Operator | DCs | States | Total nearby capacity (GW) | Median per site (GW) | Hydro (GW) | Nuclear (GW) | NG (GW) | Solar (GW) | Wind (GW) |
|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
| AWS | 93 | 5 | 397 | 4.8 | 66 | 2.5 | 201 | 4.6 | **114** |
| *(Unknown)* | 118 | 26 | 339 | 2.3 | 86 | 35 | 135 | 23 | 19 |
| Meta | 51 | 11 | 120 | 2.0 | 4.9 | 0 | 61 | 16 | 0.3 |
| Microsoft | 26 | 6 | 113 | 3.4 | 28 | **13** | 39 | 9.1 | 8.1 |
| Google | 31 | 5 | 100 | 3.9 | 14 | 0 | 43 | 3.6 | 4.7 |
| Apple | 5 | 2 | 4 | 0.6 | 1.6 | 0 | 1.1 | 0.9 | 0.4 |
| Yahoo | 2 | 1 | 7 | 3.5 | 6.4 | 0 | 0 | 0 | 0.7 |
**Distinct hyperscaler strategies, visible in the fuel mix:**
- **AWS** has aggregated 114 GW of *wind* exposure across its 93 sites — by far the most renewable-coupled portfolio. Also heavy hydro (66 GW) from its OR/WA footprint and 201 GW of natural gas as baseline.
- **Microsoft** has the highest *nuclear* exposure (12.6 GW) — almost entirely from its Goodyear, AZ campuses near Palo Verde Nuclear.
- **Meta** has the most *solar* (16 GW) among the named hyperscalers, but minimal nuclear or wind — consistent with its New Mexico (Los Lunas) and Iowa builds.
- **Google** is split — moderate hydro and natural gas, modest renewables.
### Largest non-metro grid neighborhoods (top sites by surrounding capacity)
| DC | Operator | Location | Nearby capacity | Fuel highlight |
|---|---|---|---:|---|
| PHX70 / PHX-10 / PHX-11 | Microsoft (Azure) | Goodyear, AZ (RUCA 2) | 14.0–14.6 GW | **4.2 GW nuclear (Palo Verde)** + 6.4 GW gas + 2.2 GW solar |
| Stream PHX-1 | Stream Data Centers | Goodyear, AZ | 13.4 GW | Same Palo Verde / gas mix |
| T5 Charlotte Campus | T5 | Kings Mountain, NC (RUCA 6) | 12.9 GW | **4.9 GW nuclear** (Catawba) + 5.5 GW gas + 1.5 GW coal |
| Apple Maiden | Apple | Maiden, NC (RUCA 2) | 9.1 GW | 2.4 GW nuclear + 4.6 GW gas |
| Percheron DC | Rowan Green Data | (Texas, RUCA 10) | 6.7 GW | **3.0 GW wind** + 0.9 GW hydro + 2.4 GW gas |
---
## Data quality flags
1. ~~196 of 1,833 DCs (10.7%) have null `state`~~ **Resolved 2026-05-18** by backfilling from `geoid` first-2-chars (state FIPS).
2. **`master_data_centers.power_mw` is populated for only 108 / 1,833 DCs (5.9%).** Useless as a sizing metric without imputation or alternative source. Nearby EIA capacity is the more reliable proxy (used as the per-DC scale in this analysis). A grant-funded scrape of Baxtel / Data Center Map would close this gap.
3. **50 of 190 non-metro DCs (26%) have null `operator`.** Likely hyperscalers based on geography (OR/WA) but unconfirmed.
4. **Operator-string fragmentation**: "Meta" vs. "Meta, Inc."; "Amazon Web Services" vs. "Amazon AWS" vs. "amazon web services"; "Microsoft" vs. "Microsoft Azure". Inflates distinct-operator counts and fragments per-operator totals.
5. **`avg_household_size` column has sentinel pollution** (min: −666,666,666). Use median or filter before using.
6. **7 DCs failed RUCA join** — Puerto Rico tracts or non-US locations; trivial.
7. **EIA generator coordinates had a longitude sign error for 2008-01 through 2010-11** (~11K rows with positive lower-48 longitudes). The flat-table build at [ingest_eia_energy_layers.py:839-870](../ingest_eia_energy_layers.py#L839-L870) corrects this in `longitude` and `geom`, so spatial joins are unaffected.
---
## Suggested next steps
1. **Backfill `power_mw`** from Baxtel / Data Center Map (paid scrape — grant work).
2. **Operator-string deduplication** — collapse "Meta"/"Meta, Inc.", "AWS" variants, etc., before any per-operator analysis.
3. **Watershed (HUC8) join** — `public.watershed_huc8` is loaded but unused; would let us look at water stress overlap, particularly for the 190 non-metro DCs.
4. **State-level energy demand context** — `im3_state_projected_moderate_50` and `seds_state_msn_year` are loaded; joining these would let us compute "DC nearby capacity as share of state grid" rather than absolute MW.
5. **Opposition cases overlay** — `opposition_cases_geocoded` is loaded but unused; could test whether cluster-vs-isolated demographic differences predict community opposition.