Topics · social acceptability · v0.3

Social friction index

Name: Social friction index for data centers
Creator: Datacenter Atlas
License: https://creativecommons.org/licenses/by/4.0/

Full method

Where are data centers most likely to meet local resistance? The index scores each of 8,357 facilities from 0 to 100 on six pillars: and publishes every pillar, so the number is never a black box.

Map

Data centers coloured by their rating, from green (low) to red (high). The five levels are fifths of all scored data centers, with the score thresholds shown in the legend. Each pie shows how many of its data centers fall in each level (hover for the counts; sites at approximate, town-level locations are not scored). Zoom in for individual sites and click one for its pillars.

loading map…

How to read a score. It describes the location, not public opinion or any operator's conduct: it measures the pressure that tends to come before local opposition, plus the opposition we could document. From 0 to 100, higher means more of that pressure. The score is relative: 80 means higher than 80 % of all scored data centers, so from 80 a data center is in the highest fifth. The absolute value it is drawn from is shown in each site's record. A country or hub figure is the median of its data centers.

Highest-friction countries

Bangladesh16 data centersscore 93
Chile51 data centersscore 89
Ireland90 data centersscore 89
Peru18 data centersscore 82
Netherlands214 data centersscore 79
Pakistan7 data centers (few)score 77
Thailand27 data centersscore 76
Germany390 data centersscore 74
France395 data centersscore 74
Egypt5 data centers (few)score 74

All 76 countries →

Highest-friction hubs

Frankfurt am Main, Germany129 data centersscore 98
London, United Kingdom111 data centersscore 96
Los Angeles, United States25 data centersscore 95
Dhaka, Bangladesh10 data centersscore 94
Lille, France27 data centersscore 94
Madison, United States9 data centers (few)score 94
Phoenix, United States53 data centersscore 93
Santiago, Chile41 data centersscore 92
Marseille, France19 data centersscore 92
Baku, Azerbaijan3 data centers (few)score 91

More hubs →

How the counts fit together: the inventory holds 8,460 facilities, of which 6,790 are data centers (the others are network points of presence or unclassified). The index is computed at the 8,357 facilities located precisely enough; its maps, rankings and medians use data centers only, 6,687 of the 6,790. The rest have a town-level location or fall outside a source's coverage.

Countries

Median over each country's data centers (countries with ≥ 5). Click a column to sort. Pressure is the index without the observed-conflict pillar. “Few sites” marks a median over fewer than 10 data centers, which a single site can move.

#
1	Bangladesh	93	99	94	34	78	65	–	0	16
2	Chile	89	89	82	35	82	60	39	41	51
3	Ireland	89	76	33	48	65	78	92	85	90
4	Peru	82	92	83	27	100	37	–	0	18
5	Netherlands	79	68	24	44	68	53	99	67	214
6	Pakistan few sites	77	87	93	11	100	28	–	0	7
7	Thailand	76	86	64	28	100	48	–	0	27
8	Germany	74	78	65	38	83	50	78	10	390
9	France	74	70	43	30	84	53	63	11	395
10	Egypt few sites	74	85	100	20	94	14	–	0	5
11	North Macedonia few sites	72	82	64	38	93	28	–	0	5
12	United Kingdom	71	64	69	37	83	36	39	49	305
13	Lithuania few sites	69	80	72	30	74	48	–	0	8
14	Angola few sites	68	79	84	28	72	32	–	0	5
15	Turkey	65	76	82	23	88	28	–	0	36
16	Belgium	64	68	86	29	74	36	0	45	46
17	Iran	64	76	88	24	86	7	–	0	16
18	Hong Kong	64	74	43	55	92	64	0	0	50
19	South Africa	63	71	87	42	59	53	0	5	69
20	Israel	62	72	91	31	86	32	0	0	16
21	United States	58	52	44	35	52	51	86	25	2133
22	Philippines	58	70	60	37	92	22	–	0	31
23	South Korea	58	42	42	15	96	44	39	63	36
24	Singapore	56	56	20	49	81	71	39	24	70
25	Greece	56	69	80	30	74	22	–	0	14
26	Romania	55	68	54	32	100	39	–	0	25
27	Estonia few sites	55	67	42	38	99	32	–	0	7
28	India	54	54	87	27	81	36	0	8	160
29	Moldova few sites	51	63	44	47	79	36	–	0	5
30	Portugal	51	61	74	24	81	28	–	5	21
31	Australia	48	45	62	42	68	55	0	5	169
32	Indonesia	48	58	66	42	90	39	0	0	127
33	Ukraine	48	59	27	27	100	54	–	0	36
34	Austria	47	57	17	29	94	36	95	0	48
35	Czechia	46	58	21	35	92	62	–	0	30
36	United Arab Emirates	46	55	100	23	58	35	0	0	18
37	Taiwan	44	36	27	25	96	32	39	41	21
38	Denmark	42	47	38	33	70	46	63	5	38
39	Russia	38	48	55	21	89	51	–	0	109
40	Spain	38	43	88	22	76	36	0	8	73
41	México	38	39	84	28	64	28	0	5	59
42	Serbia	37	47	27	41	96	28	–	0	11
43	Poland	36	42	36	41	90	42	0	0	76
44	Ghana few sites	36	46	23	34	100	39	–	0	6
45	Italy	34	32	58	29	74	36	0	8	114
46	Bulgaria	34	44	19	33	98	51	–	0	20
47	Luxembourg	33	39	58	45	62	36	0	0	16
48	Tanzania few sites	33	41	36	42	88	11	–	0	8
49	Kenya	32	38	50	37	72	32	0	0	13
50	Japan	29	29	49	27	75	42	0	8	155
51	Argentina	29	36	24	39	92	14	–	0	76
52	Slovakia	28	36	24	24	92	36	–	0	10
53	Malaysia	28	25	20	35	60	53	63	26	47
54	Puerto Rico few sites	28	35	14	46	76	51	–	0	9
55	Hungary	26	33	24	28	94	44	–	0	25
56	Croatia	26	32	27	32	91	44	–	0	15
57	Slovenia few sites	26	34	27	32	86	42	–	0	8
58	Brazil	23	26	39	23	81	44	0	5	250
59	Costa Rica few sites	23	30	30	29	74	39	–	0	9
60	Latvia	22	29	18	40	92	42	–	0	13
61	Finland	22	26	32	27	80	44	39	0	53
62	Georgia few sites	22	28	43	26	75	18	–	0	6
63	Dominican Republic few sites	20	26	44	35	73	0	–	0	5
64	New Zealand	20	25	13	39	80	48	–	5	70
65	Switzerland	19	23	25	31	87	42	0	0	63
66	Vietnam	17	22	44	17	72	32	–	0	17
67	Ecuador few sites	15	18	16	23	91	22	–	0	9
68	Venezuela	15	19	23	18	100	28	–	0	17
69	China	14	16	56	16	74	14	0	0	122
70	Nigeria	14	17	25	33	64	27	0	0	10
71	Canada	13	14	20	20	78	46	0	8	135
72	Colombia	13	16	19	24	96	18	–	0	28
73	Sweden	12	13	23	23	83	36	0	5	56
74	Kazakhstan few sites	10	12	20	30	62	14	–	0	8
75	Norway	6	7	8	19	68	22	39	5	38
76	Iceland	5	6	5	38	29	63	39	2	35

Pillars & weights

Water: 25 %
Energy & grid: 20 %
Neighbourhood: 20 %
Concentration: 15 %
Policy context: 10 %
Observed conflict: 10 %

Pressure = the five structural pillars only (no conflict data). Index = all six. Weights are renormalised when a pillar is unavailable for a site.

Does it work?

0.745 AUC

Using the structural pressure score alone, no conflict data, ranks the 998 data centers located within 25 km of a documented contestation above the others with a ROC-AUC of 0.745 (0.5 = chance, 1 = perfect), over 6,687 data centers.

For comparison, facility clustering only (concentration pillar) alone reaches 0.783: on this test, clustering explains much of the signal. The index is meant as a profile of several pressures, not as a better predictor. Drop-one-pillar and spatial checks →

Caveats: conflicts are documented where press coverage exists (Europe, US over-represented); clusters of facilities near one event count as several positives.

Pillars, sources and related pages

Water	WRI Aqueduct 4.0: baseline water stress, 2050 business-as-usual stress, drought risk (CC BY 4.0). A separate index on the same theme: Water stress index
Energy & grid	Data centers per TWh of national electricity demand and grid carbon intensity (Ember, CC BY 4.0); distance to the nearest ≥110 kV substation (OpenStreetMap via Overture Maps, ODbL).
Neighbourhood	Residents within 1 km (JRC GHSL GHS-POP 2020, CC BY 4.0); distance to the nearest residential area (OpenStreetMap via Overture Maps).
Concentration	Number of tracked facilities within 10 km (this inventory).
Policy context	Restrictive measures in force (zoning, grid, water, heat-reuse rules) plus documented moratoria and grid restrictions. In detail: Regulation · Compare regulation
Observed conflict	Documented contest or mixed events from the Conflict atlas within 50 km (recent events weigh double), plus a capped national component. In detail: Conflict atlas

Method: full method (v0.3, 2026-10-12). Observed conflicts: Conflict atlas. Licence CC BY 4.0. Cite: Datacenter Atlas (2026). Social friction index, v0.3. https://www.datacenter-atlas.com/social-friction