Topics · water · v0.2

Water stress index for data centers

Name: Water Stress Index for data centers (WSI-DC)
Creator: Datacenter Atlas
License: https://creativecommons.org/licenses/by/4.0/

Full method

1,060 of 6,687 data centers (16 %) sit where the index reaches 60+: places where cooling a data center competes with scarce water, today or by 2050.

Map

Data centers coloured by their index, from green (low) to red (high). Each pie shows how many of its data centers fall in each of the five levels (hover for the counts). 103 of the 6,790 data centers are not scored because their location is approximate (town level only) or outside the Aqueduct coverage, so the map counter can exceed the 6,687 scored. Zoom in for individual sites and click one for its pillars.

loading map…

How to read a score. It describes the location, not any facility's water use, which depends on its cooling (air-cooled and closed-loop sites use little water, evaporative cooling a lot). From 0 to 100, higher means scarcer water, hotter weather to cool against and more people sharing the supply; from 60 the index is called high. A country or hub figure is the median of its data centers.

Most water-stressed countries

Pakistan7 data centers (few)score 88
Bangladesh16 data centersscore 84
Angola5 data centers (few)score 84
United Arab Emirates18 data centersscore 83
Israel16 data centersscore 72
South Africa69 data centersscore 71
India160 data centersscore 71
Egypt5 data centers (few)score 71
Peru18 data centersscore 70
Chile51 data centersscore 68

All 76 countries →

Most water-stressed hubs

Dhaka, Bangladesh10 data centersscore 84
Dubai, United Arab Emirates10 data centersscore 83
Gazipur, Bangladesh5 data centers (few)score 82
Hyderabad, India10 data centersscore 82
Chennai, India19 data centersscore 78
Delhi, India21 data centersscore 76
Phoenix, United States53 data centersscore 75
Bologna, Italy8 data centers (few)score 73
San Diego, United States9 data centers (few)score 73
Madrid, Spain28 data centersscore 72

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 (≥ 5). Click a column to sort. “High” = data centers scoring ≥ 60. “Few sites” marks a median over fewer than 10 data centers, which a single site can move.

#
1	Pakistan few sites	88	6	100	100	78	73	95	98	52	7
2	Bangladesh	84	15	100	100	63	55	90	100	52	16
3	Angola few sites	84	4	100	100	77	47	93	82	55	5
4	United Arab Emirates	83	18	100	100	75	68	100	60	39	18
5	Israel	72	16	89	100	71	52	58	66	36	16
6	South Africa	71	63	100	100	77	54	27	65	38	69
7	India	71	105	70	84	54	56	96	86	45	160
8	Egypt few sites	71	5	78	100	62	31	66	97	44	5
9	Peru	70	18	90	100	77	39	22	91	44	18
10	Chile	68	45	87	100	70	57	13	78	55	51
11	Iran	67	13	94	100	66	47	34	58	50	16
12	Spain	67	46	96	100	71	58	28	73	35	73
13	Thailand	64	26	47	68	55	47	100	92	44	27
14	Belgium	62	41	75	100	78	39	17	60	29	46
15	Greece	62	10	79	97	56	40	49	58	37	14
16	México	61	38	78	100	63	41	15	66	44	59
17	Turkey	58	16	61	95	68	42	39	88	41	36
18	Portugal	58	4	68	88	59	52	21	76	30	21
19	Indonesia	56	36	46	63	35	46	93	90	44	127
20	Philippines	56	5	34	70	32	55	93	83	37	31
21	Lithuania few sites	52	0	55	74	64	51	19	67	28	8
22	Australia	51	1	51	80	56	43	34	64	20	169
23	China	48	52	38	56	41	44	58	68	36	122
24	North Macedonia few sites	48	0	49	70	40	50	25	75	32	5
25	Tanzania few sites	48	1	7	62	30	55	95	80	43	8
26	Italy	46	37	45	60	37	53	38	63	45	114
27	United Kingdom	46	0	47	74	58	53	12	68	21	305
28	Nigeria	46	4	1	1	12	48	96	68	43	10
29	Vietnam	46	0	22	38	29	42	81	87	41	17
30	Germany	45	0	47	69	56	42	20	68	18	390
31	Dominican Republic few sites	44	0	27	32	19	51	95	68	34	5
32	Hong Kong	44	0	27	38	27	41	80	89	24	50
33	Kenya	42	0	20	73	23	53	35	84	32	13
34	Japan	40	0	38	47	30	37	52	80	22	155
35	Luxembourg	40	0	40	63	45	43	19	49	19	16
36	Romania	40	0	36	50	25	55	37	76	28	25
37	Russia	39	47	35	52	23	44	18	87	30	109
38	Brazil	39	0	19	41	16	38	58	80	34	250
39	Taiwan	39	0	23	27	26	34	77	86	22	21
40	South Korea	39	0	33	43	25	30	45	91	20	36
41	United States	37	285	34	47	29	41	41	57	17	2133
42	Venezuela	35	1	7	19	22	34	90	78	36	17
43	Costa Rica few sites	35	0	19	31	31	28	68	78	18	9
44	France	34	34	22	37	27	51	20	68	19	395
45	Moldova few sites	34	0	23	25	16	55	32	75	42	5
46	Georgia few sites	34	0	28	39	27	32	32	62	25	6
47	Puerto Rico few sites	33	0	10	18	12	38	98	65	16	9
48	Estonia few sites	32	0	26	37	25	45	15	72	19	7
49	Malaysia	31	0	3	2	6	40	99	59	25	47
50	Singapore	31	0	3	2	6	40	99	88	15	70
51	Ghana few sites	31	0	0	0	6	44	96	89	28	6
52	Poland	29	3	18	21	20	58	23	73	18	76
53	Denmark	26	0	19	24	19	51	13	61	18	38
54	Croatia	26	0	13	20	16	38	32	70	18	15
55	Ireland	24	0	16	20	22	42	5	65	18	90
56	Slovenia few sites	24	0	13	20	16	38	21	65	12	8
57	Argentina	23	12	0	0	8	46	43	68	25	76
58	Ukraine	23	1	1	0	6	56	26	84	30	36
59	Finland	23	0	18	27	13	32	14	59	11	53
60	Sweden	22	8	9	15	14	41	13	63	10	56
61	Colombia	22	0	2	14	2	29	22	90	26	28
62	Serbia	22	1	1	0	4	51	37	72	18	11
63	Canada	21	2	15	25	15	25	28	71	10	135
64	Kazakhstan few sites	21	0	3	0	10	38	32	66	25	8
65	Netherlands	20	0	9	13	11	32	16	63	16	214
66	Hungary	20	0	1	0	4	47	32	72	15	25
67	Switzerland	20	0	9	15	11	31	19	65	13	63
68	Ecuador few sites	19	0	3	0	8	36	17	82	30	9
69	Bulgaria	19	0	1	0	5	48	21	77	17	20
70	Czechia	18	0	2	0	6	46	20	74	13	30
71	Slovakia	18	0	0	0	3	47	30	68	16	10
72	Austria	18	0	2	0	3	37	24	72	11	48
73	Latvia	18	0	2	0	7	44	17	57	16	13
74	New Zealand	16	0	2	2	12	34	21	64	9	70
75	Norway	10	0	1	0	8	27	9	30	10	38
76	Iceland	7	0	1	0	4	14	0	39	14	35

Pillars & weights

Scarcity: 25 %
Future (2050): 15 %
Seasonality & peaks: 15 %
Drought & groundwater: 10 %
Cooling demand: 15 %
Local competition: 10 %
Social & regulatory: 10 %

Pressure = six physical pillars; index adds the social & regulatory pillar. Cooling source: NASA POWER daily 2015–2024 + WorldClim CMIP6 2041–2060 (pattern-scaled delta).

Distribution of data centers

0–19964

20–392,508

40–592,155

60–79966

80–10094

Validation: the physical pressure score alone ranks the 167 data centers within 25 km of a documented water conflict above the others with ROC-AUC 0.662, over 6,687 scored data centers. It uses the 24 water cases located at a site or town (of those in the conflict atlas; cases located only at country or region level can't be tested), a small sample. Caveat: once spatial clustering is accounted for (resampling whole 5° cells) the interval is 0.49–0.804, which includes 0.5, so this test can't rule out chance. Spatial check →

Pillar sources

Scarcity	WRI Aqueduct 4.0 baseline water stress and depletion (CC BY 4.0).
Future (2050)	Aqueduct 4.0 water stress 2050, business-as-usual scenario.
Seasonality & peaks	Aqueduct 4.0 monthly water stress (worst month, months at high stress) and seasonal variability.
Drought & groundwater	Aqueduct 4.0 drought risk, interannual variability and groundwater table decline (where modelled).
Cooling demand	Cooling degree days today and in 2050 and design wet-bulb temperature, hotter, more humid sites need more (often evaporative) cooling.
Local competition	Residents within 5 km (JRC GHSL 2020, CC BY 4.0), people sharing the same local supply.
Social & regulatory	Aqueduct reputational water risk and electric-power-weighted water risk, plus documented water conflicts within 50 km (Conflict atlas).

Embed this water-stress map ⧉

Free to embed with attribution (the credit link is built in). Per-country versions: /embed/water/<country>, e.g. /embed/water/spain.

<iframe src="https://www.datacenter-atlas.com/embed/water" width="100%" height="480" style="border:0" loading="lazy" title="Water stress of data centers: Water stress index · Datacenter Atlas" allowfullscreen></iframe>
<p>Map: <a href="https://www.datacenter-atlas.com">Datacenter Atlas</a> (data 26.08-c2), ODbL / &copy; OpenStreetMap contributors, water risk: <a href="https://www.wri.org/aqueduct">WRI Aqueduct 4.0</a> (CC BY 4.0)</p>

Method: full method (v0.2, 2026-10-12). Related: social friction index · conflict atlas. Licence CC BY 4.0. Cite: Datacenter Atlas (2026). Water stress index for data centers, v0.2. https://www.datacenter-atlas.com/water