Skip to content

Datacenters

Planned AI data center capacity

54 infrastructures ranked by facility power.
Recorded Oct 20, 2026Checked dailySource: Epoch AIHow this board is measuredCSV / JSON

2,715 MW

Microsoft Fairwater Wisconsin ranks first

Facility power

54

Infrastructures ranked

Oct 20

Snapshot recorded

2026 · Checked daily

Follow changes

All 54 infrastructures by facility power

Microsoft Fairwater Wisconsin leads at 2,715 MW; the middle of the board sits at 464.5.

Ahead of second place
265
Held by the top three
18%
Middle of the board
464.5
Below half the leader
45 of 54
Highest 2,715 MWMiddle 464.5
1. Microsoft Fairwater Wisconsin: 2,715 MW2. OpenAI Stargate New Mexico: 2,450 MW3. Anthropic-Amazon New Carlisle: 2,310 MW4. Meta Hyperion: 2,203 MW5. Colossus 2: 2,065 MW6. OpenAI Stargate Shackelford: 1,960 MW7. QTS Cedar Rapids: 1,587 MW8. OpenAI Stargate UAE: 1,400 MW9. OpenAI Stargate Michigan: 1,383 MW10. Meta Prometheus: 1,328 MW11. OpenAI Stargate Wisconsin: 1,300 MW12. Goodnight: 1,294.1 MW13. OpenAI Stargate Milam: 1,200 MW14. OpenAI Stargate Abilene: 1,180 MW15. Google Fort Wayne: 1,035 MW16. Amazon Ridgeland: 1,008 MW17. Crusoe Abilene Expansion: 941 MW18. Amazon Madison Mega Site: 819 MW19. Coreweave Helios: 800 MW20. VNET Bayin Ulanqab: 798 MW21. Envision Galaxy Ulanqab: 612 MW22. Meta Bowling Green: 570 MW23. Huawei Wuhu: 526 MW24. CoreWeave Ellendale ND: 520 MW25. STACK Infrastructure NVA02: 487 MW26. Google Omaha: 474 MW27. Meta Montgomery: 465 MW28. Google Columbus: 464 MW29. Anthropic Lake Mariner: 441 MW30. Huawei Horinger: 414.6 MW31. QTS Eagle Mountain: 411 MW32. AWS New Albany: 405 MW33. CoreWeave Denton TX: 395 MW34. Google Storey County: 389 MW35. Google Lincoln: 339 MW36. QTS Richmond 3: 336 MW37. Google Cedar Rapids: 314 MW38. Anthropic Barber Lake: 300 MW39. OpenAI Stargate Lordstown: 300 MW40. Start Campus Sines Data Campus: 280 MW41. Meta Los Lunas: 273 MW42. Stream Phoenix: 267 MW43. Meta Cheyenne: 249 MW44. AWS Berwick: 212 MW45. CoreWeave Chester VA: 197 MW46. Nebius Mantsala: 196 MW47. Google Red Oak: 185 MW48. CoreWeave Muskogee OK: 181 MW49. Vantage TX1: 179.2 MW50. Meta Aiken: 168 MW51. CoreWeave Lancaster Greenfield site: 124 MW52. Microsoft Narvik Norway: 124 MW53. EcoDataCenter 2: 67 MW54. Nscale Keflavik: 13.28 MW
Rank 1One bar per published rowRank 54

Checked against its source every day; the numbers change when the source publishes.

Epoch AI release: 2026-10-20 · 3 checks behind each row · 60% confidence across rows.

Planned AI data center capacity current rankings
RankNameFacility powerData center capex 2025Data center H100 equivalentsChangeWhy
01Microsoft Fairwater Wisconsin2,715 MW$85.73B4.5M H100e This rank
02OpenAI Stargate New Mexico2,450 MW$66.29B5.1M H100e This rank
03Anthropic-Amazon New Carlisle2,310 MW$72.92B1.7M H100e This rank
04Meta Hyperion2,203 MW$61.82B4.1M H100e This rank
05Colossus 22,065 MW$58B1.8M H100e This rank
06OpenAI Stargate Shackelford1,960 MW$53.03B4.6M H100e This rank
07QTS Cedar Rapids1,587 MW$42.92B3.7M H100e This rank
08OpenAI Stargate UAE1,400 MW$37.88B2M H100e This rank
09OpenAI Stargate Michigan1,383 MW$37.43B2.9M H100e This rank
10Meta Prometheus1,328 MW$38.72B1.2M H100e This rank
11OpenAI Stargate Wisconsin1,300 MW$34.17B2.6M H100e Showing
12Goodnight1,294.1 MW$38.11B2.3M H100e This rank
13OpenAI Stargate Milam1,200 MW$32.46B2.5M H100e This rank
14OpenAI Stargate Abilene1,180 MW$31.93B1M H100e This rank
15Google Fort Wayne1,035 MW$32.69B2.9M H100e This rank
16Amazon Ridgeland1,008 MW$31.82B788.3K H100e This rank
17Crusoe Abilene Expansion941 MW$25.46B1.7M H100e This rank
18Amazon Madison Mega Site819 MW$25.87B514.4K H100e This rank
19Coreweave Helios800 MW$20B638.2K H100e This rank
20VNET Bayin Ulanqab798 MW$26.52B388.1K H100e This rank
21Envision Galaxy Ulanqab612 MW$17.8B169.9K H100e—This rank
22Meta Bowling Green570 MW$15.99B855.2K H100e 1This rank
23Huawei Wuhu526 MW$17.31B221.9K H100e 1This rank
24CoreWeave Ellendale ND520 MW$15.15B505.3K H100e 1This rank
25STACK Infrastructure NVA02487 MW$13.18B250.6K H100e 1This rank
26Google Omaha474 MW$14.96B904.5K H100e 1This rank
27Meta Montgomery465 MW$13.56B407.5K H100e 1This rank
28Google Columbus464 MW$14.62B576K H100e 1This rank
29Anthropic Lake Mariner441 MW$14.32B527.1K H100e 1This rank
30Huawei Horinger414.6 MW$13.12B168.1K H100e 1This rank
31QTS Eagle Mountain411 MW$11.52B347K H100e 1This rank
32AWS New Albany405 MW$12.8B351.7K H100e 1This rank
33CoreWeave Denton TX395 MW$10.68B271.9K H100e 1This rank
34Google Storey County389 MW$11.36B540.7K H100e 1This rank
35Google Lincoln339 MW$10.72B672.1K H100e 1This rank
36QTS Richmond 3336 MW$9.09B291.1K H100e 1This rank
37Google Cedar Rapids314 MW$9.89B670.1K H100e 1This rank
38Anthropic Barber Lake300 MW$7.84B288.5K H100e 1This rank
39OpenAI Stargate Lordstown300 MW$8.11B545.7K H100e 1This rank
40Start Campus Sines Data Campus280 MW$8.83B540.7K H100e 1This rank
41Meta Los Lunas273 MW$8.26B869.1K H100e 1This rank
42Stream Phoenix267 MW$8.33B160.2K H100e 1This rank
43Meta Cheyenne249 MW$7.84B236K H100e 1This rank
44AWS Berwick212 MW$6.4B145K H100e 1This rank
45CoreWeave Chester VA197 MW$4.28B125.8K H100e 1This rank
46Nebius Mantsala196 MW$5.49B226.2K H100e 1This rank
47Google Red Oak185 MW$5.83B279.4K H100e 1This rank
48CoreWeave Muskogee OK181 MW$5.8B198.1K H100e 1This rank
49Vantage TX1179.2 MW$4.85B154.6K H100e 1This rank
50Meta Aiken168 MW$4.89B147K H100e 1This rank
51CoreWeave Lancaster Greenfield site124 MW$3.49B232.9K H100e 1This rank
52Microsoft Narvik Norway124 MW$3.49B265.3K H100e 1This rank
53EcoDataCenter 267 MW$1.82B138.9K H100e 1This rank
54Nscale Keflavik13.28 MW$386.78M11.6K H100e 1This rank

Why this rank

OpenAI Stargate Wisconsin on Planned AI Data Center Capacity

Some explanation inputs are unavailable. Not published: the evidence frozen with this snapshot, comparable prior ranking.

Latest future-dated timeline milestone per center after the dataset release date, requiring positive power, capital cost, and H100-equivalent values.

11

Rank on this board

1,300 MW

Score

3

Inputs behind the score

60% confidence · 1 of 1 components evidenced

What the score is built from

1 of 1 evidenced

Every component evidenced · 1 evidence record each

Components of OpenAI Stargate Wisconsin's score
ComponentValueContribution
Data center facility power MW1300.00000000001,300

Unchanged since the previous snapshot

No change in rank

component_history_unavailable

No component moved since the previous snapshot.

Where each number comes from

3 public records
  • Facility capital costData center capex 2025 · observed Oct 20, 2026, 12:00 PM UTC$34.17B60% confidence
    Source
    Epoch AI open datasets
    Grade
    A
    Freshness
    fresh
    Locator
    {"rowCount":567,"downloadUrl":"https://epoch.ai/data/data_centers/data_center_timelines.csv","releaseDate":"2026-10-20","contentHashKind":"sha256_raw_bytes","expectedColumns":["Data center","Date","Construction status","Buildings operational","IT power (MW)","Power (MW)","H100 equivalents","Performance (8-bit OP/s)","Total capital cost (2025 USD billions)","Compute cost (2025 USD billions)","Construction cost (2025 USD billions)","Annual operating cost (2025 USD billions)","Water use (MGD)","Current H100 equivalents (from Data center)"],"documentationUrl":"https://epoch.ai/data/data-centers-documentation/downloads","documentationHash":"40bcdc602d20546fb253629beb3f25ee4df294388d5614986572e3d7ef4e1396"}
    Open the source
  • Facility powerData center facility power MW · observed Oct 20, 2026, 12:00 PM UTC1,300 MW60% confidence
    Source
    Epoch AI open datasets
    Grade
    A
    Freshness
    fresh
    Locator
    {"rowCount":567,"downloadUrl":"https://epoch.ai/data/data_centers/data_center_timelines.csv","releaseDate":"2026-10-20","contentHashKind":"sha256_raw_bytes","expectedColumns":["Data center","Date","Construction status","Buildings operational","IT power (MW)","Power (MW)","H100 equivalents","Performance (8-bit OP/s)","Total capital cost (2025 USD billions)","Compute cost (2025 USD billions)","Construction cost (2025 USD billions)","Annual operating cost (2025 USD billions)","Water use (MGD)","Current H100 equivalents (from Data center)"],"documentationUrl":"https://epoch.ai/data/data-centers-documentation/downloads","documentationHash":"40bcdc602d20546fb253629beb3f25ee4df294388d5614986572e3d7ef4e1396"}
    Open the source
  • H100-equivalent computeData center H100 equivalents · observed Oct 20, 2026, 12:00 PM UTC2,627,589.692 H100e60% confidence
    Source
    Epoch AI open datasets
    Grade
    A
    Freshness
    fresh
    Locator
    {"rowCount":567,"downloadUrl":"https://epoch.ai/data/data_centers/data_center_timelines.csv","releaseDate":"2026-10-20","contentHashKind":"sha256_raw_bytes","expectedColumns":["Data center","Date","Construction status","Buildings operational","IT power (MW)","Power (MW)","H100 equivalents","Performance (8-bit OP/s)","Total capital cost (2025 USD billions)","Compute cost (2025 USD billions)","Construction cost (2025 USD billions)","Annual operating cost (2025 USD billions)","Water use (MGD)","Current H100 equivalents (from Data center)"],"documentationUrl":"https://epoch.ai/data/data-centers-documentation/downloads","documentationHash":"40bcdc602d20546fb253629beb3f25ee4df294388d5614986572e3d7ef4e1396"}
    Open the source
Epoch AI: ai-data-centers-planned, version 1.1.0162 observations · 1 source · quality passed
Snapshot
4ae3fdf7-e727-4448-a2a0-130bef49aefa
Data hash
6082400f96d4a2ba93109ecc1f0f36dff4b0606eeb8529f3e0bf3252128ab40f
Method hash
28cc45cc30e6de2ee5e11ae978d520d181fd3f18833d9eb109d1cc71a94882d2

More in Datacenters

Operating, under-construction, and planned AI capacity by place and operator.

3 boards
Operational AI data centers

Epoch AI current positive-capacity facilities ranked by estimated total facility MW, with 2025-dollar capex and calculated H100-equivalent compute beside every row.

State AI infrastructure pipeline

A reviewed first-party project cohort with announced, under-construction, and energized stages plus separately scoped disclosed datacenter and compute MW. Permits and grid queues are never inferred.

Enter your website. In about two minutes, Rank.ai asks ChatGPT, Claude and Gemini 12 questions your customers ask and grades how often they name you.

  • Your grade out of 100How often AI names you, cites your site, and how high it ranks you.
  • Who gets namedEvery competitor in the answers, most named first.
  • The pages AI readsThe sources behind each answer.
  • Three fixesWhat to fix first, with a brief for the first article.