Beta
ComputeAI Infrastructure Situation Room

The Compute Hub

The complete state of the AI compute supply chain in one view. Five infrastructure layers, composite demand index, regional readiness rankings, and hyperscaler capex tracker · sourced from earnings reports, utility filings, sustainability reports, and analyst estimates.

Composite AICDI10 regions ranked$283B capex tracked20 chips8 nuclear deals
AI Compute Demand Index
Supply chain pressure · composite
HIGH
rising
Significant bottlenecks in AI compute supply chain. Composite of 5 supply chain layers. 0 = healthy supply · 100 = critical shortage.
0 healthy50 strained100 critical
Top supply chain constraints
  • Foundry concentration
    Critical concentration · near-monopoly on AI silicon supply
  • Hardware demand
    No bubble in sight · full parabolic
  • Energy strain
    Growing pressure on energy supply chains, especially in established hubs
How we compute this
61/100
Composite AICDI
20
Chips tracked
5
Foundries tracked
8
Memory generations
10
Systems
10
Energy regions
8
Nuclear deals
4
Constrained (of 5)
$283B
Total capex
Nordics
Top region

5 layers · Foundries Chips Memory Systems Energy · live health status

5 pressure indices · weighted to produce the composite AICDI

Composite AICDI: 61/100 · HIGH · Trend: rising
Significant bottlenecks in AI compute supply chain. Top constraints: Foundry concentration, Hardware demand, Energy strain.

10 datacenter regions ranked · compute density + energy + water + construction + regulatory

Sweden/Finland/NorwayNordics
Sweden/Finland/Norway
68/100 readiness
Component scores
Compute Density54
Energy Security87
Water84
Construction44
Regulatory76
Key metrics
DC capacity
1,800 MW
Grid share
4%
PPA price
$25/MWh
Growth
+30%
92% carbon-free powerLow PPA pricing ($25/MWh)9 months free cooling
United States flagTexas · ERCOT
United States
62/100 readiness
Component scores
Compute Density69
Energy Security56
Water28
Construction70
Regulatory86
Key metrics
DC capacity
2,200 MW
Grid share
3%
PPA price
$32/MWh
Growth
+45%
Low PPA pricing ($32/MWh)45% growth rate9 major operators presenthigh water stress
United States flagNorthern Virginia
United States
57/100 readiness
Component scores
Compute Density88
Energy Security39
Water52
Construction50
Regulatory51
Key metrics
DC capacity
4,500 MW
Grid share
25%
PPA price
$65/MWh
Growth
+28%
8 major operators present3.2 GW under construction25% grid share (saturation risk)48-month permit waitsaturating regulatory status
#4
United States flagIowa · Midwest Wind Corridor
United States
57/100 readiness
Component scores
Compute Density32
Energy Security78
Water71
Construction39
Regulatory76
Key metrics
DC capacity
900 MW
Grid share
2%
PPA price
$22/MWh
Growth
+20%
70% carbon-free powerLow PPA pricing ($22/MWh)
#5
United States flagOregon · Pacific Northwest
United States
56/100 readiness
Component scores
Compute Density42
Energy Security75
Water54
Construction38
Regulatory76
Key metrics
DC capacity
1,500 MW
Grid share
8%
PPA price
$28/MWh
Growth
+22%
74% carbon-free powerLow PPA pricing ($28/MWh)7 months free cooling
#6
Japan flagJapan
Japan
51/100 readiness
Component scores
Compute Density45
Energy Security45
Water69
Construction31
Regulatory81
Key metrics
DC capacity
1,500 MW
Grid share
1%
PPA price
$85/MWh
Growth
+25%
5 major operators presentHigh energy costs ($85/MWh)
#7
United Arab EmiratesUAE · Gulf States
United Arab Emirates
48/100 readiness
Component scores
Compute Density47
Energy Security49
Water12
Construction61
Regulatory69
Key metrics
DC capacity
500 MW
Grid share
2%
PPA price
$35/MWh
Growth
+60%
Low PPA pricing ($35/MWh)60% growth rate1.5 GW under constructionextreme water stress
#8
Netherlands/GermanyNetherlands · Frankfurt
Netherlands/Germany
47/100 readiness
Component scores
Compute Density37
Energy Security49
Water82
Construction22
Regulatory58
Key metrics
DC capacity
2,000 MW
Grid share
4%
PPA price
$68/MWh
Growth
+10%
7 months free cooling6 major operators present36-month permit waitconstrained regulatory status
#9
Ireland flagIreland
Ireland
40/100 readiness
Component scores
Compute Density24
Energy Security37
Water86
Construction19
Regulatory53
Key metrics
DC capacity
1,200 MW
Grid share
21%
PPA price
$72/MWh
Growth
+8%
10 months free cooling5 major operators present21% grid share (saturation risk)36-month permit waitconstrained regulatory status
#10
Singapore flagSingapore
Singapore
20/100 readiness
Component scores
Compute Density17
Energy Security17
Water12
Construction22
Regulatory38
Key metrics
DC capacity
400 MW
Grid share
7%
PPA price
$110/MWh
Growth
+12%
extreme water stressrestricted regulatory statusHigh energy costs ($110/MWh)

The $283B arms race · 2025 capital expenditure commitments

Microsoft logo
MicrosoftFY2025 earnings guidance
$80B
$80B
Amazon logo
Amazon2025 capex guidance
$75B
$75B
Google logo
Google2025 capex guidance
$50B
$50B
O
OracleFY2025 earnings call
$40B
$40B
Meta logo
Meta2025 capex guidance
$38B
$38B
Combined 2025 capital expenditure · primarily AI datacenter infrastructure$283B

Deep-dive into each layer of the AI compute stack

Pulled from the live dataset · schema-ready for AEO

What is the AI Compute Demand Index (AICDI)?

The AICDI is a composite index measuring overall strain on the AI compute supply chain. Currently at 61/100 (high). It combines five sub-indices: hardware demand (65), foundry concentration (83), memory pressure (57), energy strain (59), and cost pressure (30). Higher scores indicate greater supply chain strain.

How is the AI Infrastructure Readiness Score calculated?

Each datacenter region receives a 0-100 readiness score based on five weighted components: compute density (25%), energy security (25%), construction velocity (20%), water sustainability (15%), and regulatory climate (15%). Nordics ranks #1 with a score of 68, while Singapore ranks last at 20.

How much are hyperscalers spending on AI infrastructure?

The top 5 hyperscalers are investing a combined $283B in 2025 capital expenditure, primarily on AI datacenter infrastructure. Microsoft leads at $80B, followed by Amazon at $75B. This represents the largest infrastructure buildout since the transcontinental railroad.

What are the biggest bottlenecks in the AI supply chain?

The top constraints are: Foundry concentration, Hardware demand, Energy strain. Significant bottlenecks in AI compute supply chain. The supply chain flows from foundries (fabrication) through chips, memory, systems, and energy. Each layer has its own pressure index.

Which region is best for building an AI datacenter?

Based on the AI Infrastructure Readiness Score, Nordics (Sweden/Finland/Norway) ranks #1 at 68/100, with strengths including 92% carbon-free power, Low PPA pricing ($25/MWh), 9 months free cooling. Texas · ERCOT ranks #2 at 62/100. Constrained regions include UAE · Gulf States, Netherlands · Frankfurt, Ireland, Singapore.

How does this data update?

Currently updated manually from public sources (earnings reports, utility filings, sustainability reports, analyst estimates). Data quality rules: "TBD" for unknown values, never guessed. Future phases will automate via Supabase + scheduled scrapers pulling from SEC EDGAR, TrendForce, utility filings, and hyperscaler earnings transcripts.

Explore every layer of the AI infrastructure stack