AI Infrastructure SimCity
Register a site
AI Infrastructure SimCity

Every data center that exists or is coming, what it will cost and emit, whether it should be built, and a simulation to build it in.

Research agents keep a registry of current and announced facilities from public sources. An estimation engine turns each one into economics, environmental footprint and community impact. An advisor searches energy mixes, issues a go, conditional or no-go verdict, and writes a build plan you can play through in the simulation.

Sites tracked
27
29,600 MW at full build-out
Operational
10
11,600 MW
In pipeline
17
18,000 MW
Verdicts
32 / 50 / 21
go / conditional / no go
For the
Executive

Where to build next, what it costs, how long until first revenue, and which sites the advisor would not sign off on.

  • · Portfolio capex and payback
  • · Critical path by site
  • · Open conditions
For the
Policy maker

What each facility does to the grid, the water table, the air, and the local economy, and whether the operator has a viable path.

  • · Jobs, tax base, local spend
  • · Carbon, water, grid share
  • · Nuisance and sentiment
For the
Investor

Risk-adjusted ranking of the pipeline with live sensitivity to power prices and utilization, and the verdict behind each number.

  • · IRR, NPV, payback
  • · Price and utilization sliders
  • · Go / no-go with conditions
Recently updated sites
All sites
SiteStatusMWCapexVerdict
Stargate Abilene
Abilene, Texas, United States
Under construction1,200$40.6BGo
Colossus
Memphis, Tennessee, United States
Operational300$10.0BConditional
Hyperion
Richland Parish, Louisiana, United States
Under construction2,000$67.9BGo
Prometheus
New Albany, Ohio, United States
Under construction1,000$33.5BConditional
Project Rainier
New Carlisle, Indiana, United States
Operational2,200$73.0BConditional
Fairwater
Mount Pleasant, Wisconsin, United States
Under construction900$30.9BConditional
Loudoun County cluster
Ashburn, Virginia, United States
Operational5,000$172.1BNo go
Council Bluffs
Council Bluffs, Iowa, United States
Operational700$23.8BConditional
How a verdict is produced
  1. 1
    Discover. Kimi plans web searches, Tavily runs them, and the model extracts name, operator, location, status, capacity, energy supply and sources.
  2. 2
    Enrich. The same pair establishes the local factors: power price, grid carbon and headroom, interconnection lead time, labor, land, water stress, nearby population, sentiment and any moratorium.
  3. 3
    Estimate. A deterministic engine computes capex, schedule, revenue, carbon, water, grid share, jobs, tax base and nuisance for a hardware spec and an energy mix. Change the mix and it recomputes.
  4. 4
    Advise. The engine searches alternative mixes against hard floors and fixes the verdict. GLM 5.3 writes the rationale, conditions, risks, a brief per audience, and a build plan that opens in the simulation.