OperationalConditionalprofile from priorscurated
Project Rainier
Amazon / Anthropic · New Carlisle, Indiana, United States · 2,200 MW · online 2026
Cluster of data halls serving Anthropic training workloads.
Supply as reported: Grid, largely coal and gas in the regional mix; renewable PPAs
Build cost
$73.0B
Months to live
75
gated by 14 construction phases
EBITDA / yr
$22.1B
Payback
3.3 yrs
IRR
13.9%
CO₂ / yr
7,304 kt
476 g/kWh
Jobs
2,000 / 1,605
build / permanent
Composite
52
E 71 · Env 20 · C 68
Conditionalon “Baseline as reported” (60% on-site · 40% grid) · Oct 30, 2026 · meta/muse-wonder-1.0
Project Rainier is CONDITIONAL, not a go as proposed and not a rejection. As proposed at 1,760 MW IT and about 2,077 MW facility, it earns 14% IRR, about 3 years payback and about $12B NPV, but its environment score of 20 misses the 45 floor and grid draw at 138% of regional headroom breaches the 60% grid-load cap. The Gas + solar alternative is the only viable mix shown, lifting environment to 64 and community to 73, cutting emissions from about 7 Mt per year to about 3 Mt per year, and clearing payback and IRR floors albeit narrowly. It requires 2,586 ha versus 168 ha measured open, so land assembly is the critical gating item. On that mix, with phased supply-gated halls, the verdict can hold.
Conditions
- Adopt the viable Gas + solar mix at 50% gas and 50% solar, which lifts the environment score to 64 against the 45 floor and removes grid draw against the 60% grid-load-share cap, instead of proceeding as proposed.
- Assemble land well beyond the 168 ha measured open at the New Carlisle site, as the viable mix needs 2,586 ha in total, with secured title or leases before financial close.
- Hold returns above the floors on the viable mix, with payback inside 7 years and IRR at or above 8%, noting the viable case sits close to that IRR floor at about 8% and about 4 years payback.
- Keep the community score above the 40 floor through binding local agreements, building on the 73 score modelled for the viable mix.
Risks
- Land assembly is the largest execution risk, as the viable mix needs 2,586 ha against 168 ha measured open and failure would stall or fragment the 14-phase build.
- Returns on the viable mix are thin, at about 8% IRR against an 8% floor, so capex overrun on $73B, power cost or GPU price decay could breach the floor.
- Emissions remain material at about 3 Mt per year from 50% gas, inviting disclosure scrutiny and future carbon cost even though the environment floor is cleared.
- The roughly 75-month, 14-phase timetable depends on gas and solar delivery staying firm enough to sustain about 69% utilization, and slippage delays revenue while refresh cost lands in year five.
Scores
71economic
20environment
68community
For the executive
For Amazon / Anthropic leadership this is a buildable 1,760 MW B200 campus at 80 kW density, but only on the Gas + solar mix rather than as proposed, and only once the much larger land requirement is secured. Keep the 2,500 peak crew and 20% construction automation, gate each hall on firm supply, and expect a thinner return than the headline proposal case.For the policy maker
For government officials the conditional case avoids the proposal's grid draw at 138% of regional headroom by using no grid supply, while still creating about 2,000 construction jobs and 1,605 permanent jobs. Approval should be tied to land assembly, emissions disclosure for about 3 Mt per year, and enforceable noise, traffic and local-benefit commitments.For the investor
The proposal case shows 14% IRR, about 3 years payback and $12B NPV over ten years, but it fails the environment and grid floors and is not investable as drawn. The viable Gas + solar case is nearer 8% IRR and about 4 years payback, so returns clear the minimums with little headroom and $73B capex plus a year-five GPU refresh must be underwritten on that basis.How to build it so the verdict holds
Build the 1,760 MW, 2,077 MW facility as 14 phases of 148 MW using the viable 50% gas and 50% solar mix, B200 GPUs at 80 kW direct-liquid density, a 2,500 peak crew and 20% automation over about 75 months. Do not energize halls ahead of firm on-site supply, and make land assembly and binding community commitments conditions precedent so environment, grid and community floors hold.
Configuration
| Supply mix | 50% on-site · 50% solar |
| IT load | 1,760 MW |
| Hardware | B200 · 80 kW racks |
| Crew · automation | 2,500 · 20% |
| Plan scores | 63 · E 57 · Env 64 · C 73 |
| Live in · payback | 75 mo · 4.1 yrs |
| Clears the floors | yes |
Phasing
- Permitting, land assembly for the full 2,586 ha requirement and community agreement before land close and financial close.
- On-site gas and solar supply procurement in parallel with civil works, gated on executed supply, land and interconnection-free delivery for the no-grid mix.
- Phase 1: 148 MW hall, racks energized only as firm supply lands and commissioning is passed.
- Phases 2 to 7: successive 148 MW halls staggered as in the proposal, each gated on supply delivery, cooling performance at 80 kW density and community compliance.
- Phases 8 to 14: remaining 148 MW halls, each gated on the prior hall reaching target utilization, emissions reporting and local hiring and liaison commitments being met.
Community commitments
- Local-first hiring for construction trades, tied to the 9,000-person local labor pool, with published targets, apprenticeships and a community liaison.
- Binding heat-reuse or water-offset agreement with the municipality, with annual public reporting on water use and local benefit.
- Published noise and traffic limits with monitored compliance, complaint handling and remediation, reflecting the proposal noise and traffic indices and protecting sentiment.
- Local spend and tax transparency, reporting local spend and the modelled benefit of about $402M per year in local spend and about $1B per year in tax revenue, with corrective action if delivery falls short.
Energy mixes the advisor searched
| Mix | Viable | Score | Payback | IRR | CO₂ kt/yr |
|---|---|---|---|---|---|
| Gas + solar 50% on-site · 50% solar | yes | 63 | 4.1 yrs | 8% | 3,190 |
| Renewables + SMR 50% small · 30% solar · 20% wind | no | 64 | 4.4 yrs | 7% | 215 |
| Hydro PPA + grid 60% hydro · 40% grid | no | 58 | 3.4 yrs | 13% | 3,575 |
| SMR long-term 70% small · 30% grid | no | 57 | 3.7 yrs | 11% | 2,700 |
| Solar + wind + grid backup 40% solar · 30% grid · 30% wind | no | 55 | 4.4 yrs | 6% | 2,546 |
| Gas bridge 70% on-site · 30% grid | no | 53 | 3.3 yrs | 14% | 7,086 |
| Grid + wind 60% grid · 40% wind | no | 53 | 3.8 yrs | 10% | 4,993 |
| As proposed (60% on-site · 40% grid) 60% on-site · 40% grid | no | 52 | 3.3 yrs | 14% | 7,304 |
| Grid only 100% grid | no | 51 | 3.3 yrs | 14% | 8,615 |
| Grid + solar 60% grid · 40% solar | no | 51 | 3.9 yrs | 9% | 5,042 |
Energy what-if
Energy supply mix
Sliders are weights; shares are normalized to 100%.
Live estimate
vs. baselineComposite
52
Economic
71
Environment
20
Community
68
| Metric | This mix | Baseline |
|---|---|---|
| Build cost | $73.0B | $73.0B |
| Months to live | 75 | 75 |
| Blended power | $0.055/kWh | $0.055/kWh |
| Utilization | 69% | 69% |
| EBITDA / yr | $22.1B | $22.1B |
| Payback | 3.3 yrs | 3.3 yrs |
| IRR | 13.9% | 13.9% |
| CO₂ / yr | 7,303,986 t | 7,303,986 t |
| Carbon intensity | 476 g/kWh | 476 g/kWh |
| Water / yr | 28,584,640 m³ | 28,584,640 m³ |
| Grid draw vs headroom | 138% | 138% |
| Land | 328 ha | 328 ha |
| Permanent jobs | 1,605 | 1,605 |
| Nuisance index | 51 | 51 |
- Measured open land is 168 ha; the facility and its supply need 328 ha, so land must be assembled beyond the site (sam-3.1).
- Majority of supply is high-carbon; expect scrutiny under disclosure rules.
- Grid draw is 138% of regional headroom; rate impacts for other customers are likely.
Region profile country priors
| Industrial power | $0.065/kWh |
| Grid carbon | 560 g/kWh |
| Interconnect lead | 22 months |
| Grid headroom | 600 MW |
| Grid reliability | 96% |
| Construction trades | 9,000 @ $40/h |
| Land index | 1.00× |
| Sentiment | 58 / 100 |
| Climate · water | temperate · low stress |
| Solar · wind | 0.60 · 0.60 |
| Geothermal · flare gas | no · yes |
| People within 5 km | unknown |
Baseline build cost
Site measured by sam-3.1 over a 1.4 km view: 168 ha open land, 0 ha built, 0 ha water, 0 ha solar · Built as 14 phases of 148 MW, staggered 4 months · PUE 1.18 from 80 kW · direct liquid in a temperate climate · GPU-hour rate $3.20 decaying 7%/yr per generation; utilization 69% from supply firmness and fill · Blended power $0.055/kWh, 476 gCO2/kWh · Ten-year cash flow at 10% discount with a GPU refresh in year 5
Scenarios
| Name | Mix | Score | Payback | Verdict | |
|---|---|---|---|---|---|
| Baseline as reportedbaseline | 60% on-site · 40% grid | 52 | 3.3 yrs | Conditional |
Loading map…