Under constructionConditionalprofile from priorscurated
Fairwater
Microsoft · Mount Pleasant, Wisconsin, United States · 900 MW · online 2026
AI datacenter on the former Foxconn site.
Supply as reported: Grid with utility solar expansion
Build cost
$30.9B
Months to live
42
gated by 6 construction phases
EBITDA / yr
$7.9B
Payback
3.9 yrs
IRR
9.2%
CO₂ / yr
1,801 kt
301 g/kWh
Jobs
2,000 / 663
build / permanent
Composite
53
E 63 · Env 30 · C 68
Conditionalon “Baseline as reported” (65% grid · 35% solar) · Oct 30, 2026 · meta/muse-wonder-1.0
The verdict is conditional, not a go as proposed: the 65% grid and 35% solar proposal scores 30 on environment against a 45 floor and draws 110% of regional grid headroom against a 60% ceiling, so it cannot proceed in that form. The search found one viable alternative, the SMR long-term mix of 70% nuclear and 30% grid, which lifts the environment score to 51, the composite to 63 and the community score to 73, with an 11% IRR and payback inside 4 years. That configuration also cuts annual emissions from about 1.8 Mt to about 0.9 Mt and reduces the grid draw to 51% of headroom, though rate impacts remain likely. Economics stay demanding at roughly $31B of capex for the proposed build, 504,000 GPUs and about $8.6B a year of revenue at 61% utilization. The conditions below therefore lock in the SMR mix, the land, the grid agreement and the community commitments before and during the phased build.
Conditions
- Adopt the viable SMR long-term supply mix — 70% nuclear and 30% grid — in place of the proposed 65% grid and 35% solar mix, lifting the environment score from 30 to 51, above the 45 floor, and cutting emissions from about 1.8 Mt a year to about 0.9 Mt a year.
- Assemble land beyond the 41 ha measured open on site to the 167 ha the SMR configuration needs, with permitting and a signed community agreement completed before land close.
- Secure firm SMR licensing, procurement and delivery commitments before racks are energized in any phase, so no hall defaults back to majority high-carbon grid supply.
- Lock a grid interconnection and tariff agreement for the reduced grid share, which still draws 51% of regional headroom, with protections against rate impacts for other customers.
- Hold the community score above the 40 floor — the SMR configuration is scored at 73 — by funding and publishing the community measures in the plan for the life of construction.
Risks
- SMR licensing or delivery slips beyond the roughly 42-month, six-phase schedule, leaving halls built but unable to energize on low-carbon supply.
- Only 41 ha of open land is measured on site against 167 ha needed for the viable configuration, so land assembly could delay or shrink the project.
- Even at 51% of regional headroom, the grid share may raise rates for other customers and attract disclosure and regulatory scrutiny, especially if any fallback to grid-heavy supply occurs.
- Capital intensity is high at about $31B with roughly negative $1B of ten-year NPV on the proposal's assumptions, leaving returns sensitive to utilization, GPU-hour price decay and the year-5 GPU refresh.
- Construction at a 2,500-person peak crew brings noise and traffic pressure, and any failure on local hiring, water offsets or published limits could erode the community score toward its 40 floor.
- Annual water use of about 11 million cubic metres and residual emissions of about 0.9 Mt a year will face ongoing public and disclosure scrutiny.
Scores
63economic
30environment
68community
For the executive
For Microsoft's leadership, this is a conditional build at 720 MW of IT load with B200 GPUs at 80 kW density, not a build as proposed, because the grid-and-solar mix fails the environment floor and overdraws regional grid headroom. The SMR-led mix restores a viable configuration with an 11% IRR and payback inside 4 years, but only if SMR delivery is contracted before halls are energized.For the policy maker
For government officials, the conditional case cuts annual emissions roughly in half versus the proposal and keeps the grid draw at 51% of regional headroom, within the 60% ceiling, rather than 110% as proposed. Approval should still be tied to land assembly beyond the measured site, ratepayer protections, and enforceable noise, traffic and local-benefit commitments given construction traffic and the site's history.For the investor
Investors should underwrite the SMR configuration, scored at 11% IRR with payback inside 4 years, rather than the proposal at 9% IRR, about $31B of capex and roughly negative $1B of ten-year NPV on the stated assumptions. Returns depend on holding utilization, GPU-hour pricing and the phased 42-month schedule, with the year-5 GPU refresh already in the cash-flow assumptions.How to build it so the verdict holds
Build the 720 MW IT facility, 850 MW in total, in six staggered phases with B200 GPUs at 80 kW density and direct-liquid cooling, powered by the viable SMR long-term mix of 70% nuclear and 30% grid. Keep the 2,500-person peak crew with 20% construction automation, and gate every hall's energization on firm low-carbon supply actually landing so the verdict's floors hold.
Configuration
| Supply mix | 70% small · 30% grid |
| IT load | 720 MW |
| Hardware | B200 · 80 kW racks |
| Crew · automation | 2,500 · 20% |
| Plan scores | 63 · E 67 · Env 51 · C 73 |
| Live in · payback | 42 mo · 3.7 yrs |
| Clears the floors | yes |
Phasing
- Permitting and community agreement before land close, including assembly of land beyond the 41 ha measured open toward the 167 ha required.
- SMR licensing and on-site supply procurement, plus the grid interconnection and tariff agreement, in parallel with civil works and gated before rack energization.
- Phase 1: 142 MW facility hall, racks energized only as contracted SMR and grid supply lands.
- Phase 2: 142 MW facility hall, gated on Phase 1 supply performance and published noise and traffic compliance.
- Phase 3: 142 MW facility hall, gated on continued SMR delivery milestones and community-score reporting.
- Phase 4: 142 MW facility hall, gated on grid-draw compliance within the agreed headroom share.
- Phase 5: 142 MW facility hall, gated on heat-reuse or water-offset agreement delivery.
- Phase 6: 142 MW facility hall, gated on full supply firmness at the 70% nuclear and 30% grid mix before final energization.
Community commitments
- Local-first hiring for construction trades, tied to the 2,000 construction jobs and 663 permanent jobs in the estimate, with published local hiring and local spend reporting.
- Heat reuse or water offset agreement with the municipality, addressing about 11 million cubic metres of annual water use despite low local water stress.
- Published noise and traffic limits with a named community liaison, regular public reporting against those limits, and a remediation process for breaches.
- Binding ratepayer-protection and disclosure commitments for the 30% grid share, including public reporting of carbon intensity and grid draw.
- Community benefit reporting against the scored community outcome, maintaining a buffer above the 40 floor through construction and into operations.
Energy mixes the advisor searched
| Mix | Viable | Score | Payback | IRR | CO₂ kt/yr |
|---|---|---|---|---|---|
| SMR long-term 70% small · 30% grid | yes | 63 | 3.7 yrs | 11% | 896 |
| Renewables + SMR 50% small · 30% solar · 20% wind | no | 64 | 4.4 yrs | 6% | 88 |
| Gas + solar 50% on-site · 50% solar | no | 63 | 4.1 yrs | 8% | 1,297 |
| Hydro PPA + grid 60% hydro · 40% grid | no | 62 | 3.4 yrs | 13% | 1,190 |
| Gas bridge 70% on-site · 30% grid | no | 58 | 3.3 yrs | 14% | 2,692 |
| Solar + wind + grid backup 40% solar · 30% grid · 30% wind | no | 58 | 4.5 yrs | 6% | 848 |
| Grid only 100% grid | no | 55 | 3.4 yrs | 13% | 2,832 |
| Grid + wind 60% grid · 40% wind | no | 55 | 3.8 yrs | 9% | 1,647 |
| As proposed (65% grid · 35% solar) 65% grid · 35% solar | no | 53 | 3.9 yrs | 9% | 1,801 |
| Grid + solar 60% grid · 40% solar | no | 53 | 4.0 yrs | 9% | 1,662 |
Energy what-if
Energy supply mix
Sliders are weights; shares are normalized to 100%.
Live estimate
vs. baselineComposite
53
Economic
63
Environment
30
Community
68
| Metric | This mix | Baseline |
|---|---|---|
| Build cost | $30.9B | $30.9B |
| Months to live | 42 | 42 |
| Blended power | $0.054/kWh | $0.054/kWh |
| Utilization | 61% | 61% |
| EBITDA / yr | $7.9B | $7.9B |
| Payback | 3.9 yrs | 3.9 yrs |
| IRR | 9.2% | 9.2% |
| CO₂ / yr | 1,801,242 t | 1,801,242 t |
| Carbon intensity | 301 g/kWh | 301 g/kWh |
| Water / yr | 10,884,652 m³ | 10,884,652 m³ |
| Grid draw vs headroom | 110% | 110% |
| Land | 767 ha | 767 ha |
| Permanent jobs | 663 | 663 |
| Nuisance index | 46 | 46 |
- Measured open land is 41 ha; the facility and its supply need 767 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 110% of regional headroom; rate impacts for other customers are likely.
Region profile country priors
| Industrial power | $0.075/kWh |
| Grid carbon | 450 g/kWh |
| Interconnect lead | 22 months |
| Grid headroom | 500 MW |
| Grid reliability | 96% |
| Construction trades | 8,000 @ $40/h |
| Land index | 1.10× |
| Sentiment | 56 / 100 |
| Climate · water | temperate · low stress |
| Solar · wind | 0.50 · 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: 41 ha open land, 0 ha built, 0 ha water, 0 ha solar · Built as 6 phases of 142 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 61% from supply firmness and fill · Blended power $0.054/kWh, 301 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 | 65% grid · 35% solar | 53 | 3.9 yrs | Conditional |
Loading map…