Under constructionConditionalprofile from priorscurated

Prometheus

Meta · New Albany, Ohio, United States · 1,000 MW · online 2026

AI training cluster at Meta's existing New Albany footprint.

Supply as reported: Grid with on-site gas

about.fb.com

Open in the simulation

Build cost

$33.5B

Months to live

gated by 7 construction phases

EBITDA / yr

$10.0B

Payback

3.3 yrs

IRR

13.6%

CO₂ / yr

2,929 kt

420 g/kWh

Jobs

2,000 / 735

build / permanent

Composite

E 73 · Env 30 · C 59

Conditionalon “Baseline as reported” (60% on-site · 40% grid) · Oct 30, 2026 · meta/muse-wonder-1.0

Prometheus in New Albany, Ohio is CONDITIONAL: do not build it as proposed, but it can proceed if it switches to the one viable alternative searched. As proposed, with 60% gas and 40% grid for the 800 MW B200 facility, it earns 14% IRR and payback in just over 3 years, yet fails the floors with a 30-point environment score against a 45-point minimum, about 3 Mt of CO2 per year, and grid draw at 63% of regional headroom against a 60% maximum. The SMR long-term alternative — 30% grid and 70% nuclear — lifts the environment score to 53, the community score to 66 and the composite to 61, cuts emissions to about 933 kt per year, and still delivers 11% IRR with payback just under 4 years. That trade is acceptable against an 8% IRR floor and a 7-year payback ceiling, provided land, licensing and phased firm power are secured first. The $34B build therefore only holds together on the SMR mix, not on gas-led economics.

Conditions

Switch from the proposed 60% on-site gas and 40% grid mix to the viable SMR long-term mix of 30% grid and 70% nuclear, which clears the 45-point environment floor at 53 points and keeps grid draw within the 60% maximum.
Assemble and permit land beyond the 35 ha of measured open land — 185 ha is needed for the facility and its SMR supply — before full construction proceeds.
Lock SMR supply, licensing and delivery milestones so each 135 MW phase is only energised when firm low-carbon power is available.
Maintain the measured community measures below so the community score stays well above its 40-point floor, as it does at 66 points under the SMR alternative.

Risks

Land is the first hard constraint: only 35 ha of open land is measured, while the SMR configuration needs 185 ha, so assembly or permitting failure could strand phased halls.
SMR licensing, procurement and delivery could slip behind the seven-phase, roughly 46-month programme, leaving built capacity without firm low-carbon power.
Even on the SMR path annual emissions remain about 933 kt and the 30% grid share will draw disclosure and rate-impact scrutiny from regulators and other grid customers.
Construction and operations depend on the base-case assumptions of 69% utilisation, $3 GPU-hour pricing decaying 7% per year, a year-5 GPU refresh and the $34B capex holding, any of which could push returns toward the 8% floor.
Community tolerance could erode from noise, traffic and water use despite a 59-point community score as proposed and 66 points under SMR, especially while sentiment is already softening.

Scores

73economic

30environment

59community

For the executive

For Meta's leadership, this is a buildable 800 MW B200 cluster only on the SMR path, with 11% IRR and payback in just under 4 years rather than the stronger but non-compliant economics of gas. Keep the 80 kW liquid-cooled design and 46-month, seven-phase programme, but do not let halls outrun licensed nuclear supply or land assembly.

For the policy maker

For New Albany and Ohio officials, the conditional path cuts annual emissions from about 3 Mt to about 933 kt, lifts the environment score from 30 to 53, and avoids making grid draw the dominant claim on regional headroom. In return for approvals, require enforceable local hiring, water, noise and traffic commitments and transparent disclosure of the power mix and rate impacts.

For the investor

The SMR case still clears the 8% IRR floor at 11%, with payback under 4 years against the 7-year maximum, despite giving up return versus the 14% IRR proposed gas case. Underwrite the $34B capex, year-5 GPU refresh and 69% utilisation assumptions in the base proposal, and price in SMR licensing and land-assembly delay risk.

How to build it so the verdict holds

Build the 800 MW, 944 MW facility as a seven-phase B200 cluster at 80 kW density, powered 30% by grid and 70% by SMR nuclear, with first power on the roughly 46-month programme only as firm low-carbon supply lands. This keeps the viable alternative's profile — 53-point environment score, 66-point community score, 11% IRR and payback just under 4 years — rather than the non-compliant gas-led proposal.

Play this plan in the simulation

Configuration

Supply mix	70% small · 30% grid
IT load	800 MW
Hardware	B200 · 80 kW racks
Crew · automation	2,500 · 20%
Plan scores	61 · E 66 · Env 53 · C 66
Live in · payback	46 mo · 3.7 yrs
Clears the floors	yes

Phasing

Permitting, SMR licensing pathway and community agreement before land close — gated by assembled land for the full 185 ha need and approval of the 30% grid and 70% nuclear mix.
On-site SMR supply procurement and grid interconnection in parallel with civil works — gated by firm supply contracts and confirmation grid draw stays within the 60% floor.
Phase 1: 135 MW hall, racks energised as supply lands — gated by first firm power, commissioning and community monitoring being live.
Phases 2 to 7: each further 135 MW hall staggered as in the base programme — each gated by available firm low-carbon power, GPU delivery, and noise, traffic and water compliance from the prior phase.

Community commitments

Local-first hiring for construction trades, tied to the 2,000 construction jobs and 735 permanent jobs in the base estimate, with published local-hiring and local-spend reporting.
Heat reuse or water offset agreement with the municipality, reflecting about 13 million cubic metres of annual water use in the base proposal and the surface water measured 58 m from the site.
Published noise and traffic limits with a community liaison, independent monitoring, and pause-and-remediate triggers to protect the 66-point community score under the SMR alternative.
Grid rate-impact mitigation and public disclosure agreement covering the 30% grid share, so other customers are not left bearing unassigned upgrade costs.
Land-assembly and host-community benefit agreement agreed before land close for the area beyond the 35 ha of measured open land.

Energy mixes the advisor searched

Mix	Viable	Score	Payback	IRR	CO₂ kt/yr
SMR long-term 70% small · 30% grid	yes	61	3.7 yrs	11%	933
Renewables + SMR 50% small · 30% solar · 20% wind	no	62	4.4 yrs	7%	98
Gas + solar 50% on-site · 50% solar	no	61	4.1 yrs	8%	1,450
Hydro PPA + grid 60% hydro · 40% grid	no	60	3.5 yrs	13%	1,239
Solar + wind + grid backup 40% solar · 30% grid · 30% wind	no	57	4.4 yrs	6%	890
Gas bridge 70% on-site · 30% grid	no	56	3.3 yrs	14%	2,928
As proposed (60% on-site · 40% grid) 60% on-site · 40% grid	no	54	3.3 yrs	14%	2,929
Grid only 100% grid	no	53	3.3 yrs	14%	2,937
Grid + wind 60% grid · 40% wind	no	52	3.8 yrs	10%	1,710
Grid + solar 60% grid · 40% solar	no	51	3.9 yrs	9%	1,735

Energy what-if

Energy supply mix

Grid40%On-site gas60%Solar + storage0%Wind + storage0%Small modular reactor0%Geothermalno mapped resource0%Hydro PPA0%

Sliders are weights; shares are normalized to 100%.

IT load (MW)GPU classRack densityConstruction crewAutomation share 20%

Live estimate

vs. baseline

Composite

Economic

Environment

Community

Metric	This mix	Baseline
Build cost	$33.5B	$33.5B
Months to live	46	46
Blended power	$0.055/kWh	$0.055/kWh
Utilization	69%	69%
EBITDA / yr	$10.0B	$10.0B
Payback	3.3 yrs	3.3 yrs
IRR	13.6%	13.6%
CO₂ / yr	2,929,406 t	2,929,406 t
Carbon intensity	420 g/kWh	420 g/kWh
Water / yr	12,993,018 m³	12,993,018 m³
Grid draw vs headroom	63%	63%
Land	149 ha	149 ha
Permanent jobs	735	735
Nuisance index	41	41

Measured open land is 35 ha; the facility and its supply need 149 ha, so land must be assembled beyond the site (sam-3.1).
Majority of supply is high-carbon; expect scrutiny under disclosure rules.
3 ha of surface water measured 58 m from the site; cooling make-up supply is plausible (sam-3.1).
Grid draw is 63% of regional headroom; rate impacts for other customers are likely.

Region profile country priors

Industrial power	$0.065/kWh
Grid carbon	420 g/kWh
Interconnect lead	24 months
Grid headroom	600 MW
Grid reliability	96%
Construction trades	11,000 @ $42/h
Land index	1.40×
Sentiment	58 / 100
Climate · water	temperate · low stress
Solar · wind	0.60 · 0.60
Geothermal · flare gas	no · yes
People within 5 km	unknown

sam-3.1 segmentation of Esri World Imagery

Baseline build cost

Site measured by sam-3.1 over a 1.4 km view: 35 ha open land, 13 ha built, 3 ha water 58 m away, 0 ha solar · Built as 7 phases of 135 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, 420 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	54	3.3 yrs	Conditional

Loading map…