Skip to content

#15 of 30 · AI cloud

Oracle Cloud (OCI)

100% carbon-free for the public cloud; the AI campuses run on grid plus gas.

Austin, Texas·oracle.com ·Verified Oct 16, 2026
BGood confidence
Index score
Good confidence

In plain words

How scores work

Oracle scores 47 out of 100 — middle of the pack. Some disclosure or clean-energy commitments, offset by rising emissions or blind spots. Confidence in this score is good confidence (B).

What helps

  • Carbon · 100% carbon-free electricity matched annually for offices and public cloud regions; 4.3 Mt avoided since 2020, per Oracle.
  • Water · Closed-loop liquid cooling at Abilene with no evaporation in normal operation; target to halve water use in water-stressed regions by 2035.
  • Commitments · 100% clean energy for AI clusters by 2035; 20% cut in value-chain emissions intensity per MW by 2030; 90% waste diversion and halved water use in stressed regions by 2035.

What hurts

  • Carbon · AI clusters are excluded until a 2035 target; Abilene draws on ERCOT plus ≈360 MW of on-site gas turbines, with 41 more proposed.
  • Water · No fleet WUE disclosed.
  • Commitments · 2035 is five years behind peers' 2030 goals, for the fastest-growing part of the fleet.

What nobody knows

  • Per-prompt energy disclosed
  • Per-prompt water disclosed
  • Training emissions in model cards
  • Water-use effectiveness reported

Oracle Cloud Infrastructure hosts OpenAI's Stargate Abilene (up to 450,000 GB200 GPUs, 1.2 GW) and serves Cohere, xAI and Meta training runs. Oracle reports that its offices and public cloud regions match 100% of electricity with carbon-free sources — but it has carved its fast-growing 'non-public cloud and AI clusters' out of that claim with a 2035 target, and Abilene pairs the Texas grid with one of the largest on-site gas build-outs in the country, permitted with 62 diesel generators and 10 turbines under rules written for small businesses.

Products
OCI Supercluster · OCI Generative AI · Stargate (with OpenAI and SoftBank)
Hosting
Own OCI regions (50+) plus Crusoe-built Stargate campuses (Abilene) and partner capacity.

Key figures

Per-prompt and fleet metrics. Hover a title for what it means; open “How we got this number” for the working.

How to read the labels:Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Measured independentlyMeasured independently. Measured or estimated by researchers or analysts with no stake in the result.ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.— hover any label.
Energy per promptElectricity used to answer one typical request. Under 0.3 Wh is good. A 10 W LED bulb uses 0.3 Wh in about two minutes.
0.34
Wh / typical prompt
≈ 28.3 smartphone charges per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.Sam Altman (blog)Data Center Dynamics

ecogpt.co estimate; no company figure.

How we got this number

Oracle is a host, not a model maker; the dominant workload on OCI's AI clusters is OpenAI's, whose own figure is 0.34 Wh per average query. Uncertainty roughly ±3×.

Water per promptWater evaporated cooling the servers for one request. Scope matters: on-site cooling only, or also the water used to generate the electricity (which adds ≈1.8 mL per Wh).
0.07
mL / typical prompt
≈ 0.14 500 mL bottles per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.Sam Altman (blog)Data Center DynamicsTechRepublic

On-site cooling only, so comparable with Google and OpenAI's figures.

How we got this number

0.34 Wh per prompt × 0.2 L/kWh (estimated WUE: closed-loop at Abilene, undisclosed elsewhere). Counting the water evaporated to generate the electricity adds ≈0.61 mL.

Carbon per promptGreenhouse gas from one request on the grid where it actually runs. Below 0.1 g is clean. A kilometre in a petrol car is about 170 g.
0.143
gCO₂e / typical prompt
≈ 0.84 km driven (petrol car) per 1,000 prompts
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.Sam Altman (blog)Data Center Dynamics

Location-based: what the grid actually emitted. Renewable certificates would bring the market-based figure near zero.

How we got this number

0.34 Wh × 420 gCO₂/kWh (Texas grid plus on-site gas, location-based).

Annual emissionsThe company's whole yearly footprint, in million tonnes of CO₂e. Clouds and device makers dwarf pure labs because they also count shops, logistics or products. Compare the trend, not the size.
3.83
MtCO₂e / year (AI clusters only)
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.Network WorldCleanview

Location-based estimate for the AI fleet only; not comparable with a full corporate inventory.

How we got this number

1300 MW × 8,760 h × 80% utilisation × 420 gCO₂/kWh (Texas grid plus on-site gas). Compute only — nothing else the company does is counted.

Emissions trendDirection of travel against last year or the company's own baseline. Negative is good. Positive means the AI build-out is outrunning clean energy.
+150
% (Abilene from 200 MW to 1.2 GW)
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.Data Center DynamicsCleanview
How we got this number

Abilene energised 200 MW in September 2025 and is building toward 1.2 GW across eight buildings by mid-2026; AI-cluster emissions scale with it.

Water per kWh (WUE)Litres of water consumed per kilowatt-hour of server electricity, across the fleet. 0.15 is excellent, 0.3 is good, 1.8 is the US data-centre average.
0.20
L/kWh
ecogpt.co estimateecogpt.co estimate. Our own estimate from published factors. The working is shown next to the number; treat it as an order of magnitude, not a measurement.TechRepublic
How we got this number

Crusoe's closed-loop cooling at Abilene evaporates no water in normal operation; older OCI regions are undisclosed. 0.2 is a blended assumption.

Clean electricityShare of electricity matched with renewable or carbon-free sources. Annual matching counts certificates bought anywhere; hourly matching on the same grid is far stricter.
92.0
% renewable electricity in the cloud (public cloud regions)
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Oracle

AI clusters excluded; target 100% for them by 2035.

Hardware & materials

The chips, power, copper and future e-waste behind the fleet — the part sustainability reports leave out

Factors we use
AI chips in service
1.0 million
H100-equivalents (Epoch AI, Q4 2025)
≈ 13,889 racks of 72 GPUs
Measured independentlyMeasured independently. Measured or estimated by researchers or analysts with no stake in the result.Network WorldData Center Dynamics

GPUs, TPUs or custom accelerators the company runs. Analyst figures are in H100-equivalents, company figures in physical chips.

How we got this number

Stargate Abilene alone is planned for 450,000 GB200s.

Power draw
1,300
MW
≈ 975,000 US homes' worth of power
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldData Center DynamicsSemiAnalysis

Megawatts the AI fleet pulls. One megawatt runs roughly 750 US homes.

How we got this number

1,000,000 accelerators × 1.3 kW each, all-in. Fleet power implied by the chip count.

Electricity a year
9.11
TWh / year
≈ 867,619 US homes for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldData Center DynamicsSemiAnalysis

Terawatt-hours per year. One TWh is a year of electricity for about 95,000 US homes.

How we got this number

1300 MW × 8,760 h × 80% utilisation. Compute fleet only.

Carbon baked into the hardware
514
ktCO₂e embodied in chips and servers
≈ 111,739 petrol cars driven for a year
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldData Center DynamicsarXiv 2607.05408NVIDIA developer blog

CO₂e emitted making the chips and servers before they answer a single prompt.

How we got this number

1,000,000 accelerators × 514 kg CO₂e each (chip plus share of the host server). Emitted in manufacturing, before the first prompt is served.

Future e-waste
17,000
tonnes of servers to retire within ~5 years
≈ 94 empty Boeing 747s by weight
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldData Center DynamicsNVIDIAThe Register

Tonnes of server hardware that will be retired in three to five years.

How we got this number

1,000,000 accelerators × 17 kg of server hardware each. Tomorrow's e-waste unless reused; generative AI could produce 2.5 Mt of it a year by 2030.

Copper in the build
35,100
tonnes of copper in the build
≈ copper for 422,892 electric cars
DerivedDerived. Arithmetic on a disclosed or independent input using a published conversion factor.Network WorldData Center DynamicsSemiAnalysisBHPS&P Global Market Intelligence

Tonnes of copper in cabling, busbars and transformers.

How we got this number

1,300 MW × 27 t of copper per MW (cabling, busbars, transformers).

Fossil generators on site
72.0
fossil units permitted at Stargate Abilene (62 diesel generators, 10 gas turbines)
≈ 187 MW of backup capacity
Company figureCompany figure. Published by the company itself. We take it as given and record the scope next to it.Latitude Media / FloodlightThe Texas Tribune

Diesel backup units or gas turbines at the data centres, permitted or observed.

How we got this number

Shared with OpenAI; 41 more turbines and 18 more generators proposed.

Hardware reused or recycled
—
No hardware reuse or e-waste figure published.
Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.

Share of retired servers and parts given a second life.

Mineral depletion
—
Only Mistral has published a mineral-depletion figure.
Not disclosedNot disclosed. The company publishes nothing on this and no defensible estimate exists.

Kilograms of antimony-equivalent, the life-cycle unit for scarce minerals consumed.

Score breakdown

Each pillar is 0–100; weights are fixed across the index

Energy efficiency — How much electricity one answer costs.weight 20% · confidence C
50
Carbon intensity — How clean that electricity is, and whether total emissions are rising.weight 25% · confidence B
40
  • 100% carbon-free electricity matched annually for offices and public cloud regions; 4.3 Mt avoided since 2020, per Oracle. Oracle
  • AI clusters are excluded until a 2035 target; Abilene draws on ERCOT plus ≈360 MW of on-site gas turbines, with 41 more proposed. CleanviewOracleLatitude Media / Floodlight
Water — How much water the data centres consume, and whether it is replaced.weight 15% · confidence B
55
  • Closed-loop liquid cooling at Abilene with no evaporation in normal operation; target to halve water use in water-stressed regions by 2035. TechRepublicOracle
  • No fleet WUE disclosed. Oracle
Transparency — How much the company actually publishes.weight 20% · confidence B
48
  • Annual sustainability update with targets and avoided-emissions claims; no WUE, no electricity total, no AI-specific energy. OracleOracle
Commitments — Whether it has dated, measurable climate and water targets.weight 10% · confidence B
55
  • 100% clean energy for AI clusters by 2035; 20% cut in value-chain emissions intensity per MW by 2030; 90% waste diversion and halved water use in stressed regions by 2035. Oracle
  • 2035 is five years behind peers' 2030 goals, for the fastest-growing part of the fleet. Oracle
Conduct — How it behaves: permits, lawsuits, gas turbines, community water.weight 10% · confidence B
40

Models

What this company ships, and what each prompt costs

No model-level entries — this company hosts or supplies others rather than serving its own models.

Where the compute lives

Named sites from company disclosures and press reporting

OperationalConstructionPlanned
  • Stargate AbileneTexas, USA · 450,000 GB200s planned; 1.2 GW; on-site gas.
  • AshburnVirginia, USA
  • PhoenixArizona, USA
  • FrankfurtGermany
  • Stargate Shackelford CountyTexas, USA

Timeline

Incidents and developments involving this company

HighAir & power·Jul 9, 2026

Stargate Abilene adds 41 gas turbines and 62 diesel generators under a permit written for dry cleaners

A Floodlight and Texas Tribune investigation found Stargate's Abilene campus was authorised to run 10 gas turbines and 62 backup diesel generators through Texas 'permits by rule' that need no environmental study, public notice or comment — and has applied to add 41 turbines and 18 generators more.

OpenAI, Oracle, Crusoe