#23 of 30 · Inference chipmaker & cloud
Groq
Claims a tenth of the energy per token; measured by no one; now licensed to NVIDIA.
In plain words
How scores workGroq scores 37 out of 100 — laggard. Little or no disclosure, fossil-backed capacity, or active disputes with regulators and communities. Confidence in this score is little data (D).
What helps
- Efficiency · SRAM-based LPU architecture; Groq claims up to 10× the speed at a tenth of the energy of GPUs, and picked Helsinki for 'high-efficiency, low-cost inference'.
- Carbon · Helsinki sits on one of Europe's lowest-carbon grids.
What hurts
- Efficiency · No independent per-token energy measurement exists.
- Carbon · Dammam runs on Saudi Arabia's gas-fired grid; US sites on the grid with no procurement disclosed.
- Water · No water data; colocation sites in Finland, Saudi Arabia, Australia and the US.
What nobody knows
- Per-prompt energy disclosed
- Per-prompt water disclosed
- Training emissions in model cards
- Annual environmental report
- Location-based Scope 2 reported
- +3 more
Groq's LPU serves open models at very high speed from its own GroqCloud sites in the United States, Dammam, Saudi Arabia (backed by a $1.5 billion Saudi commitment), Helsinki and Sydney. It claims up to ten times the speed at a tenth of the energy of GPUs and has published no measurement to support it. In December 2025 NVIDIA paid a reported $20 billion to license Groq's technology and hire its founders; Groq continues as an independent company with no environmental report.
Key figures
Per-prompt and fleet metrics. Hover a title for what it means; open “How we got this number” for the working.
ecogpt.co estimate; no company figure.
How we got this number
GroqCloud serves 70B-class open models (GPU baseline ≈0.25 Wh). Groq claims a tenth of the energy per token; with no independent measurement we credit a 2.5× improvement. Uncertainty roughly ±3×.
On-site cooling only, so comparable with Google and OpenAI's figures.
How we got this number
0.1 Wh per prompt × 0.5 L/kWh (estimated WUE for colocation sites across four countries). Counting the water evaporated to generate the electricity adds ≈0.18 mL.
Location-based: what the grid actually emitted. Renewable certificates would bring the market-based figure near zero.
How we got this number
0.1 Wh × 480 gCO₂/kWh (world average grid, location-based). Dammam's grid is almost entirely gas; Helsinki's is largely carbon-free.
Location-based estimate for the AI fleet only; not comparable with a full corporate inventory.
How we got this number
60 MW × 8,760 h × 80% utilisation × 480 gCO₂/kWh (world average grid). Compute only — nothing else the company does is counted.
How we got this number
Groq said in October 2025 it would open more than a dozen data centres in the following year, several times its 2025 footprint.
How we got this number
No disclosure. Equinix colocation in Helsinki and Sydney plus desert sites in Dammam; 0.5 L/kWh is a blended assumption.
How we got this number
Saudi Arabia's grid is about 1% renewable, the US about 40% non-fossil and Finland about 90%; weighted by where Groq's capacity sits.
Hardware & materials
The chips, power, copper and future e-waste behind the fleet — the part sustainability reports leave out
GPUs, TPUs or custom accelerators the company runs. Analyst figures are in H100-equivalents, company figures in physical chips.
How we got this number
Groq publishes no count. A $1.5bn Saudi commitment at roughly $20,000 per deployed LPU, plus US and Helsinki sites and a plan for 'more than a dozen' new data centres, points to around 100,000 chips. ±2×.
Megawatts the AI fleet pulls. One megawatt runs roughly 750 US homes.
How we got this number
LPU racks draw far less per chip than GPU racks; ≈100,000 chips at about 0.5 kW all-in, across Dammam, Helsinki, Sydney and US sites.
Terawatt-hours per year. One TWh is a year of electricity for about 95,000 US homes.
How we got this number
60 MW × 8,760 h × 80% utilisation. Compute fleet only.
CO₂e emitted making the chips and servers before they answer a single prompt.
How we got this number
100,000 accelerators × 514 kg CO₂e each (chip plus share of the host server). Emitted in manufacturing, before the first prompt is served.
Tonnes of server hardware that will be retired in three to five years.
How we got this number
100,000 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.
Tonnes of copper in cabling, busbars and transformers.
How we got this number
60 MW × 27 t of copper per MW (cabling, busbars, transformers).
Diesel backup units or gas turbines at the data centres, permitted or observed.
How we got this number
60 MW ÷ 2.6 MW per generator, the average size permitted at US data centres. Permit counts are not published fleet-wide; this is the number a fleet this size normally carries.
Share of retired servers and parts given a second life.
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
- SRAM-based LPU architecture; Groq claims up to 10× the speed at a tenth of the energy of GPUs, and picked Helsinki for 'high-efficiency, low-cost inference'. Data Center DynamicsTom's Hardware
- No independent per-token energy measurement exists. Tom's Hardware
- Dammam runs on Saudi Arabia's gas-fired grid; US sites on the grid with no procurement disclosed. Reuters via Yahoo Finance
- Helsinki sits on one of Europe's lowest-carbon grids. Data Center Dynamics
- No water data; colocation sites in Finland, Saudi Arabia, Australia and the US. Data Center Dynamics
- No environmental report, no chip count, no energy measurement behind the efficiency claim. Tom's Hardware
- No targets. Data Center Dynamics
- No enforcement actions on record. Data Center Dynamics
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
- Mountain View HQCalifornia, USA
- DammamSaudi Arabia · $1.5bn Saudi commitment; gas grid.
- Helsinki (Equinix)Finland · Low-carbon grid.
- Sydney (Equinix)Australia
Timeline
Incidents and developments involving this company
