7. Assessment and Calculation
The calculation components and their outputs.
7.1 IT Electricity Usage Calculation
IT Electricity Usage is calculated by mapping each unit of cloud usage to its underlying hardware components and summing the energy contribution of each. The energy for a given unit of usage is built up from the energy of the components identified for that SKU.
Component power figures come from published benchmarks and manufacturer data, as set out in Section 5.2. Server power is aligned with methodologies used by the US EPA, the EU Ecodesign Regulation, and the Standard Performance Evaluation Corporation's server power benchmarks, and accounts for the whole server rather than the processor alone. Cloud usage data states no processor utilisation, so a fixed default operating point is applied to each unit of usage, or the cloud provider's published baseline allowance for a burstable instance.
The electricity of a compute instance, covering its processor, memory, attached drives, and accelerators, is set out in the companion Greenpixie Methodology for Cloud Compute Energy Measurement.
7.2 Facility Energy and Emissions Calculation
Total Facility Energy is calculated by multiplying IT Electricity Usage by the applicable PUE factor for the cloud region. Where a cloud provider does not publish a region-specific PUE value, the best available published value is used, which may be a global average. Usage Emissions are then calculated by multiplying Total Facility Energy by the Grid Carbon Intensity for the relevant region and time period.
Usage Emissions are reported on a location-based basis in accordance with the GHG Protocol Scope 2 Guidance. Greenpixie follows the US EPA's guidance on indirect emissions from purchased electricity, giving precedence to regional or sub-national GCI factors where available.
The Grid Carbon Intensity and Power Usage Effectiveness factors are set out in the companion Greenpixie Methodology for Regional Carbon, Energy, and Water Factors.
7.3 Embodied Emissions Calculation
Embodied emissions are estimated on a component-by-component basis, using hardware specifications for the instance type, semiconductor die area data, life cycle assessment data from Dell's server assessments, NVIDIA's product carbon footprints, the ACT semiconductor model of Gupta et al., open life cycle inventory datasets, Gröger et al., and assumed server operational lifespans based on publicly available cloud provider statements.
Components covered include CPU, RAM, SSD storage, HDD, GPU where applicable, power supply units, chassis and fans, motherboard, expansion cards and assembly.
The embodied emissions of a CPU or GPU die are calculated by multiplying its die area by a coefficient for its process node, taken from the ACT model of Gupta et al. Fabrication energy and process gas per cm² of wafer are higher on newer nodes. A cm² of 5 nm die carries about twice the embodied emissions of a cm² of 28 nm die. The coefficient includes fabrication yield, so it is multiplied by the die area as published.
For CPUs, the die area is the total area of every die in the processor package, being the compute chiplets and, in a chiplet processor, the separate IO die. Die areas come from third-party hardware specification databases. The total die area is multiplied by the die coefficient for the process node of the compute chiplets. A baseline value for packaging materials, from Gröger et al., is then added. Where no die area is published, the die area is estimated from the process node and thread count.
For RAM, the capacity of each module is divided by the die density of its DDR generation to give its die area. Die densities come from open life cycle inventory datasets and third-party semiconductor die analyses. The die area is multiplied by a per-cm² memory emissions factor from Gröger et al. A baseline value is added for the circuit board, connector strip, and packaging.
For SSDs, the capacity of each drive is divided by the die density of its NAND generation to give its die area. That die area is multiplied by a die coefficient for the same NAND generation. Newer, taller NAND stacks hold more data per cm² and carry higher embodied emissions per cm². The NAND generation of an instance's drives is inferred from the launch year of its CPU. Each HDD takes a fixed embodied emissions value per device, from open life cycle datasets. Power supply units, chassis, fans, motherboard and expansion cards are assigned fixed values from Dell's critically reviewed life cycle assessments of its PowerEdge servers.
For GPUs, the embodied emissions are the sum of four terms, being the die area multiplied by the die coefficient for the GPU's process node, the capacity of its high bandwidth or graphics double data rate memory multiplied by a per-GB factor, a fixed value for the rest of the board, and a cooling solution term that scales with the card's thermal design power. The per-GB factor, the board value and the cooling solution term are derived from NVIDIA's HGX H100 and HGX B200 product carbon footprints. Each GPU model is recorded for each memory size it is sold in. Each instance family is mapped to the memory size it carries.
Allowances for transport to the data centre and for end-of-life processing are added to the component manufacturing totals as a percentage. Each type of hardware, being servers, disks, GPUs and networking equipment, takes its own pair of percentages from published life cycle assessments of that type. End of life follows the recycled content method of the GHG Protocol Product Standard, so no recycling credits are taken. The component emissions and these two allowances are allocated to the client by the ratio of reserved capacity to total server capacity. The result is amortised over the server's assumed operational lifespan and pro-rated by usage hours. An allowance for the data centre shell is then added to the embodied emissions of each unit of cloud usage.
Where a server is built to a cloud provider's own design and not sold commercially, its non-chip components are estimated from published assessments of comparable standard servers. The transport allowances for GPUs and networking equipment assume air freight. Hardware moved by sea has lower transport emissions, so those two allowances are an upper bound.
The embodied emissions calculation is set out in the companion Greenpixie Methodology for Cloud Embodied Emissions Measurement.
7.4 Water Consumption Calculation
Total water consumption comprises two components. The first is IT cooling water consumption, the water consumed on site in cooling the IT equipment, estimated as the cloud provider's published WUE multiplied by IT Electricity Usage. The second is electricity generation water consumption, the water consumed during the generation of that electricity, estimated using regional water intensity factors that account for the mix of energy sources powering the local grid, including water withdrawn and consumed for fuel extraction, processing, and power generation.
Where cloud providers publish site-specific or region-specific WUE data, these values are used directly. Where only global averages are available, the global average is applied. Where regional electricity generation water data is not available from the producing country, a global average from the World Resources Institute is used.
The water consumption methodology is developed in collaboration with Dr Shaolei Ren at the University of California, Riverside, and informed by the published work Making AI Less Thirsty by Li et al.