1. Home
2. Guides
3. Account info
4. Sustainability

This guide explains how we calculate electricity-related Greenhouse Gas (GHG) emissions from facilities that house Fastly PoPs. It outlines the steps in which one or more outputs and metrics are created for consumption in later steps.

The outputs and metrics referred to in this guide are used as the basis for the metrics exposed to customers in the Fastly control panel, downloadable CSVs, and the API. However, they aren't exposed at this level of detail. Our guide to the Sustainability dashboard describes the metrics that are available to Fastly customers in more detail.

Also, keep in mind that some source data required to calculate emissions (i.e., Power Usage Effectiveness (PUE), renewable electricity coverage percentage, and emission factors) is only ever available for the previous year. As such, calculations using these variables use the most recently available year's figures until data from a more recent year is provided by the facility operator.

Step 1: Calculating Fastly IT equipment electricity consumption

Fastly-operated IT equipment (e.g., cache servers, switches) resides in leased, colocation facilities. We measure the our equipment's electricity consumption at these facilities in kWh based on monitoring by power distribution units (PDUs), which are also located at these sites.

To calculate Fastly IT equipment electricity consumption, we obtain each site's kWh data from all PDUs and also from all cache servers on a daily basis at the facility level and then compare the numbers. If the number of kWh recorded by PDUs is lower than the kWh recorded by servers on a particular day, this implies that a PDU is not reporting data correctly.

For each day this discrepancy occurs, we add 25% to the electricity consumption associated with running cache servers to account for the additional electricity required for equipment such as routers, switches, and other networking infrastructure. By incorporating this increase, Fastly ensures a more accurate estimation of total electricity consumption associated with its edge cloud platform. We've based this 25% increase on empirical observations of Fastly's deployments, making it a reliable representation of infrastructure-related overhead across diverse facility environments.

Outputs and metrics

• Daily electricity consumption in kWh by Fastly IT equipment in each facility (kWh_IT Equipment_Facility_Day)

Step 2: Calculating Fastly share of facility electricity consumption by non-IT equipment

Colocation facilities consume electricity beyond what IT equipment requires, including for environmental controls and lighting (facility-operated non-IT equipment). We annually request each facility's average PUE for the previous year and use that number to determine Fastly's share of the total electricity consumption there. If PUE is not provided for a specific facility, the average PUE across all facilities in that year is used in our calculations.

We calculate Fastly's PUE by dividing the each facility's average, annual electricity consumption by our total annual IT equipment electricity consumption in that facility. Daily electricity consumption by Fastly-operated IT equipment in each facility (from Step 1) is multiplied by the facility's average annual PUE to calculate Fastly's share of the combined Fastly-operated IT equipment and facility-operated non-IT equipment electricity consumption in the facility.

To be able to calculate Scope 3 GHG emissions from non-IT electricity consumption, we need a distinct metric for non-IT equipment (overhead) electricity consumption. We obtain this by subtracting the daily electricity consumption by Fastly IT equipment in each facility from Fastly's share of the combined IT equipment and non-IT equipment electricity consumption in the facility.

Outputs and metrics

• Daily electricity consumption in kWh by non-IT equipment in each facility (kWh_Non-IT Equipment_Facility_Day)

Step 3: Calculating facility renewable electricity mix

Co-located facility operators may procure electricity from renewable sources through a variety of different mechanisms. Depending on the physical location of the individual facility, they may:

• establish contracts such as Power Purchase Agreements (PPAs) directly with low-carbon generators,
• negotiate with their supplier or utility to supply low-carbon electricity,
• purchase green energy attribution certificates (EACs) such as RECs, i-RECs, or GOs, or
• generate renewable electricity on-site for their own consumption.

Regardless of how the renewable power is obtained, we request the percentage procured annually for the previous year and use that number to determine the portion of electricity consumed by Fastly that is covered by renewables and grid electricity at each facility for that year.

We require Facility operators to substantiate the renewable energy procurement they report to us with acceptable evidence such as attestations, contracts for PPAs, utility green tariffs, or EAC certificates.

Outputs and metrics

• Daily renewable electricity consumption in kWh by Fastly IT equipment in each facility (kWh_IT Equipment_Renewable_Facility_Day)
• Daily grid electricity consumption in kWh by Fastly IT equipment in each facility (kWh_IT Equipment_Grid_Facility_Day)
• Daily renewable electricity consumption in kWh by non-IT equipment in each facility (kWh_Non IT Equipment_Renewable_Facility_Day)
• Daily grid electricity consumption in kWh by non-IT equipment in each facility (kWh_Non IT Equipment_Grid_Facility_Day)

Step 4: Calculating Fastly Scope 2 emissions from Fastly IT Equipment

Fastly uses Scenario 1 from BSR's Future of Internet Power Report to calculate Scope 2 emissions from Fastly IT equipment based on a combination of location-based and market-based emissions data for each relevant country. This means we account for electricity consumed by Fastly-operated IT equipment in colocated data centers as Scope 2, and electricity consumption by equipment that is owned by colocated data center operators as Scope 3 category 8.

Fastly's Scope 2 GHG emissions associated with electricity consumption in facilities are calculated according to the GHG Protocol Scope 2 Guidance using emission factors from CaDI, the carbon database initiative, published by Carbonfootprint.com.

Location-based emissions omit the effect of renewable energy procurement at facilities by facility operators. We calculate them by adding together daily electricity consumption in kWh by Fastly IT equipment in each facility (from Step 1) for all facilities in the same country to calculate the total daily kWh consumed by Fastly IT equipment in each relevant country. In the USA, Canada, and Australia, electricity consumption is totaled at the state or regional basis rather than at the country level.

The location-based emissions total for that area (the country or the state or region in USA, Canada, and Australia) is then multiplied by the electricity generation emission factor for the relevant area and year to calculate the kg of carbon dioxide equivalent (kgCO2e) resulting from electricity generation caused by consumption of electricity by Fastly IT equipment across facilities in that country, state, or region on a particular day.

Market-based emissions include the effect of renewable energy procurement at facilities by facility operators, which are zero emissions under the Greenhouse Gas Protocol. We calculate them based only on grid electricity consumption by first adding together daily grid electricity consumption in kWh by Fastly IT equipment in each facility (from Step 3) for all facilities in the same location (the country or the state or region in USA, Canada, and Australia) and use that to calculate the total daily grid kWh consumed by Fastly IT equipment in each relevant area.

The market-based emissions total for that area (the country or the state or region in USA, Canada, and Australia) is then multiplied by the residual emission factor for the relevant area and year to calculate the kg of carbon dioxide equivalent (kgCO2e) resulting from electricity generation caused by consumption of grid electricity by Fastly IT equipment across facilities in that country, state, or region on that day.

Finally, we add together all country, state, or region-specific kgCO2e metrics to obtain a global metric, and convert into metric tons CO2e (MTCO2e) by dividing all kgCO2e metrics by 1000.

Outputs and metrics

• Scope 2 Location-Based (Country): Daily kgCO2e from total electricity consumption by Fastly IT equipment in each country (KgCO2e_IT Equipment_ElecGen_LocationBased_Country_Day)
• Scope 2 Market-Based (Country): Daily kgCO2e from total grid electricity consumption by Fastly IT equipment in each country (KgCO2e_IT Equipment_Residual_MarketBased_Country_Day)
• Scope 2 Location-Based (Global): Daily MTCO2e from total electricity consumption by Fastly IT equipment across all countries (mtCO2e_Scope 2_LocationBased_Global_Day)
• Scope 2 Market-Based (Global): Daily MTCO2e from total grid electricity consumption by Fastly IT equipment across all countries (mtCO2e_Scope 2_MarketBased_Global_Day)

Step 5: Calculating Fastly Scope 3.8 emissions from non-IT equipment in upstream leased assets

Location-based and market-based emissions from electricity consumption by non-IT equipment follow an identical approach to that described in step 4 above, except the starting point is daily electricity consumption by non-IT equipment in each facility (from step 2).

Outputs and metrics

• Scope 3.8 Location-Based (Country): Daily kgCO2e from total electricity consumption by non-IT equipment in each country (KgCO2e_Non IT Equipment_ElecGen_LocationBased_Country_Day)
• Scope 3.8 Market-Based (Country): Daily kgCO2e from total grid electricity consumption by non-IT equipment in each country (KgCO2e_ Non IT Equipment_Residual_MarketBased_Country_Day)
• Scope 3.8 Location-Based (Global): Daily MTCO2e from total electricity consumption by non-IT equipment across all countries (mtCO2e_Scope 3.8_LocationBased_Global_Day)
• Scope 3.8 Market-Based (Global): Daily MTCO2e from total grid electricity consumption by non-IT equipment across all countries (mtCO2e_Scope 3.8_MarketBased_Global_Day)

Step 6: Calculating Scope 3.3 emissions (fuel and energy-related emissions: T&D losses, well to tank, WTT of T&D losses)

• Emissions from Transmission and Distribution (T&D) Losses: kWh_IT Equipment_Country_Day (from step 4) is multiplied by the electricity transmission and distribution losses emission factor for that area (the country or the state or region in USA, Canada, and Australia) to calculate the kg of carbon dioxide equivalent (kgCO2e) resulting from losses of electricity consumption by Fastly IT equipment during transmission and distribution across all facilities in that area on that day.
• Well to Tank (WTT) Emissions: kWh_IT Equipment_Country_Day (from step 4) is multiplied by the well to tank emission factor for that area (the country or the state or region in USA, Canada, and Australia) to calculate kgCO2e resulting from extraction, refining, and transportation of fuel used to generate the electricity consumed by Fastly IT equipment across all facilities in that area on that day.
• Well to Tank (WTT) of T&D Loss Emissions: kWh_IT Equipment_Country_Day (from step 4) is multiplied by the well to tank of T&D emission factor for the relevant country (state or region in USA, Canada, and Australia) to calculate kgCO2e of well to tank: emissions associated with electricity lost in transmission and distribution of electricity consumed by Fastly IT equipment across all facilities in that country, state, or region on that day.

Finally, we add together all country, state, or region-specific kgCO2e metrics to obtain a single global metric, and convert into metric tons CO2e (MTCO2e) by dividing all kgCO2e metrics by 1000.

Outputs and metrics

• Scope 3.3 T&D (Country): Daily kgCO2e resulting from losses of electricity consumption by Fastly IT equipment during transmission and distribution across all facilities in each country (KgCO2e_T&D_Country_Day)
• Scope 3.3 WTT (Country): Daily kgCO2e resulting from extraction, refining, and transportation of fuel used to generate the electricity consumed by Fastly IT equipment across all facilities in each country (KgCO2e_WTT_Country_Day)
• Scope 3.3 WTT of T&D (Country): Daily kgCO2e resulting from extraction, refining, and transportation of fuel used to generate the electricity consumed by Fastly IT equipment lost during transmission and distribution across all facilities in each country (KgCO2e_WTT T&D _Country_Day)
• Scope 3.3 (Global): Daily MTCO2e associated with T&D, WTT and WTT of T&D across all countries (mtCO2e_Scope 3.3_Global_Day)

Annual threshold-based comparison of facility-level Fastly IT equipment electricity consumption

On an annual basis, we compare the Fastly IT equipment electricity consumption data gathered from PDUs in each facility (via step 1) for the previous year with the data reported by facility operators for the same time period. Where the facility-operator-provided kWh is more than 10% higher than the kWh recorded by PDUs, the previous year's kWh is replaced with the data reported by the facility operator. Where this occurs, in-year daily data will be increased by the same proportion in order to maintain the ability to show representative trends in graphs and data.

How we attribute electricity consumption and GHG emissions to individual customer accounts

Fastly tracks customer usage of its platform features and products using billing statistics metered by Fastly service and PoP (see Historical Stats). At the edge, Fastly platform features and products run a collection of processes on Fastly IT equipment located in facilities around the world. Fastly tracks how much CPU time each process uses on each server and apportions electricity consumption per process according to aggregate daily CPU time. We do not consider variation in fleet CPU utilization or use a time- or hardware-based power consumption model. Every process that is not dedicated to either Delivery or Compute is categorized as shared. Specifically:

• The electricity consumed by Delivery-dedicated processes is apportioned to Fastly Delivery services based on their volume of network transfer (ingress and egress) that we commonly refer to as “bandwidth” in the billing statistics.
• The electricity consumed by Compute-dedicated processes is apportioned to Fastly Compute services using the compute request time billing statistic since they operate on a single virtual WebAssembly CPU.
• The electricity consumed by processes in the Shared category is apportioned to Fastly services as an overhead.

Dashboard emission boundary

While the Sustainability dashboard provides insights into data center electricity consumption, renewable energy coverage, and associated GHG emissions, it does not provide data on other environmental factors such as water, diesel, or other categories of Scope 3, such as category 1 (purchased goods and services, which includes emissions associated with third party cloud hosting), category 5 (waste generated in operations), category 6 (business travel), or category 7 (employee commuting). It is not designed to attribute Fastly's full GHG inventory to individual customer accounts.

This table summarizes what is and is not included in the displayed GHG emissions calculations.