The final factor to account for when calculating embodied emissions is the lifespan of the hardware, and the customer’s use as a percentage of that lifespan. Because cloud services are abstracted from the underlying hardware, and server resources can be used by different organizations over time, an organization that does not use a service or virtual machine for the full lifespan of the underlying hardware not responsible for the entire quantity of embodied emissions.
Server hardware is in operation for six years, on average. The three major Cloud Service Providers (CSPs) – AWS, Azure, and GCP – all commit to this six-year lifespan (WatersTechnology, 2024). Tailpipe therefore usually accounts for an organization’s usage as a proportion of that six-year lifespan. However, in the case of GPU and machine learning enabled virtual machines, Tailpipe uses a shorter five-year lifespan. This is based on a 2025 securities filing from AWS which stated that:
We completed our most recent servers and networking equipment useful life study in Q4 2024, and are changing the useful lives of a subset of our servers and networking equipment, effective January 1, 2025, from six years to five years… These two changes above are due to an increased pace of technology development, particularly in the area of artificial intelligence and machine learning.
Tailpipe finds the organization’s usage period in the billing reports from the organization’s cloud service provider. It divides the organization’s period of use by the total five or six year lifespan, and multiplies the total embodied emissions figure by the resulting fraction.
For example, if an organization had paid for the use of a regular (non-GPU) service for one year, and that service is responsible for 50 kg of embodied CO2e, the organization would be responsible for a sixth of those emissions:
50 * (1/6) = 8.33 kgCO2e
