Paper is often saved with digital applications, see our blog article on the subject of ‘Protect the environment with paperless accounting‘: In the ideal case, letters are no longer sent, or at least internal company correspondence is eliminated. But what about the ecological effects of the server operation of cloud solutions? Is artificial intelligence really as green as it first appears?
Energy saving via cloud
Emissions from Microsoft servers are measured in Metric Tonnes of Carbon Dioxide Equivalent (MTCO2e) to allow comparison of the global warming impact of different gases. Because the total emissions are made up of different gases.
Digitization as a signpost
Microsoft divides the emissions generated when operating the servers into 3 areas: direct emissions from data center operation, indirect emissions from purchased energy or indirect emissions within the value chain (everything that relates to hardware, e.g. the construction of the data center).
Smart Future
Over a period of 10 days in August, only 0.19 MTCO2 emissions were caused for our cloud service. Almost all of this is attributed to the hardware, i.e. the operation is so low in comparison that it cannot be represented in 10 days. These emissions are equivalent to running one LED lamp for 1727 days, or 679 dishwasher cycles, or 4308 liters of boiled water.
- (0.441 conversion factor kWh in kgCO2)
- LED: 10 watts * 24 h = 0.240 kWh * 0.441 = 0.10584 kgCO2; 190 kgCO2/0.11kgCO2 = 1727 days
- Dishwasher: 1.25kW * 0.5 (since approx. one wash cycle takes 2 hours) = 0.625kWh * 0.441 = 0.275625 kgCO2; 190 kCO2/0.28kgCO2 = 679 wash cycles
- Kettle: 0.1kWh * 0.441 = 0.0441kgCO2; 190 kgCO2/0.0441 kCO2 = 4308 liters
AI in the Cloud is more Environmentally Friendly than On-Premise
These examples clearly show that the operation of our cloud solution on rented servers is extremely energy-efficient, see also this Microsoft article.
The energy consumption resulting from the operation can hardly be represented in comparison to the emissions caused by the hardware.
And even these emissions are shared among many different companies and users through hardware sharing.
This means that the energy required can be used in an even more resource-efficient manner than with on-premise solutions.
