An investigation from the British news outlet The Guarding uncovered that carbon emissions from the data centers owned by companies like Microsoft, Alphabet, and Meta Platforms may drastically exceed what they officially reported between 2020 and 2022.
The situation has been exacerbated by the rise of artificial intelligence (AI) as data from Goldman Sachs indicates that a ChatGPT query demands 10 times more energy than a Google search.
Amazon and Apple lead the chart in terms of carbon emission although it is hard to compare the carbon footprint of the ecommerce giant with that of its rivals as it does not provide a breakdown of how its different businesses contribute to polluting the environment.
Even though the International Energy Agency (IEA) says that data centers account for just 1.5% of the globe’s electricity consumption, Goldman Sachs’s research indicates that the demand for power coming from this industry will nearly triple just 5 years from now.
However, despite these alarming trends, all of the five tech companies mentioned earlier said that their operations had reached carbon neutrality at some point.
Skeptics have pointed out that they have managed to accomplish these goals only on paper and through the use of creative accounting techniques that mask reality.
The Amazon Employees for Climate Justice advocacy group, which is made up of employees who are discontent with how the company keeps track of its carbon footprint, boldly stated: “It’s down to creative accounting. Amazon – despite all the PR and propaganda that you’re seeing about their solar farms, about their electric vans – is expanding its fossil fuel use, whether it’s in data centers or whether it’s in diesel trucks.”
Creative Accounting and the Role of Renewable Energy Certificates
It appears that companies like Amazon rely on Renewable Energy Certificates (RECs), which are documents that attest that the company is buying renewable energy for its operations to the point that matches the electricity consumed from fossil-fuel-generated sources.
However, this energy is not consumed by the company itself and can be generated in any other location, meaning that its actual carbon footprint is not being offset in reality. Some estimates claim that, once RECs are eliminated from the calculation, these five tech companies generate as much combined carbon emissions as entire countries.
In fact, based on the figures provided by these five firms in 2022, they would rank 33th in the list of countries with the highest carbon emissions in the world.
Experts within the data center industry believe that location-based carbon emissions are the right way to assess how much pollution these facilities are generating. “Location-based [accounting] gives an accurate picture of the emissions associated with the energy that’s actually being consumed to run the data center,” commented Jay Dietrich, a research director from the Uptime Institute.
Data Centers Obscure Their Carbon-Emission Stats via Third-Party Providers
There seems to be a stark difference in the results that companies would be reporting if they adopt the location-based method. In the case of Meta Platforms (META), they reported just 273 metric tons of CO2 emitted in 2022. However, if the location-based methodology is implemented, that number would jump to 3.8 million tons coming out of its data centers alone.
Meanwhile, Microsoft reportedly emitted nearly 281,000 metric tons of CO2 during that same year via the Scope 2 methodology. However, if the calculations are run based on the location-based parameters, the amount skyrockets to 3.8 million. The buck doesn’t stop with big tech either.
The largest tech companies in the world accounted for 37% of the global data center capacity by 2022 according to the Synergy Research Group and around half of that installed capacity comes from third-party providers.
This makes it much more difficult to estimate their actual carbon footprint as they can exclude these amounts from their reports and claim that they have no decision power over the energy sources that these other companies rely on and use.
Chris Dietrich explains: “Scope 3 emissions are hugely uncertain. This area is a mess just in terms of accounting.”
AI Pushes Data Centers to Its Boundaries
Moreover, the rise of AI is dramatically increasing energy consumption by data centers as workloads require exponentially more computation. Google and Microsoft have already confirmed that this new technology is responsible for the recent upticks in market-based emissions.
The Chief Executive Officer of DigitalBridge, a private equity firm that owns large data centers, made an alarming claim as he warned that the sector could run out of power two years from now. If it continues to grow at this pace, the easiest and cheapest power sources, namely fossil fuels, will likely be ramped up to keep big tech satiated, no matter the effects on the planet and its inhabitants.
The sector’s expansion has also impacted specific locations, even small cities and towns, where water shortages have occurred as data centers need significant amounts to cool down their systems.
In places like Dublin, Ireland, data centers are the fifth largest consumers of electricity. Authorities have opted to temporarily suspend the construction of new facilities to avoid a crisis.
Also read: How Much Data is Created Every Day in 2024?
Moreover, certain residential areas like those in Northern Virginia are being transformed into industrial zones as large expansions of land are needed and state officials want to attract new businesses to the area.
However, residents have complained that this could affect their health and way of life and could potentially threaten their ability to access public services if the local infrastructure is not prepared to supply the needs of new huge data centers.
Statistics show that a third of the world’s data centers are located in the US while Europe and China only account for 16% and 10% of the total respectively. That said, facilities in China are draining the country’s resources fast as they consume 1.3 billion cubic meters of water per year.
Regulators Are Getting Ready to Curve the Indiscriminate Growth of Data Centers
Google claimed that its DeepMind AI lab found a way to reduce the energy consumption of its data center cooling systems by 40%. This could be a substantial improvement but most of the energy consumption is used to power the hardware itself, not the cooling systems.
Industry observers have also stated that data center companies could be forced in the future to build renewable sources of energy that power their facilities. These would include nuclear or solar-powered solutions that can reduce their reliance on fossil-fuel-generated electricity.
Some countries seem to be getting ready to enforce the adoption of sustainable solutions to this issue. For example, Singapore announced that it was creating a sustainability standard while the European Commission is drafting regulations on data center sustainability as well.
Also read: AI in Engineering in 2024: Creating the Future with Artificial Intelligence
Sasha Luccioni from the AI platform Hugging Face emphasized the importance of transparency in the industry: “Transparency is currently our main hurdle, because many people don’t realize the environmental impacts of AI: and not only consumers, but lawmakers and companies. When we establish transparency and start having a little bit more information, then we can start trying to regulate.”
As the tech industry continues to innovate and expand, particularly in the AI space, it faces significant challenges in finding the right balance between moving forward with its ambitions and protecting the environment.
The discrepancies between reported and actual emissions, the massive energy demands of data centers in the wake of the AI revolution, and the local and global impacts of these facilities all point to a pressing need for more accurate reporting, improved efficiency, and potentially, stricter regulation.