Google’s environmental goals are facing major hurdles due to its growing use of energy-hungry data centers.
On Tuesday, Google reported a 48% increase in its greenhouse gas emissions over the last five years, mainly due to the high electricity use of its data centers and supply chain.
In its latest environmental report, Google reported a 13% increase in emissions from 2022 to 2023, reaching a total of 14.3 million metric tons.
Data centers: The energy guzzlers
At the core of Google’s AI advancements, like the Gemini and OpenAI’s GPT-4 models, are data centers that consume huge amounts of electricity. According to the International Energy Agency, data centers’ total electricity consumption is projected to double by 2026, reaching 1,000 TWh—equivalent to Japan’s current electricity demand.
Furthermore, research firm SemiAnalysis forecasts that AI could account for 4.5% of global energy generation by 2030.
Google’s goal to achieve net-zero emissions by 2030 has been made more challenging by these energy demands. The company recognizes the “significant uncertainty” regarding the environmental impact of AI, which adds to the difficulty of reaching the net-zero goal.
“It won’t be easy,” Google admitted, highlighting the challenges posed by the unpredictable future of AI’s energy consumption.
Microsoft, another tech giant deeply invested in AI, has also felt the strain. The company’s president, Brad Smith, acknowledged in May that their AI strategy has made their carbon-negative target by 2030 increasingly difficult, stating, “the moon has moved.”
Microsoft’s total carbon emissions have risen nearly 30% since 2020, primarily due to the construction of new data centers. This increase shows how much energy AI systems need.
AI’s environmental toll: More than just electricity
The environmental impact of AI extends beyond electricity consumption. Manufacturing and transporting the computer servers and chips necessary for AI also contribute to carbon emissions. Additionally, water usage is a growing concern.
A study estimates that AI could account for up to 6.6 billion cubic meters of water use by 2027—almost two-thirds of England’s annual consumption .
Cooling systems for data centers are a significant part of this water consumption. As data processors become more efficient, they also generate more heat, necessitating increased cooling resources.
In 2022 alone, Google and Microsoft consumed a staggering 32 billion liters of water in their data centers.
Big tech companies have become major purchasers of renewable energy in an effort to meet their climate goals. Bill Gates recently highlighted that AI could aid in combating the climate crisis, as tech companies are “seriously willing” to pay extra for clean electricity to claim they’re using green energy.
However, since wind and solar energy aren’t always available, data centers need a steady power supply, which can strain the existing power grid.
Balancing AI growth and sustainability
As AI continues to expand, individual companies’ electricity usage is expected to rise. Traditional data center servers typically run on 7 kilowatts of electricity, whereas AI servers can use between 30 and 100 kilowatts due to their increased processing power, according to reports.
Nvidia , a leading AI server provider, shipped 100,000 units last year, with an annual energy consumption 7.3 times higher than traditional servers.
Even though computing is becoming more efficient, data centers still face challenges. More efficient processors generate more heat, which puts extra demand on cooling systems and water resources.
To tackle these issues, the U.S. needs to manage its resources wisely. Big companies like Microsoft are investing in wind and solar farms to balance their electricity use. While these efforts are good, there’s still a big problem with needing constant power when renewable sources aren’t producing energy.
Most Popular
Comments / 0