Data: The New Waste Crisis – The Hidden Impact of Storage Growth (2/4)

In my previous blog, I highlighted how essential sufficient data storage is to deploy AI in a reliable and scalable manner. In this follow-up, we take a closer look at the darker side of our insatiable storage appetite: its environmental impact.

Recent estimates suggest that by 2025, the global data volume will exceed 200 zettabytes. This explosive growth is driven by:

• Generative AI and deep learning applications
• IoT and edge devices continuously generating data
• High-resolution streaming services
• Smart industries and sensor-based technology
• Scientific simulations and research data

These trends are not only pushing our digital infrastructure to its limits, but also forcing organizations to rethink their data strategy, architecture, and governance. And that strategic rethink is urgently needed — because the environmental consequences of this explosive growth in strorage are far from hypothetical.

1. Electricity Consumption by Data Centers

Data centers already consume 1 to 3% of the world’s total electricity (IEA, 2023), and that share is growing each year. The increasing demand for storage — particularly from AI, video, and IoT — is driving energy use to unprecedented levels.
Source: International Energy Agency (IEA) – Electricity 2023 Report

• 2. CO₂ Emissions from Fossil Power

Many data centers still run on fossil fuels (coal and gas), especially in regions with cheap, carbon-intensive energy. According to Nature, data centers generate 200 to 300 megatons of CO₂ per year — comparable to the aviation sector in its entirety.
Source: Nature, 2020 – How to stop data centres from gobbling up the world’s electricity

•  3. Water Consumption for Cooling

To prevent overheating, hyperscale data centers consume millions of liters of water every day for the purpose of cooling. This is a growing concern in water-stressed areas like for example Arizona or parts of Spain.
Source: Uptime Institute (2022) – Data Center Water Usage

• 4. Electricity Demand for AI Training

AI models like GPT-4 require immense amounts of storage and processing power. Training just one model can consume several gigawatt-hours of electricity — equivalent to the annual energy usage of a small town.
Source: MIT Technology Review (2022) – Training a single AI model can emit as much carbon as five cars in their lifetimes

These figures may sound abstract, but they’re closely tied to our daily lives:

• Watching Netflix in 4K burns about 7 GB per hour. Watching 2 hours per day adds up to ~420 GB per month per user. Across 100,000 users, that’s 42 petabytes per month.
  Source: Netflix Help Center + Techradar
• Spotify hosts over 100 million tracks. At ~5 MB per song, that translates into ~500 terabytes of music storage.
  Spotify newsroom
• Facebook processes over 4 petabytes of data every day, including photos, videos, and real-time interactions.
  Source: Meta Engineering Blog

Data is the life blood of our digital economy — but it also needs oxygen: electricity, water, space, and energy. If we want AI to scale sustainably, we must rethink the foundations of our storage infrastructure. Smarter, greener, and more purpseful approaches are the way forward.

The era of “store everything forever” is over. The future isn’t just data-driven — it’s impact-aware.

That will be a conversation for my next blog.