The Evolving Landscape of Data Centers: potential consequences for the energy sector 2025

Data centers have become indispensable to the modern digital economy, powering the cloud services, AI applications, and online experiences that underpin our interconnected world.

Hyperscale data centers, operated by major technology companies such as Google , Amazon , Microsoft or Oracle , exemplify this trend. These facilities, often occupying expansive campuses exceeding 1 million square feet (approximately 93,000 square meters), are strategically located in regions with abundant and relatively inexpensive electricity sources, such as the Pacific Northwest in the United States, China and some European countries such as Iceland.

The China Telecom-Inner Mongolia Information Park currently holds the distinction of being the world's largest data center, spanning approximative 1 million sqm. This may change within a couple of months. To put this into perspective, consider that this single facility occupies an area larger than many shopping malls or even some airports.

In fact, it is mandatory to go behind our screens to better measure the impact of the transformation we are currently going through: The amount of data stored within data centers has been experiencing explosive growth.

In 2010, the global datasphere was estimated to be around 2 zettabytes (ZB). By 2025, this figure is projected to reach an astonishing 180 ZB, representing a 150-fold increase in just 15 years. This exponential growth is anticipated to continue unabated. By 2035, the global datasphere is projected to reach a staggering 2,100 ZB.

For the people of my generation, 2000 ZB represent approximatively the volume of 450 quadrillion DVDs and 1/9th the volume of the entire Earth! If you prefer books, it represents 10 times the volume of the entire Earth. These figures should be multiplied by 50 by 2040 and by 500 by 2050. Welcome to the Matrix Mr Anderson!

The emergence of cloud computing and Artificial Intelligence (AI) have significantly amplified energy consumption within data centers. Training sophisticated AI models demands an order of magnitude more computational power compared to traditional web searches. For instance, training a large language model can consume as much energy as a typical household over its entire lifetime. This surge in computational intensity stems from the complex algorithms and massive datasets employed in AI training, requiring extensive processing power and memory resources. As AI models become increasingly complex and sophisticated, their energy demands are expected to escalate further, posing a significant challenge to the sustainability of data center operations.

In addition, due to the datasphere size and our constant needs to accelerate treatments, quantum computers are poised to play a pivotal role in the evolution by offering unique capabilities. Their inherent parallelism allows them to process and analyze massive datasets with unprecedented speed and efficiency, enabling faster insights and more accurate predictions.

In a nutshell, we face a skyrocketed energy demand which places a considerable strain on power generation, electricity grids, potentially leading to grid congestion, and future energy costs, especially when green.

Data centers consumption is currently estimated to be around 2% of total global electricity consumption. To put this in perspective, this is roughly equivalent to the electricity demand of a country like Brazil. Projections suggest that data center energy consumption should reach 5% of global electricity consumption by 2030 and exceed 10% by 2040, approximatively 50% of the US demand.

The development of the datasphere has created a novel and unprecedented challenge for the energy industry. The sheer scale and rapid growth of data consumption, fueled by AI, cloud computing, and the proliferation of connected devices, demand a significant and rapidly increasing supply of energy. This presents a unique challenge for the energy sector, where development cycles for new power plants and infrastructure projects are typically measured in years, if not decades.

In contrast, the pace of innovation within the digital sector, particularly among major technology companies, is breathtaking. These companies, driven by the demands of a hyper-competitive market and fueled by substantial market capitalization, can rapidly adapt and scale their operations. Their ability to innovate and deploy new technologies at an unprecedented pace allows them to exert significant influence over the energy supply chain. This could lead to shifts in the energy landscape, potentially disrupting traditional energy providers and fostering the emergence of new players, much like Tesla disrupted the automotive industry which was over-saturated when arriving. The competitive dynamics within the energy sector may undergo a significant transformation, with changes in ownership structures and the emergence of new market players.

Given the current dynamics of the US stock market, the upcoming new US administration and the anticipated acceleration of technological advancements, 2025 promises to be a particularly interesting year. We can expect to see increased competition and innovation within the energy sector as companies strive to meet the growing energy demands of the digital age.