Interestana
Home/News/Underwater Data Centers: A Sustainable Frontier for AI's Growing Demands?
Fast Company5 min read

By Interestana AI Editorial — AI-drafted, human-overseen. How we report

Underwater Data Centers: A Sustainable Frontier for AI's Growing Demands?

Underwater Data Centers: A Sustainable Frontier for AI's Growing Demands?

The burgeoning field of artificial intelligence (AI) is creating an insatiable appetite for data processing power, driving an unprecedented demand for data centers. This rapid expansion, however, is accompanied by significant environmental concerns, including escalating energy consumption, substantial water usage for cooling, and a considerable carbon footprint associated with their operation. These mounting challenges are compelling organizations to explore innovative solutions beyond the confines of conventional land-based data centers. Consequently, some forward-thinking developers are now investigating the vast potential of the ocean as a new frontier for housing critical AI infrastructure.

The core concept behind underwater data centers is to leverage the ocean's natural cooling capabilities. By submerging servers in the deep sea, proponents aim to enhance energy efficiency and cooling effectiveness, potentially reducing the reliance on energy-intensive cooling systems common in terrestrial facilities. This approach also offers the benefit of minimizing the demand for precious fresh water resources and reducing the land area footprint required for such massive computing operations. My own research delves into the intricate societal, organizational, and environmental implications of emerging technologies, with a specific emphasis on artificial intelligence and the foundational digital infrastructure, including data centers, that underpins its continuous development and widespread deployment. From this perspective, underwater data centers emerge as a promising new avenue to support the relentless growth of AI.

However, the transition of servers to an oceanic environment does not magically resolve the fundamental environmental issues inherent in data center operations, such as the overall energy consumption and the associated carbon emissions. Moreover, this innovative approach introduces a new set of complex concerns. These include the potential for ecological harm to delicate marine environments, the significant challenge of establishing effective regulatory frameworks to govern these novel offshore installations, and the practicalities of maintaining, repairing, and expanding these submerged facilities should the need arise. Ultimately, the long-term viability of underwater data centers as truly sustainable alternatives to traditional data centers hinges on the successful resolution of these multifaceted economic, technical, and environmental hurdles.

To gauge the feasibility and potential benefits, Microsoft initiated a significant research project in 2015. This endeavor was designed to thoroughly assess the practicalities, advantages, and inherent obstacles associated with deploying data centers beneath the waves. A key milestone in this initiative was the deployment of a waterproof data center on the seabed near Scotland’s remote Orkney Islands in 2018. This experimental facility was equipped with 864 servers and maintained a vital connection to the mainland via an underwater fiber optic cable. After an operational period of two years, Microsoft released findings indicating that the servers housed within this submerged environment experienced failure rates approximately one-eighth that of servers in comparable land-based data centers. While Microsoft's engineers are still actively investigating the precise reasons behind this marked improvement in reliability, preliminary hypotheses suggest that the stable, sealed underwater environment may play a crucial role in reducing the incidence of server failures, potentially by mitigating factors like dust ingress and extreme temperature fluctuations common in land-based facilities.

Original source — read the full reporting at the publisher:

Read on Fast Company

Get the weekly AI digest

AI news + new model releases, weekly. Drafted by our agents, reviewed by humans.

Read next