In the grand scheme of things, data centers account for a fraction of the world’s electrical usage. However, they are growing more resource-hungry as demands on cyberspace increase.
Tech pundits project that the electrical consumption of these industrial facilities will increase significantly in the near future. This upward trajectory stems from the aggressive investment by hyperscalers to boost data center capacity, but it’s just one of the many growth drivers. Understanding where else future demand will come from, the prevailing supply challenges, and unresolved inefficiencies is the key to averting an energy crisis.
Cryptocurrency’s resilience
The crypto industry is notoriously energy-intensive. Most cryptocurrencies use proof-of-work consensus mechanisms, which require considerable electricity by design. Miners must expend so much energy to solve complex mathematical puzzles, computationally securing networks from fraudsters who want to tamper with transactions but are unwilling to invest the same amount of resources.
Energy expenditure is intrinsically tied to the security of respected crypto networks — take Bitcoin as an example. It adjusts the difficulty of puzzles miners must solve relative to the available computing power to ensure only one block is added to the chain roughly every 10 minutes. This built-in feature incentivizes network participants to invest in power-hungry ASICs that run continuously to outguess an increasing number of competitors.
Crypto detractors used to weaponize this fact against Bitcoin, highlighting its design’s perceived detrimental impact on the environment to influence bearish sentiments. Despite headwinds, the cryptocurrency market remains resilient, and bitcoin’s legitimacy is stronger than ever.
Bitcoin has become a reserve asset by the United States, sending a message to the world that it’s here to stay. This development would drive up and stabilize the price of BTC over the long term, enticing more crypto mining companies, energy and infrastructure enterprises, engineering and design firms, and hardware manufacturers to build more bitcoin farms.
AI’s cooling crisis
The AI revolution is underway, forcing hyperscalers to rethink the architecture of IT infrastructure. Traditional enterprise data centers, designed with 5-15 kW rack densities and air cooling, are severely underequipped to handle the rising volumes of AI workloads powered by GPUs. Individual racks for AI can start at 30 kW and exceed 80-100 kW.
Liquid cooling is a viable method to efficiently prevent high-performance hardware from overheating. Direct-to-chip liquid cooling systems remove significantly more heat than air, as they utilize cold plates to pump coolant directly to processors. This method promises greater energy efficiency gains and sustained performance, allowing data centers to deploy more GPU clusters for AI workloads while moderating electric consumption.
Immersion cooling is an alternative approach. This method involves submerging entire servers in a nonconductive dielectric fluid for total heat dissipation, rather than circulating coolant through pipes to lower the temperature levels of specific GPU spots.
Hyperscalers’ sustainability pledges
In the past, ramping up the consumption of fossil fuels could effectively address surging power demand. Although the U.S. is confident that its crude oil supplies will be sufficient to meet energy demand through 2050, tech giants operating hyperscale data centers are promising a greener approach.
Sustainability is a top priority for Big Tech to reinforce eco-friendly brand images, generate capital from green-minded investors, and prepare for a future when fossil fuels are no longer available. The problem is that generating clean electricity is challenging due to the intermittent nature of primary renewable energy sources, such as solar and wind. Photovoltaic panels and wind turbines heavily rely on minerals, making them vulnerable to logistical and geopolitical supply chain disruptions.
The good news is that Amazon, Microsoft, and Google have the means to fund the construction of large-scale solar farms. Still, will their partners generate enough renewable power to meet the sharp power demand from data centers before it results in an energy crisis? The International Energy Agency predicts that electricity generation from renewables will reach 16,200 TWh by 2030 — a reassuring projection.
Moreover, new players present a sustainable business case for AI factories. Many crypto miners that never operated traditional data centers have migrated to the space, especially after ChatGPT’s rise to fame, leveraging their land zoned for industrial use and immediate access to low-cost renewable energy sources. Former crypto mining facilities operate as neoclouds that rent out GPUs or serve as wholesale colocation providers for AI model developers, helping to meet some computing demand sustainably.
European Union’s (EU) blind spots
Failing to address operational inefficiencies diminishes the merits of harnessing clean electricity. The EU is guilty of this — a cause for concern, as the union accounts for a significant chunk of hyperscale data center capacity worldwide.
EU policymakers know little about how individual data centers within the jurisdictions of the union’s member states operate because of the lack of reliable, specific information about the natural gas imports from Russia. Insufficient business-to-government data sharing and commercial confidentiality rules lead to estimates rather than precise reported data.
The EU acknowledges its lack of precise consumption data and aims to address this through legislative frameworks, which should clarify the legal requirements and incentives for businesses to report their energy consumption data.
Mitigating an energy crisis in the making
Underestimating the skyrocketing power demand of data centers is a recipe for disaster. Fortunately, vital stakeholders have taken measures to prevent an impending crisis from occurring and mitigate its impact on other industries.
Jack Shaw is
TNGlobal INSIDER publishes contributions relevant to entrepreneurship and innovation. You may submit your own original or published contributions subject to editorial discretion.
Featured image: imgix on Unsplash
How precision sorting is changing the way the supply chain moves goods