In Honor of Nobel Laureate Dr. Avram Hershko
With artificial intelligence or AI and cryptocurrency mining expanding globally at a phenomenal rate, data centers are expected to double their electricity consumption from 460 terawatt-hours (TWh) in 2022 to more than 1000 TWh in 2026 [1]. Both have significant cooling needs, with 30 to 55% of the electricity powering the cooling systems [2]. Data centers directly use water in the cooling systems to extract heat, and the humidification systems—maintain 40-60% relative humidity to prevent static electricity buildup, bathrooms, and fire sprinkler systems; and indirectly use water from using thermoelectric energy generation.
Each year, the world uses more than 4.3 trillion cubic meters (~1.1 quadrillion gallons) of water; and data centers are among the top ten consumers. In 2022, Google’s data centers alone consumed 4.3 billion gallons of water [3]. There are 7,069 hyperscale data centers in 140 countries [4]. In addition, there are hundred-thousands of data centers globally, and their water consumption adds up to a significant amount. We are in the midst of a global drought and it is not sustainable to continue supporting the data centers water needs. We urgently need to find low-water alternatives to data center cooling methods.
In 2023, the global energy mix comprised 60% fossil fuel contribution from 35% coal (10,434 TWh), 23% natural gas (634 TWh), and 2.7% oil & petroleum products (786 TWh), clean energy from 14% hydro (4,210 TWh), 9.1% nuclear (2,686 TWh), 7.8% wind (2,304 TWh), 5.5% solar (1,631 TWh), 2.4% bioenergy (697 TWh), and 0.3% or 90 TWh other renewables—mostly geothermal generation, with tidal and wave energy providing a small fraction [5].
Per the Food & Water Watch Institute, the water withdrawal intensities (and lifecycle water consumption) to produce 1 MWh of electricity using natural gas, coal, and nuclear, wind, and solar are 45.110 m3 (0.600 m3), 84.413 m3 (1.671 m3), 93.600 m3 (2.000 m3), 0.001 m3 (0.001 m3), and 0.040 m3 (0.020 m3) respectively [6]. According to the United Nations Environment Program, some 1.8 billion people will likely face absolute water scarcity by 2025 [7], and for them, power generation is the fourth priority after water for food production, safe drinking water, and sanitation.
Disregarding the global water crisis, the fossil fuel industry, and nuclear energy continue to expand their energy footprint. Recently, the electric regulatory board in the U.S. State of Georgia approved the construction of 2.4 GW of coal and gas power plants for various utilities [8]. However, there is hope from multiple fronts.
With the rapid expansion of “attribution science” [9] correlating extreme weather to climate change induced by human activity—power generation from nuclear and fossil fuels is facing backlash and litigation for economic and human losses [10].
The United States Congress has recently introduced the “Artificial Intelligence Environmental Impacts Act of 2024” mandating voluntary reporting of energy and water consumption, pollution, and electronic waste associated with the full lifecycle of artificial models and hardware, etc. [11]. This act is likely to pass and many nations will follow suit.
We urgently need an expeditiously deployable—viable clean energy—low water utilization alternative to thermoelectric power generation technology and financiers. Here’s an idea.
Install solar microgrids on the municipal water and wastewater treatment facilities to produce the cheapest clean electricity and green hydrogen; and install geothermal district cooling networks to provide heating, ventilation, and air conditioning or HVAC services. Interconnect multiple solar microgrids and district thermal energy networks to meet the data centers' energy and cooling requirements. The data centers would benefit from highly reliable cheapest clean electricity, a noise-free environment, extended operational life of electronic equipment due to the absence of combustion products from geothermal cooling, and significantly reduced operating expenditures and downtime.
Next is routing the high-quality waste heat from the data centers to chemical manufacturing plants to drive their energy efficiency and reduce energy usage. All chemical reactions occur within a temperature and pressure window, whose maintenance is energy-intensive and a source of GHG emissions. The waste heat from chemical manufacturing plants can be transported back to the municipal water & wastewater treatment facilities to aid in temperature control of wastewater during microbial decomposition of sludge, saving significant energy expenditure for this step. This route will drive energy efficiency, lower operating costs, and minimize GHG emissions at the data centers, chemical manufacturing plants, and municipal water and wastewater treatment facilities. The solar microgrids can also meet the electricity needs of the three entities. We can approach the nuclear and fossil fuel industries to finance these projects as they look to pivot.
