Table of Contents
Sustainable Finance - This article is part of a series.
Sustainable finance has become a pivotal factor in the tech industry. As we witness an exponential rise in the use of technology, the associated resource consumption and power usage have escalated, leading to a significant environmental impact. This post will explore the history of power usage and resource utilization by the tech industry, touching on the impact of cryptocurrency mining and AI/ML LLMs on global power consumption. Then it will highlight emerging technologies aiming to reduce power utilization and dependence on non-renewables.
The Power Hungry Tech Industry: A Historical Perspective #
The timeline of power usage in the tech industry is closely tied to the evolution of electrical power generation and distribution systems. The journey begins in the mid-18th century with the inception of glass friction generators, marking the dawn of the electrification era. By the early 19th century, rudimentary forms of electric motors and generators had been invented, laying the foundation for modern power systems.
The invention of the dynamo in 1832 by Frenchman Hippolyte Pixii marked a significant milestone in power generation. Antonio Pacinotti further improved this to provide continuous direct-current (DC) power by 1860. Another notable development was the 1870 invention of a dynamo that produced a steady DC source suitable for powering motors, this time by Belgian inventor Zenobe Gramme. By 1877, Charles F. Brush developed and began selling the most reliable dynamo design up to that point, and many were exploring the potential of large-scale electricity distribution. Pacinotti’s innovation around DC in 1860 fueled enthusiasm about electricity’s potential to power the world. This vision became more tangible with Thomas Edison’s invention of an incandescent lamp in 1879, and in 1882, his establishment of the first central power generating station in on Pearl St. in Lower Manhattan.
The late 19th century also saw the emergence of alternating current (AC) technology, developed and refined by Nikola Tesla, who pioneered the modern AC electricity supply system. His polyphase system of AC dynamos, transformers, and motors played a crucial role in the widespread adoption of electricity. Tesla’s AC system caught the eye of George Westinghouse, another AC power pioneer, who bought the rights to Tesla’s patents. This led to the demonstration of an AC power system at the World’s Columbian Exposition in Chicago in 1893, showing the potential of AC power for long-distance transmission. By then, coal power generation had become a significant part of power generation history.
During this time, coal-fired steam generators were also being utilized to provide steam for steam engines driving DC dynamos. However, the environmental impact of such power sources has led to a shift towards more sustainable forms of power generation in recent years.
Modern Day #
Fast forward to the present day, and power has become an overlooked, yet utterly necessary, utility. It is used for a myriad of purposes, including the computation necessary for artificial intelligence and machine learning. As the needs of humanity grow, our power requirements will continue to balloon well-beyond our current generation capabilities. While one half of the solution is to increase and improve power generation, the other half is around power saving. Remember, a watt saved is a watt earned!
Power Usage in Data Centers: The Modern Energy Challenge #
Data centers (DCs) are the heart of the tech industry, hosting the infrastructure that powers everything from online shopping and social media to cloud computing and cryptocurrency mining. However, they also consume a significant amount of power. To understand the magnitude, it’s crucial to delve deeper into the specifics of power usage in data centers and how they are supported by the grid. We will go deeper into those specifics in a future post.
With growing awareness of environmental issues, data centers have been making concerted efforts to reduce power usage and their carbon footprint. Both hardware and software optimizations play a critical role in this transition towards sustainability.
The Energy Impact of Cryptocurrency Mining and LLMs #
Recent tech trends that have reached the general public have had huge impacts on global power consumption. Technologies such as cryptocurrency mining and AI/ML LLMs (large language models) require large amounts of compute, which requires lots of energy. These resource-intensive operations have a significant impact on global power consumption and pose new challenges to sustainable finance in the tech sector.
Cryptocurrency Mining #
Cryptocurrencies like Bitcoin and Ethereum, and their recent volatile periods of boom and bust, have brought the technology into the public eye. Cryptocurrency mining is a process that requires solving complex mathematical problems to create new cryptocurrency coins. This process involves the creation of a blockchain, and miners contribute their computing power to solve these problems and validate the blockchain process. As these mathematical problems become increasingly complex over time, the computational performance required, and therefore the power needed to continue validating the process, scales upward.
Cryptocurrency mining facilities, which are buildings housing the hardware used in the mining process, are significant consumers of energy. Some estimates indicate that energy consumption by these facilities could grow to 100 TWh annually. There are a few ways that the sustainability of cryptocurrency mining could be increased. One method involves optimizing the efficiency of the mining hardware. The higher the efficiency of the hardware, the less energy is required for each unit of cryptocurrency produced, which in turn increases productivity. Another method involves getting creative with the location of mining facilities. For instance, targeting locations with colder climates can reduce the need for cooling, which is a significant energy consumer in these facilities. Additionally, locating facilities in areas with ample hydroelectric and geothermal renewable energy availability can help reduce the carbon footprint of the mining process.
ChatGPT and Other LLMs #
Recently, ChatGPT has exploded into the general public, introducing the concept of LLMs to the masses. Large language Models (LLMs) like GPT-4, which ChatGPT is based on, require a significant amount of computational power, and therefore energy, especially during the training phase. This energy demand comes from the sheer size of the models (GPT-3, for example, has 175 billion parameters), the complexity of the algorithms, and the amount of data processed during training. The actual amount of energy consumed can vary based on many factors, including the specific hardware used, the efficiency of the software, and the duration of training.
Therefore, as ML models continue to grow in number, size, and prevalence, their energy requirements will grow as well. If AI will truly become the future for mankind, it will be crucial to optimize compute efficiency and find renewable sources of energy to continue powering the industry.
Emerging Technologies and the Future of Sustainable Tech #
As we move forward, emerging technologies hold the potential to significantly reduce the tech industry’s power consumption. These technologies include energy-efficient hardware, renewable energy sources, and AI-driven power management systems. For example, Red Hat’s open-source project Kepler, or Kubernetes-based Efficient Power Level Exporter, brings sustainability to the cloud native landscape. Stay tuned for a deeper dive into Kepler and Red Hat’s integrations of it with OpenShift!
The tech industry’s journey from power usage to resource optimization is a testament to its commitment to sustainability. As we continue to push technological boundaries, the balance between innovation and sustainability will be paramount. The future of tech lies not just in its ability to solve complex problems but also in its potential to do so sustainably.