Introduction
Artificial Intelligence (AI) has had an immense impact on the global economy. The advancement of technology across industries is rapidly progressing with AI taking the lead in innovation for healthcare, finance, manufacturing and transportation industries.
During this transitional period, one major factor that is often overlooked is how much energy the demand for AI will ultimately cost in terms of increased electricity prices in the United States.
OpenAI, Google and Microsoft are major players in establishing the necessary infrastructure to support the growing need for AI, and this infrastructure is expanding at a greater speed than ever before.
Data centers are a key driver of electricity demand due to the amount of energy consumed by data centers, and as the amount of AI being processed within data centers continues to grow, so too will the demand for energy from the data centers that support AI.
Therefore, there is growing concern that electricity prices will increase as a result of the additional demand created by AI in America.
In this article, we will discuss the interrelated nature of AI and electricity costs and how various driving forces such as the demand for energy, the role that data centers serve, the limitations on infrastructure, potential policy actions and if technological advancements will assist with the rising demand for electricity caused by AI.
The Energy Demands of Artificial Intelligence
The way AI systems consume electricity requires a large amount of computing power to run. Processing millions of parameters, and multiple billion and trillion AI parameters use thousands of best-performing graphical processing units (GPUs) to train a newly developed artificial intelligence model.
Data centers use an enormous amount of power for performing computations, cooling computer servers, and maintaining data center operations.
The need for high amounts of power also comes from running AI inference workloads — which refer to using AI models in live applications. With the continued development of AI applications (e.g., virtual assistants, search engine results), the amount of energy consumed by these systems will accelerate at an exponential rate.
Data Centers: The Core of AI
Physical data centers are the backbone for AI. Many of the leading tech companies are spending billions of dollars over the next decade to expand their data center capacity across America. The majority of these data centers, which run 24 hours a day seven days a week, require an uninterrupted power supply.
In addition to electricity, another aspect that is driving the growth of data centers is that many areas, such as Virginia, Texas and California, will have a data center proliferation due to favorable conditions in terms of their access to fiber optic networks and existing electrical infrastructure. This clustering of energy-intensive facilities will increase the amount of stress placed on the local power grids.
The Supply Side: Can the U.S. Continue to Provide Power to Data Centers?
Capacity and Infrastructure Issues with Electrical Grids
The United States electrical grid was not originally meant to handle the additional electricity loads created by rampant demand for AI technologies currently affecting American society. Sections of this electrical grid have been operating for significant lengths of time and are in need of substantial upgrades to fully support the growth resulting from the creation and utilization of these new energy-intensive applications.
There are also challenges on the transmission side of the electricity industry, such as transmission bottlenecks that will limit the effective distribution of electricity from the generation side to areas that utilize large amounts of electricity when using AI technologies.
As more and more data centers gather into certain locations, local utility companies may find themselves unable to supply enough electricity for the growing demand unless they build new infrastructure.
Renewable Energy and its impact on demand
Renewable energy sources such as solar power and wind energy are two potential solutions to increasing demands for electricity. Technology companies (including technology giants) are among the largest users of renewable energy and sign long-term contracts to purchase it in order to offset the carbon footprint associated with their operations.
Renewable energy comes with its own unique challenges. For example, the primary challenge associated with renewable energy is that it has an intermittent supply, which means (for example) that during cloudy days there may be very little solar energy produced and during still days there may be no wind generated. In both of these cases the utility provider will need to have some means (e.g., batteries or gas) to provide electricity when the renewable energy generation is not available.
How will the electricity prices react?
Immediate demand increases will likely put upward pressure on the price of electricity in some regions due to the fact that electricity prices are generally tied to the balance of supply and demand in the market, and when demand exceeds supply prices generally rise. This is particularly true in areas where data centres are located in large amounts.
Utilities may choose to pass on costs associated with the necessary infrastructure and generator upgrades to their customers therefore causing residential customers and small business customers to see a higher electric bill.
Differences in Price by Region
The price of electricity varies by region across the United States. The states or jurisdiction with deregulated energy markets will see their electricity prices fluctuate more quickly than those states or jurisdiction with regulated energy markets.
In areas with high amounts of renewable energy or excess resources available, there’s generally a better opportunity to accommodate more demand without raising prices to unacceptable levels.
All of these are counterarguments to what efficiency could mean to consumers.
Renewable Energy Optimized By Artificial Intelligence
Interestingly, artificial intelligence may provide some relief from higher rates created by itself. Using sophisticated computer algorithms, AI can find optimum use patterns for electricity throughout the grid, resulting in a more efficient operation and minimization of energy waste.
An example of this would be to accurately predict demand used by electric utilities so they can properly allocate resources. AI can also be used to optimize air conditioning systems located in data centres, helping to reduce total energy consumption.
Hardware Improvements
Ongoing advances in hardware will assist in further reducing the amount of energy required to run AI computations. Current generations of chips are reducing energy usage associated with running AI applications.
As a result, chip manufacturers are continuously striving to build more energy-efficient chips, while providing higher performance levels per watt.
Therefore, many businesses are making substantial investments in the development of custom silicon chips that will be optimized for running AI workloads. Eventually, these efforts should lead to a significant reduction in total consumption of energy associated with running AI applications.
Policy & Regulation
Government Influence over an Energy Market
The role of local, state and federal governments cannot be overlooked when discussing how to affect change in the energy market. Energy policy reform will include promoting the development of renewable energy, modernization of the electric grid, and energy efficiency – all which will serve to lessen the cost to consumers who rely on electricity for their daily needs as well as help to lessen the ultimate burden on the utilities.
In addition, state and/or federal funds associated with clean energy projects and/or electric infrastructure upgrades will help reduce the costs incurred by utilities and their customers.
Regulatory Issues
There’s a challenge in balancing affordability with the development of new technologies. On the one hand, supporting the development of AI will support U.S. competitiveness, while at the same time, we must ensure that all Americans have access to affordable electricity.
The Role of Large Technology Companies
Corporate Responsibility
Technology companies have become very conscious of their use of energy and many have committed to becoming either carbon-neutral or carbon-negative.
The commitments by these corporations typically require investments in renewables; improving efficiencies in data centers; and seeking out alternative sources of energy such as nuclear or geothermal systems.
Strategic Investments
There are several companies, such as Amazon and Meta, spending money on significant renewable energy investments to operate their companies. These large-scale renewable energy investments not only reduce emissions but give additional energy to the grid as well.
Long-Term Considerations
A Balanced View of the Future
While AI will result in increased electricity demand, it does not mean that all prices will rise significantly; the outcome will vary based on many factors such as infrastructure investments, technology advances and policy decisions.
Alternatives For The Future
Higher Cost Alternative: Traffic congestion could result from inadequate physical and digital infrastructures while at least 10% price increases in the future could result due to failing infrastructures within 2028.
Balanced Development Alternative: Consistent investment and innovating solutions, will lead supply to match demand in a stable way, housing and job creation can mitigate and equalize our demands and thereby maintain stable energy prices
Improved Efficiency Alternative: New AI-enhanced technologies could greatly improve energy efficiency and increase energy supply and thus, create a better price for electricity.
Conclusion
AI provides us with significant opportunities to change our energy system and the cost of electricity moving forward. AI has changed the method in which society consumes power by providing unprecedented levels of power consumption through the introduction of large scale AI based data centers around the world. Some of this growth will occur as a result of changes in efficiency brought about through innovation.
The mobility and flexibility created through advanced telecommunications and artificial intelligence has provided us with vast opportunities to fundamentally change how we consume electricity.
The major issue we face is not alone when we gauge the actual expenses associated with the future potential for electric utilities, but also whether the United States can adapt our energy policies and energy infrastructure to respond to these future realities, as well as respond effectively to the current economic and environmental realities created by rapidly evolving technology and consumer preferences.
