Title: Can We Increase Sustainable Energy with AI?
Resource URL: https://www.youtube.com/watch?v=cgzTueXGKKk
Publication Date: 2025-03-28
Format Type: Video
Reading Time: 5 minutes
Contributors: Jonathan Koomey;Trevor Noah;
Source: Microsoft (YouTube)
Keywords: [Artificial Intelligence, Sustainability, Data Center Energy Use, Greenhouse Gas Emissions, AI and Climate Change]
Job Profiles: Academic/Researcher;Academic and Research Collaborations Unit;Artificial Intelligence Engineer;Chief Technology Officer (CTO);Chief Executive Officer (CEO);
Synopsis: In this video, host Trevor Noah speaks with energy analyst Jonathan Koomey about the real energy impact of AI and data centers. Koomey discusses their potential role in either accelerating or delaying net-zero goals.
Takeaways: [Data centers and AI currently account for about 5% of global electricity use, but this enables significant efficiencies in how other energy is consumed., Public concern over AI’s energy use often stems from localized spikes, not nationwide trends., AI is already being used to optimize energy usage in data centers, such as improving cooling efficiency., The direct energy footprint of AI systems like ChatGPT is relatively small compared to other daily electricity uses, like TV consumption., The key metric to evaluate AI's net impact is not energy use alone, but whether total greenhouse gas emissions are decreasing.]
Summary: AI and datacenter-related electricity use is not the dominant factor in rising global energy demand. Although local increases in power usage have raised public concern, these are not reflective of national or global trends. Datacenters use about 5% of total electricity, a figure that includes AI workloads. However, this usage contributes significantly to improving the efficiency of the remaining 95% of energy consumption across sectors.

Direct comparisons, such as the energy needed to run ChatGPT for a day versus powering all U.S. televisions for the same period, reveal that AI’s energy footprint remains modest. Furthermore, AI is increasingly being leveraged to enhance the energy efficiency of datacenters themselves, particularly in cooling systems.

The broader and more consequential question is whether AI will catalyze systemic changes that drive down overall emissions. While the energy consumption of AI is growing, its potential to optimize systems, inform better decision-making, and accelerate decarbonization in other sectors may outweigh its direct impact. The definitive measure of progress remains whether total greenhouse gas emissions are decreasing across society
Content: ## Introduction

### Context
The following is a structured dialogue conducted from Redmond, Washington, between a host and an energy and data analytics specialist. The conversation centers on the global pursuit of net-zero greenhouse-gas emissions and the role that artificial intelligence (AI) may play in either accelerating or impeding this effort.

### Objective
The principal question addressed is whether AI—and the data centers that support it—will ultimately facilitate or delay progress toward net-zero emissions.

## Energy Consumption of AI and Data Centers

### Fundamental Principles
All computing devices require electricity. When a user submits an AI prompt or a search query, that request traverses a network to a large data-center facility. These facilities consume power not only for computational tasks but also for cooling systems essential to maintain optimal operating temperatures.

### Current Consumption Metrics
Across the economy, information-technology systems—including data centers—account for approximately 5 percent of total electricity use. Despite headlines suggesting runaway energy growth, this share has enabled significant efficiency improvements in other sectors.

### Efficiency Gains
Since the early 2000s, the energy efficiency of computing has doubled roughly every 2.6 years. This trend illustrates how investments in data-center infrastructure can yield broader productivity and efficiency benefits across transportation, manufacturing, and building operations.

## Assessing AI’s Contribution to Electricity Demand

### Direct vs. Indirect Impacts
Public concern frequently focuses on the direct electricity consumption of data centers. Yet projections by international energy authorities indicate that over the next decade, the principal drivers of increased electricity demand will be:

- Electrification of road transport
- Electrification of heating and buildings
- Electrification of various industrial processes

Data centers are anticipated to represent only a small fraction of this growth.

### Comparative Examples
To contextualize AI’s footprint, one illustrative comparison is the energy required to operate a leading AI service for a full day versus powering all television sets in the United States for the same period. Although exact figures vary, the AI service’s daily consumption remains modest relative to national appliance use.

## Weighing Costs and Benefits

### Evaluating Net Effects
Rather than judging data-center electricity use in isolation, it is crucial to consider the benefits derived. AI applications can optimize logistics, enhance process controls, and reduce waste across many industries. These indirect effects often outweigh the initial energy investment.

### AI-Driven Efficiency Innovations
AI methodologies are already being deployed to improve the cooling efficiency of data-center facilities. Such advances suggest that AI could help address its own energy demands and contribute to smarter energy generation, distribution, and consumption at scale.

## Tracking Progress Toward Net Zero

### Key Performance Indicators
Society’s advancement toward net zero hinges on aggregate greenhouse-gas emission trends. Metrics should encompass all emission sources—transportation, industry, buildings, and energy supply—rather than isolating AI or data centers.

### The Enabling Role of AI
While precise forecasting is challenging, continued investment in AI and associated infrastructure is expected to catalyze transformative, economy-wide changes. These structural shifts may prove decisive in reducing emissions and stabilizing the global climate.

## Conclusion
AI and data centers represent both an energy demand and a powerful tool for efficiency gains. By monitoring total emissions and encouraging AI applications that optimize energy use, society can harness technology to accelerate progress toward net-zero targets rather than impede it.