AI’s Hidden Energy problem nobody talks about

Have you ever thought about the AI energy consumption behind a simple ChatGPT question or a DALL-E image generation? While we are impressed with AI’s capabilities, a crisis related to data centers is going silent all over the globe and is threatening the environment and energy grids substantially.

The figures are frightening. In its April 2025 report, the International Energy Agency, estimates that the worldwide electricity demand of data centers will increase over two times by 2030 and will be around 945 terawatt-hours, which is just a bit more than the total energy consumption of Japan . At the same time, Goldman Sachs Research predicts that nearly 60% of the rising electricity demand will be satisfied through the burning of fossil fuels, thus elevating the global carbon emissions by approximately 220 million tons . This AI environmental impact issue, being among the most significant tech challenges of our time, is hardly ever acknowledged.

So, what will be the effect on our planet, and why is this matter not being discussed by more people? In this long-read, we will uncover the concealed energy costs of AI, analyze the potential solutions, and reflect on what this means for AI development in the future.

What is the Problem with AI Energy?

The heart of the AI energy problem is the huge computational power needed to train and run complex models, in particular Large Language Models (LLMs). You may compare an AI model to a digital brain having billions or even trillions of “neurons” (parameters). The process of teaching this brain means that you have to give it enormous data and it has to do a huge number of calculations to change its parameters. This is a very energy-intensive process.

It is said that training of one single large-scale AI model can be a task of thousands of high-powered Graphics Processing Units (GPUs) that have to run non-stop for weeks or even months. The amount of electricity that is used is very high. Also, it doesn’t stop here. The moment when the user interacts with generative AI, which is a process called “inference”, is also energy-consuming. There is an estimate that a ChatGPT query uses nearly ten times the energy of a simple Google search.

This surge in AI energy consumption is having a tangible impact:

• Exponential Growth: AI-related electricity use is projected to grow by as much as 50% annually between 2023 and 2030.​

• Strained Power Grids: Data centers, the backbone of AI, could consume up to 20% of the world’s electricity by 2035, placing immense strain on power grids. In the US alone, data centers could account for 12% of the nation’s power by 2028.​

• Carbon Footprint: Much of this electricity is generated from fossil fuels, which means the AI boom is directly contributing to a significant increase in greenhouse gas emissions and environmental degradation.​

The problem is compounded by the physical infrastructure. Data centers require constant cooling to prevent servers from overheating, which itself consumes a massive amount of energy—sometimes as much as the servers themselves. It’s a compounding issue that has made the AI energy problem a top concern for environmentalists and tech leaders alike.​

Why is AI Bad for Global Warming?

The link between AI and global warming is direct and concerning. The vast majority of the electricity powering the data centers that fuel AI is still generated by burning fossil fuels. This releases enormous amounts of carbon dioxide (CO2) into the atmosphere, which is the primary driver of climate change.​

Studies have shown a clear correlation between AI activities and increased CO2 emissions, particularly in countries with weaker environmental policies. This is sometimes referred to as the “Digital Rebound Effect,” where efficiency gains from technology are offset by a massive increase in consumption.​

However, the environmental impact isn’t just about energy. The AI industry also has a significant water footprint. Data centers use vast quantities of fresh water for their cooling systems, which can strain local water resources, especially in arid regions.​

The paradox is that AI also holds the potential to help fight climate change. AI can be used to optimize energy grids, improve the efficiency of renewable energy sources, and model climate change with greater accuracy. For instance, AI could improve the efficiency of wind and solar energy systems by up to 20% by enhancing grid management. The challenge is to ensure that the environmental benefits of applying AI outweigh the significant carbon footprint of developing and running it. Have you considered how the AI tools you use might be contributing to this complex environmental equation?​

Understanding the Scale of AI’s Energy Appetite

The Numbers Behind the Crisis

Let’s put AI’s energy consumption of AI into perspective. Training OpenAI’s GPT-4 consumed an estimated 50 gigawatt-hours of energy—enough to power San Francisco for three days . To visualize this, imagine leaving a household lightbulb on continuously for over 5,700 years.

The infrastructure demands are equally breathtaking. Tech giants are making unprecedented investments:

• OpenAI and President Donald Trump announced the Stargate initiative, a $500 billion project to build up to 10 data centers

• Apple plans to spend $500 billion on manufacturing and data centers in the US over the next four years

• Google expects to spend $75 billion on AI infrastructure alone in 2025 

These projects represent a fundamental shift in energy demand patterns. From 2005 to 2017, electricity consumption by data centers remained relatively flat despite massive digital expansion, thanks to efficiency improvements. But since 2017—coinciding with the AI boom—data center electricity consumption has doubled by 2023 .

Why AI Demands So Much Power

Why is AI bad for global warming? The reason can be traced back to its core design. AI, unlike traditional computing, needs special kind of chips known as Graphics Processing Units (GPUs), which use a huge amount of energy while calculating trillions of operations.

It’s quite possible that one AI model is spread over several GPUs. As a result, huge data centers might have more than 10,000 such extremely power-consuming chips linked together. These kind of devices produce so much heat that in order to cool them, more energy is needed. Also, cooling systems sometimes use millions of gallons of fresh, potable water daily.

The issue is aggravated by what is going to happen after training. Building AI models is quite an energy-intensive task, however, when these models are used (referred to as “inference”), the energy consumption is 80-90% of total AI computing power. As ChatGPT is rapidly becoming the globally fifth-most visited website, the carbon footprint of AI is increasing with every one of those millions of daily queries.

From Training to Query: AI’s Energy Lifecycle

The Hidden Costs of Model Training

Before you can interact with an AI model, it undergoes an energy-intensive training process. Racks of servers hum along for months, ingesting training data and performing complex computations. This phase represents the initial energy consumption of AI investment.

But what many don’t realize is that much of this energy goes toward marginal gains. Research from MIT’s Lincoln Laboratory Supercomputing Center found that approximately half the electricity used for training an AI model is spent achieving the last 2-3 percentage points in accuracy . For many applications, slightly lower accuracy would be perfectly acceptable at a fraction of the energy cost.

The Cumulative Impact of AI Queries

Does AI usage consume energy unnecessarily? The response varies with the question and the energy grid that supplies it. For instance, a simple text-based question to a chatbot might consume relatively little energy, whereas a complex image or video creation could require significantly more energy.

It is an issue of scale. Individual requests might be so small that they are hardly noticeable, but in total, they have a huge energy footprint. According to Lawrence Berkeley National Laboratory, the energy consumption of the AI/data center sector will be so significant by 2028 that it will account for more than half of the electricity use in data centers. At that time, AI alone could be using as much electricity annually as 22% of all U.S. households.

The pollution caused by the energy used for this is a very big factor. Data centers are known to consume energy that is 48% more carbon-intensive than the U.S. average. This is because, to meet their huge and continuous demand, they usually take the power that is readily available—mostly from fossil fuels.

Solutions on the Horizon: Addressing AI’s Energy Challenge

Technical and Operational Improvements

The news isn’t all dire. Researchers worldwide are developing innovative solutions to mitigate AI’s environmental impact:

1. Hardware Efficiency: Constant innovation in computing hardware continues to deliver dramatic improvements. The amount of computation GPUs can perform per joule of energy has been improving by 50-60% each year .

2. Algorithmic Efficiency: Perhaps even more significant are gains from new model architectures. Neil Thompson, director of the FutureTech Research Project at MIT’s Computer Science and Artificial Intelligence Laboratory, notes that efficiency gains from better algorithms are doubling every eight or nine months . His team coined the term “negaflop” to describe computing operations avoided through algorithmic improvements.

3. Strategic Operation: Researchers are developing methods to schedule computing operations for times when renewable energy is more plentiful on the grid . This “energy-aware computing” can significantly reduce carbon emissions without changing the underlying technology.

Systemic and Infrastructure Solutions

Beyond technical improvements, systemic changes show tremendous promise:

• Smarter Data Centers: Researchers at the MIT Energy Initiative are studying “smarter” data centers where AI workloads of multiple companies are flexibly adjusted to improve overall energy efficiency .

• Long-Duration Energy Storage: Implementing energy storage systems allows data centers to rely more on renewable sources, even during high-demand periods .

• Location Strategy: Placing data centers in cooler climates (like Meta’s facility in Lulea, Sweden) reduces cooling demands, while locating near renewable energy sources decreases carbon intensity .

AI as Part of the Solution

In a promising twist, AI itself could help address its energy challenges. Researchers are exploring how generative AI could streamline the process of connecting new renewable energy systems to power grids—a step that currently takes years . AI could also optimize prediction of solar and wind generation, identify ideal locations for new facilities, and perform predictive maintenance on green energy infrastructure .

Beyond Energy: Other Critical AI Challenges

While energy consumption represents a significant issue, it’s far from the only challenge facing AI development. Understanding these related concerns provides important context for the energy discussion.

Does Using AI Waste Energy?

Calling it a “waste” depends on the value derived from the AI’s output. However, from a purely technical standpoint, there is a lot of inefficiency in the current state of AI energy waste. AI model training is often a brute-force process. Researchers might train dozens of similar models, tweaking small parameters, and discarding all but the best-performing one. This trial-and-error approach consumes immense resources for models that are never even used.

Furthermore, the hardware itself is not perfectly efficient. GPUs and other processors generate a lot of heat, which is essentially wasted energy that then requires even more energy to dissipate through cooling systems.​

However, the industry is keenly aware of this issue. Researchers and tech giants like Google and Microsoft are actively working on solutions. They are developing:​

• More efficient algorithms that require less computational power.

• Specialized hardware (like TPUs) designed for AI tasks.

• Advanced cooling techniques that reduce energy consumption.

These efforts are leading to what some call the 30% rule in AI, which we will explore next.

What is the 30% Rule in AI?

The 30% rule of AI is not a law, but rather an industry benchmark that signifies the energy savings in percentage that can be achieved through “clever” software and hardware optimizations. For instance, the researchers at the University of Michigan have engineered algorithms that can measure the exact amount of electricity an AI chip should be given for a particular task, thus, cutting its energy consumption by 20-30% while maintaining performance.​

Indeed, the idea of energy savings by AI extends beyond merely the chip level. Through the use of smarter programming, the data processing method optimization, and cooling system enhancement, the data center industry is looking at similar efficiency gains in their entire infrastructure. The tech companies are, however, not solely motivated by the desire to lessen the energy footprint as a way to combat the global warming trend, though it is a big reason, but also by economic ones—electrical energy is a major component of the operating costs.​

These 20-30% reductions in energy consumption are an essential factor in the alleviation of the AI energy problem. It exhibits that the problem of AI energy waste caused by existing models can be dealt with through dedicated innovation. This is a proof of principle that AI technological growth does not have to be accompanied by an equivalent increase in power consumption.

As an example, a corporate lawyer could employ AI to scan through contracts in bulk within minutes rather than hours, yet still be dependable on human discretion to interpret unusual cases, evaluate business risks, and offer clients advice. Such a model allows us to see AI as a means to enhance human capabilities rather than completely replacing them.

How Will AI Affect the Energy Industry?

The relationship between AI and the energy industry is a two-way street. While AI is a massive consumer of energy, it is also becoming a transformative tool for the sector.

On one hand, the explosive growth of AI is forcing energy providers to rapidly scale up. Utility companies, especially those in regions with a high concentration of data centers like Northern Virginia, are seeing unprecedented demand. This is creating a boom for both traditional and renewable energy producers.​

On the other hand, AI is being deployed to make the energy industry smarter and more efficient:

• Grid Optimization: AI algorithms can predict energy demand and supply in real-time, helping to balance the power grid, reduce outages, and integrate variable renewable sources like wind and solar more effectively.​

• Energy Efficiency: AI can monitor and control energy use in industrial processes and commercial buildings, identifying and eliminating waste.

• Discovery of New Materials: AI is accelerating the research and development of new materials for batteries, solar panels, and other green technologies.

The industry is at a crossroads. The demand from AI is pushing it to grow, while the intelligence of AI is providing the tools to do so more sustainably. Do you think the positive impacts of AI on the energy industry can ultimately outweigh its own consumption?

For an in-depth report on this topic, read the latest analysis from the International Energy Agency.

Investing in AI Energy: A New Frontier

With great demand comes great opportunity. The AI energy problem has created a new and compelling sector for investors. As Wall Street and Big Tech pour trillions of dollars into AI, they are realizing that none of it is possible without a massive expansion of our energy infrastructure. This has opened up several avenues for those looking to invest in AI energy.​

Utilities and Infrastructure

The most direct beneficiaries are the utility companies that supply power to data centers. Companies like Dominion Energy (D), which services the largest data center market in the U.S., are poised for significant growth. Investing in a broad utility-focused ETF like the Utilities Select Sector SPDR Fund (XLU) can provide diversified exposure to this trend.​

The Nuclear Renaissance

Nuclear energy is experiencing a revival, driven by its ability to provide massive amounts of reliable, zero-carbon power. Tech giants like Amazon and Microsoft are actively investing in the development of next-generation small modular reactors (SMRs) to power their data centers. This has brought renewed attention to companies in the nuclear sector, such as Constellation Energy (CEG), a major operator of nuclear plants.​

Energy Infrastructure and Engineering

Building the infrastructure to support this energy demand is a massive undertaking. Companies like MasTec, Inc. (MTZ), which specialize in engineering and construction for energy infrastructure, are also well-positioned to benefit from this boom. As the grid expands and modernizes, these companies will be on the front lines.​

For investors, the key is to look beyond the AI software companies and identify the businesses that are building the physical backbone of the AI revolution. The AI energy stock landscape is broad and full of potential for those who know where to look.

For a deeper analysis of market trends, visit 2025 Tech Investment Outlook.

Warnings from Visionaries: Hawking and Musk

Long before the current generative AI boom, some of the world’s brightest minds were sounding the alarm about the potential dangers of artificial intelligence. Their warnings were not about energy consumption, but about the existential risks that superintelligence could pose to humanity.

What did Stephen Hawking have to say about AI?

The late, great physicist Stephen Hawking was unequivocal in his concerns. He famously told the BBC in 2014, “The development of full artificial intelligence could spell the end of the human race”. His fear was that AI could “take off on its own, and re-design itself at an ever increasing rate”.​

Hawking’s primary warning was that humans, “limited by slow biological evolution, couldn’t compete, and would be superseded”. He saw the creation of an intelligence far exceeding our own as a monumental event, one that could be “potentially our worst mistake ever”. His Stephen Hawking AI warning was a call for caution and foresight, urging humanity to consider the consequences before creating something it could not control.​

What is Elon Musk’s warning about AI?

Elon Musk, a key figure in the tech world, shares many of Hawking’s concerns, even while actively developing his own AI through his company xAI. Musk has repeatedly stated that there is a non-trivial chance that AI could pose an existential threat. His Elon Musk AI warning often comes with a specific probability: he estimates there’s a “10% to 20% chance that it goes bad”.​

Musk has also speculated that humans might just be a “biological bootloader” for digital superintelligence—a transitional step in the evolution of intelligence on Earth. He advocates for proactive regulation and the development of “safe” AI to mitigate these risks. Both his warnings and his investments underscore the dual nature of AI: it holds immense promise and significant peril.​

What does the Bible say about artificial intelligence?

This might seem like an unusual question, but many are seeking guidance from ancient wisdom to navigate this new technological frontier. The Bible does not, of course, explicitly mention AI. However, its principles on creation, wisdom, and human responsibility can offer a framework for ethical consideration.

Some theologians see the human drive to create AI as an expression of the creativity and intelligence given to humanity as part of being made in “God’s image” (Genesis 1:27). In this view, technology is a natural extension of God-given abilities.​

However, the Bible also contains cautionary tales about human hubris, such as the Tower of Babel, where humanity’s ambition led to its downfall. There are also passages that mock idols as man-made creations that are lifeless and without true understanding, which some might draw parallels to when considering the nature of AI’s “intelligence”.​

Ultimately, a biblical perspective on artificial intelligence would likely focus on stewardship and discernment. It would ask whether we are using this powerful tool for good, to uplift humanity and care for creation, or for selfish or destructive purposes. It encourages wisdom in how we develop and deploy technologies, ensuring they align with values of love, justice, and humility. The question isn’t whether AI is “good” or “bad,” but how we, as its creators, choose to wield it. What are your thoughts on how ancient wisdom should guide modern innovation?

Transparency and Bias Concerns

Another major problem of AI revolves around transparency and bias. Many AI systems operate as “black boxes,” making decisions in ways that aren’t easily explainable to human users . This lack of transparency is particularly problematic in high-stakes domains like healthcare, finance, and criminal justice.

Additionally, AI models trained on biased data can perpetuate or even amplify societal discrimination. When AI systems make unfair decisions in hiring, lending, or law enforcement, they raise serious ethical questions that the industry is still grappling with .

Expert Warnings About AI’s Trajectory

What did Stephen Hawking have to say about AI? The late physicist was famously cautious about superintelligent AI, warning that it “could spell the end of the human race” if such systems began modifying themselves beyond human control . However, his relationship with AI was nuanced—he simultaneously relied on basic AI technology to communicate after losing his ability to speak .

Similarly, Elon Musk’s warning about AI has been consistent: he believes superhuman AI represents an existential risk if deployed incautiously. Both Hawking and Musk were signatories to a 2015 open letter calling for research on AI’s societal impacts . The letter affirmed AI’s potential benefits—including eradicating disease and poverty—while emphasizing the importance of ensuring these systems remain safe and controllable .

Ethical and Spiritual Perspectives

What does the Bible say about artificial intelligence? While scripture doesn’t mention AI directly, some biblical principles apply to this technology. The creation story emphasizes humans being made “in God’s image” with creativity and intelligence . From this perspective, AI can be seen as an expression of God-given abilities.

At the same time, the story of the Tower of Babel serves as a cautionary tale about human pride and technological overreach . This suggests the importance of humility and ethical consideration in AI development. As with any tool, the moral valence of AI depends largely on how it’s used—whether to help others or to elevate human pride beyond appropriate bounds.

The Path Forward: Balancing Innovation and Responsibility

The Role of Policy and Regulation

As AI’s energy demands and other impacts become clearer, governments worldwide are taking notice. In 2024 alone, U.S. federal agencies introduced 59 AI-related regulations—more than double the number in 2023 . Global legislative mentions of AI have risen 21.3% across 75 countries since 2023, marking a ninefold increase since 2016 .

This regulatory attention is essential for creating frameworks that encourage innovation while mitigating harm. Standardized reporting of AI energy use, carbon emissions, and efficiency metrics would help researchers and policymakers make better decisions about AI’s development and deployment.

Industry Initiatives and Best Practices

The tech industry itself is increasingly aware of these challenges. Many companies are exploring more sustainable building materials for data centers  and investing in renewable energy sources. Some are even considering radically innovative locations—including the moon—where data centers could operate with nearly all renewable energy .

Adopting best practices like those identified by researchers—such as stopping training processes early when accuracy thresholds are met, or using less precision in computing hardware for certain applications—could significantly reduce AI’s environmental impact without compromising utility for most use cases .

Investment Opportunities in AI Energy Solutions

The Growing Market for AI Efficiency

How do I invest in AI energy? This question is increasingly relevant as the field grows. The massive energy demands of AI are creating opportunities in several sectors:

• Advanced Cooling Technologies: Companies developing more efficient cooling systems for data centers stand to benefit as AI expands.

• Renewable Energy Infrastructure: With tech companies seeking clean power for their operations, renewable energy providers face growing demand.

• Energy Storage Solutions: As data centers look to manage intermittent renewable sources, advanced battery and storage technologies become increasingly valuable.

• AI Optimization Tools: Companies creating software to improve AI efficiency or reduce computational requirements represent another investment avenue.

The Big Picture on AI Investments

While specific AI energy investments hold promise, it’s worth noting that overall U.S. private AI investment reached $109.1 billion in 2024—nearly 12 times China’s $9.3 billion . Generative AI attracted particularly strong momentum at $33.9 billion globally, an 18.7% increase from 2023 .

This investment landscape suggests broad confidence in AI’s future, even as the industry grapples with its environmental impacts. For savvy investors, companies that successfully address AI’s energy challenges may represent particularly promising opportunities.

Conclusion

The hidden energy problem of AI can no longer remain in the shadows. As this technology becomes increasingly embedded in our lives—from healthcare and transportation to education and entertainment—we must confront its environmental costs with honesty and determination.

The path forward requires collaboration among researchers, companies, policymakers, and the public. We need:

• Greater transparency about AI’s energy use and carbon emissions

• Continued investment in efficiency improvements for both hardware and algorithms

• Strategic placement and operation of data centers to minimize environmental impact

• Thoughtful regulation that balances innovation with responsibility

• Consumer awareness about the digital environmental footprint

The challenge is significant, but so is the potential. With concerted effort, we can work toward a future where AI’s benefits don’t come at an unacceptable environmental cost. The question is, will we make the necessary choices before the energy impact becomes irreversible?

What role will you play in addressing this critical issue? However, the conversation about AI’s future is one we all need to participate in—our collective decisions today will shape the environmental impact of our digital tomorrow.

Frequently Asked Questions

What is the problem with AI energy?

The primary problem with AI energy is its massive and growing consumption of electricity, which often comes from carbon-intensive sources. Training and running large AI models requires enormous computational resources that demand significant power. For example, training OpenAI’s GPT-4 consumed an estimated 50 gigawatt-hours of energy—enough to power San Francisco for three days . As AI becomes more integrated into everyday applications, its energy demands are projected to increase dramatically, potentially accounting for half of all data center electricity by 2028 .

What is the 30% rule in AI?

The 30% rule in AI suggests that in most complex professional roles, about one third of tasks can be automated with current AI technology, while the remaining two-thirds require human expertise, judgment, and emotional intelligence . For instance, a lawyer might use AI to quickly review contracts but rely on human skills for client counseling and strategy. This rule highlights how AI typically augments rather than replaces human capabilities, automating repetitive tasks while leaving complex decision-making to people.

What did Stephen Hawking have to say about AI?

Stephen Hawking warned that superintelligent AI “could spell the end of the human race” if such systems become self-improving beyond human control . However, his relationship with AI was complex—while fearing potential long-term risks, he simultaneously relied on basic AI technology to communicate after losing his ability to speak . Hawking called for more research on both benefits and dangers of AI, believing it could help eradicate war, poverty, and disease if properly managed .

What is Elon Musk’s warning about AI?

Elon Musk’s warning about AI aligns with Hawking’s concerns about superintelligent systems. Musk has repeatedly cautioned that advanced AI could pose existential risks to humanity if deployed without sufficient safety measures. He was a signatory to the 2015 open letter on artificial intelligence that called for research on AI’s societal impacts and how to ensure these systems remain beneficial and controllable . Musk emphasizes the importance of proactive safety research rather than reacting to problems after they occur.

How do I invest in AI energy?

To invest in AI energy, consider companies developing energy-efficient AI hardware, advanced cooling systems for data centers, renewable energy infrastructure, and AI optimization software. The massive energy demands of AI are creating opportunities in these sectors as tech companies seek solutions to power their AI operations sustainably. With U.S. private AI investment reaching $109.1 billion in 2024 , companies that successfully address AI’s energy challenges represent promising investment opportunities.

What does the Bible say about artificial intelligence?

The Bible doesn’t mention artificial intelligence directly, but it offers principles that apply to AI development. Genesis states humans are made “in God’s image” with creativity and intelligence , suggesting AI could be an expression of God-given abilities. However, the Tower of Babel story cautions against technological pride . AI, like any tool, can be used for good (like medical advances) or misused (like promoting pride or creating idols). The technology itself is neutral—its moral value depends on how humans choose to apply it.

What is the major problem of AI?

A major problem of AI encompasses several challenges, with energy consumption being significant. Other critical issues include:

• Lack of transparency in how AI models make decisions (the “black box” problem) 

• Bias in AI systems that can perpetuate discrimination 

• High implementation costs for many businesses 

• Data privacy concerns 

• Integration challenges with legacy systems 

• Ethical concerns about appropriate use cases 
  These problems require comprehensive approaches addressing both technical and societal dimensions.

How will AI affect the energy industry?

AI will affect the energy industry in multiple ways. It will dramatically increase electricity demand—data centers alone may double their consumption by 2030 . This will require significant grid upgrades and new power generation. Simultaneously, AI can help optimize energy distribution, predict renewable output, and improve grid management. The energy industry faces both a challenge in meeting AI’s power needs and an opportunity to use AI for improving efficiency and integrating renewable sources .

Why is AI bad for global warming?

AI is bad for global warming primarily because its massive electricity consumption often relies on fossil fuels, particularly during peak demand or when renewables are unavailable. Current projections suggest about 60% of new electricity demand from data centers will be met by burning fossil fuels, potentially adding 220 million tons of global carbon emissions . Additionally, building data centers creates substantial “embodied carbon” from manufacturing and construction materials . Without mitigation, AI’s carbon footprint could significantly contribute to climate change.

Does using AI waste energy?

Using AI wastes energy when it’s applied to tasks where less energy-intensive alternatives would suffice, or when inefficient models are used. However, not all AI use is wasteful—it can optimize systems to save more energy than it consumes in some applications. The key is using the right tool for the job: simple tasks might not need complex AI, while energy-intensive AI might be justified for important medical or climate research. Researchers are working on more efficient AI to reduce energy waste while maintaining performance .