Artificial intelligence is accelerating at breakneck speed, transforming almost every industry from healthcare through to logistics. Yet, as AI continues to reshape the world, it’s placing a rapidly growing strain on one crucial resource:
Energy.
As AI’s energy requirements are on track to eclipse that of entire developed nations within the next few decades, tech companies are being forced to choose between adhering to environmental targets or being able to meet the insatiable energy demands of AI, a decision-point that has led the industry as a whole to regress into a worrying dependence on older, non-sustainable power sources.
In an era of climate urgency, the question is no longer just what AI can do for us, but instead, what price we’re willing to pay for its power.
The Hunger of AI
AI’s energy requirements are massive - and climbing rapidly. Running a single data centre for AI training, cloud storage, and processing demands an extraordinary amount of electricity, much more than traditional computing workloads.
By 2035, US-based data centres alone could consume 480 terawatt-hours (TWh), approaching a whopping 10% of the country’s (already vast) total energy use. Globally, data centres may reach 1,000 TWh by 2026, close to double Germany’s entire annual power consumption.
As AI capabilities grow, these demands are expected to continue to rise exponentially. Microsoft’s environmental report already reflects this reality, with its next-gen AI data centres using 29% more electricity than just four years ago, reversing the trendline we were all hoping to see.
Renewable energy has made massive strides but is struggling to keep pace with this rapidly spiralling demand. Solar and wind power, while critical to future energy systems, remain subject to weather fluctuations that can’t currently reliably meet AI’s 24/7 power needs. As a result, tech giants are increasingly turning to alternatives, including nuclear power and, worryingly – fossil fuels, particularly natural gas. This is a troubling backwards slide toward unsustainability that’s sparking complex debates about AI’s true environmental cost.
Nuclear Power: Clean but Contentious
To support AI’s energy appetite without further inflating their carbon footprints, some tech companies are leaning toward the oft-maligned nuclear power. Google recently announced plans to power its data centres with energy from small modular reactors (SMRs), promoting them as clean and dependable. Similarly, Microsoft has invested in a partnership with Constellation Energy to bring a previously defunct Pennsylvania nuclear plant back online, ensuring a consistent, emissions-free energy source for its AI infrastructure.
Nuclear power offers a key advantage: a steady, carbon-free power supply that doesn’t depend on weather. Yet, the choice remains fraught. Nuclear waste, safety risks, and disasters like Fukushima & Chernobyl loom large in public memory, making nuclear power a polarising option.
For companies like Microsoft, the stability of nuclear power outweighs these risks, especially as they race to expand their AI capabilities. But the move isn’t without complications. Microsoft’s deepening partnerships with fossil fuel giants show that even as they embrace nuclear and renewable power, tech companies still struggle with competing interests in the journey toward sustainability.
AI and the Fossil Fuel Revival
Amidst the clamour to scale clean energy, the fossil fuel industry is quietly benefiting from AI’s enormous power needs, particularly in the realm of natural gas. Toby Rice, CEO of EQT - the largest gas producer in the US - views AI’s rise as a “boon” for natural gas, comparing its potential to the US shale boom. It’s a perspective shared across the entire industry, with energy executives arguing that renewables alone can’t yet satisfy the colossal energy appetite of AI-driven data centres.
In particular, Microsoft’s dual role is emblematic of the tension between AI’s environmental promise and its environmental cost. Whilst the company is publicly committed to sustainability but continues to work closely with fossil fuel giants like Chevron and ExxonMobil, promoting AI tools designed to optimise oil and gas production. These partnerships are enormously lucrative, yet they tarnish Microsoft’s green credentials, illustrating the broader struggle within the tech industry to balance growth and shareholder appeasement with responsible environmental stewardship, a balance that it has, thus far, been very poor at maintaining.
The Nuance
As if it wasn’t already complex enough, It’s not all a one-way street. AI isn’t just a voracious energy consumer it’s also helping to make energy systems more efficient and resilient. As renewable energy sources expand, balancing supply and demand becomes increasingly complex, requiring intelligent solutions to smooth out fluctuations in energy production and consumption. AI systems are helping to stabilise the grid by optimising energy storage and distribution.
In Texas, for example, an additional 5 gigawatts (the equivalent of about four large nuclear power plants’ worth) of battery storage since 2021 has helped ensure energy reliability by capturing excess power and releasing it during peak demand. AI tools from companies like WattTime use predictive algorithms to control battery storage, ensuring the batteries are charging and disseminating power at optimal times. Meanwhile, platforms like Electricity Maps forecast green energy output, enabling companies such as Google to power data centres during periods of high renewable availability. These AI-driven innovations represent a significant step forward, helping to bring the vision of a low-carbon energy grid closer to reality.
AI is also monitoring power infrastructure to prevent outages and environmental hazards. Buzz Solutions, for example, deploys AI to inspect power lines for corrosion, vegetation encroachment, and wildlife interference—issues that can cause fires and other disruptions, especially in high-risk areas like California. This capability to handle unpredictability, especially in an environmental context, is going to be essential in our likely destabilised future.
Navigating Tomorrow
The rapid rise of AI in the tech landscape has triggered an energy crisis that demands both immediate attention and innovation. As AI reshapes industries and enables breakthroughs, it also intensifies our dependence on high-powered data centres, forcing tech companies to make tough choices about energy sources. The reliance on nuclear energy and partnerships with fossil fuel industries highlight a critical truth:
Without strategic intervention, the environmental costs of AI could conceivably outpace its benefits.
To fully unlock AI’s promise without compromising the planet, companies must go beyond temporary fixes and adopt a multi-layered approach. This means investing heavily in cleaner, scalable energy sources, accelerating advancements in energy efficiency, and addressing the ethical complexities of fossil fuel partnerships.
We’re at an inflection point: AI can either become a catalyst for sustainable progress or entrench us deeper into outdated energy dependencies at a point when time is a luxury we simply do not have.
As the stewards of this technology, the tech industry has a responsibility to lead, ensuring that the future of AI aligns with our environmental goals. Only by pursuing a path of responsible innovation - one that actually balances AI’s power with a real commitment to sustainable energy, backed by action - can we create a digital landscape that truly benefits both industry and the planet.