Is Our Tech Obsession Destroying the Planet?
Imagine if watching your next AI-generated Hollywood blockbuster meant charging 266 smartphones every day for five years. Sounds absurd, right? Yet, that’s the scale of energy consumption we could soon face as AI-driven movies go mainstream later this decade. While digital advances like AI and blockchain transform industries and daily lives, they also amplify our climate and resource crises. Consider this: generating one AI image uses as much energy as fully charging a smartphone, escalating quickly when multiplied over millions of daily interactions.
As technology reshapes our world, the environmental costs surge too. Cryptocurrencies like Bitcoin now consume as much electricity annually as entire countries, and the next wave of humanoid robots threatens to flood us with e-waste far beyond current recycling capacities. However, alongside these challenges lie tremendous opportunities. Innovations such as sleek AR glasses and smart lenses could drastically reduce our e-waste footprint, while breakthroughs in fusion energy promise limitless, clean power.
The key lies in making thoughtful decisions today. By adopting circular economies, investing in renewable energy, and committing to sustainable innovation, we can balance technology’s benefits against its environmental impact. Are we prepared to make bold choices now, or will we risk our future chasing progress without considering the cost?
The Energy Required for AI Images
First, the required energy consumption of one AI-generated AI will prove to become problematic once we transition to entire AI-generated Hollywood movies, something that I foresee before the end of this decade. A 2024 study by researchers at Hugging Face and Carnegie Mellon University found that generating a single image with a state-of-the-art AI model, such as Stable Diffusion XL, consumes as much energy as fully charging a smartphone.
This figure may seem insignificant at first glance, but when scaled across billions of daily interactions globally, the carbon footprint becomes staggering, especially when we move into the territory of AI-generated movies. A 90-minute Hollywood movie shot at 24 frames per seconds (or 23.976 if you want to get technical, but for clarity let's just say 24), means there are 1,440 frames per minute of film, or roughly 129,600 frames (i.e. images) for a feature film. However, this does not include the countless images required to get one useful image, so we can probably easily triple that number, resulting in roughly 388.800 images, or the equivalent of 266 smartphones charged daily for five years, illustrating the substantial energy demands of generative AI tools.
Although the study revealed that image generation is the most energy-intensive task compared to others like text generation, which requires significantly less power, new models that incorporate reasoning when providing their answers significantly increases the energy consumption. OpenAI’s o3 model, announced late 2024, exemplifies the growing environmental concerns tied to next-generation AI systems.
Benchmark results reveal that a single task on o3’s high-compute version consumes approximately 1,785 kWh of energy, the equivalent to the electricity usage of an average U.S. household over two months. This translates into 684 kilograms of CO₂ emissions, comparable to the carbon output of burning over five full tanks of gasoline. These figures represent only the GPU energy consumption during computation, excluding embodied carbon and other infrastructure costs, meaning the actual environmental toll will likely be higher.
The Increasing Energy Demands of Bitcoin
While AI models like o3 demonstrate the immense energy demands of emerging technologies, Bitcoin exemplifies how such requirements can escalate to a global scale. Bitcoin’s annual energy consumption ranges from 91 to 172 terawatt-hours (TWh), placing it on par with entire nations like Finland or Poland. Each Bitcoin transaction alone consumes approximately 851.77 kWh, equivalent to nearly a month’s electricity usage for the average U.S. household, a stark contrast to the minimal energy required for conventional financial transactions, such as those conducted via Visa.
The root of Bitcoin’s energy intensity lies in its Proof of Work (PoW) consensus mechanism. This decentralized process drives miners to compete in solving complex cryptographic puzzles, necessitating vast computational power and contributing to its high energy consumption.
Despite efforts to offset this impact, over half of Bitcoin mining operations reportedly utilize renewable energy, its environmental toll remains significant, with carbon emissions estimated at 22 to 23 million metric tons annually. These emissions rival those of entire nations, such as Greece, underscoring the critical need for more sustainable approaches to blockchain technologies, such as moving to a Proof of Stake (PoS) consensus mechanism (when Ethereum switched to PoS, it slashed its energy consumption by 99.99%).
Humanoids and e-Waste
Taking it up another level, we move from the immense energy demands of technologies like AI models and Bitcoin to an equally pressing concern: the monumental e-waste challenge posed by the hardware that supports these innovations. While the environmental costs of energy consumption are staggering, the sheer volume of physical waste from devices like humanoid robots, projected to 10 billion by 2040 as we discussed earlier, threatens to overwhelm global recycling systems and exacerbate ecological degradation.
The average smartphone generates about 90-110 grams of e-waste at the end of its life due to components like batteries, circuit boards, and casings, or roughly 50% of its total weight. Humanoid robots, being more complex, could produce substantially more e-waste. A single humanoid robot like Tesla’s Optimus 2 weighs approximately 57 kg. If we take a conservative amount of 60% it equals around 34.2 kg of e-waste per humanoid.
E-waste often contains hazardous materials such as lead, mercury, and cadmium, which pose environmental and health risks if not disposed of correctly. The introduction of billions of humanoid robots into the market could exacerbate these issues unless effective recycling systems are established. If each humanoid robot generates an estimated 34.2 kilograms of e-waste at the end of its lifecycle, this would result in a staggering 342 million tonnes of extra e-waste. To contextualize, this is almost six times the total global e-waste produced in 2022, which stood at 62 million tonnes, and represents a significant addition to the already unsustainable trend projected to hit 82 million tonnes annually by 2030.
The increasing amounts of e-waste underscores the urgent need for scalable recycling innovations, such as robotics-enabled disassembly lines or AI-driven waste sorting, to manage this deluge effectively. However, all this e-waste also presents an enormous opportunity, as the global worth of raw materials from e-waste was an estimated $57 billion annually in 2020.
As we edge closer to this predicted future, it is clear that the scale of the challenge demands immediate action. Solutions must prioritize circular economies, modular designs for easier repair and recycling, and global policies to standardize and enforce e-waste management practices. The rise of humanoid robots, while technologically transformative, serves as a stark reminder of the environmental costs of progress and the urgency of addressing sustainability in innovation.
A Sustainable Digital Future
Fortunately, a more sustainable future for consumer electronics is on the horizon. Within the next decade, I believe that the bulky electronics of today, laptops, tablets, and smartphones, are likely to give way to sleek AR glasses and even smart AR lenses. These devices, driven by advances in miniaturization and material efficiency, promise to significantly reduce e-waste.
By shrinking electronic components and designing for durability, AR glasses could offer a more environmentally friendly alternative, lessening the strain on global recycling systems, landfills, and ecosystems. While the current e-waste surge from traditional and future devices remains a challenge, this transition marks a hopeful shift toward sustainable technology.
Scaling renewable energy is the ultimate answer to technology’s energy challenge. Solar power is already a game changer, with projects like PowerWells in Queensland blending recycled e-waste with solar solutions to tackle energy poverty. But the real revolution lies in fusion energy, a source so abundant and clean it could power the insatiable demands of AI, blockchain, and any technology we might invent in the centuries ahead.
Fusion energy promises a future where energy scarcity becomes a distant memory, complementing solar and wind to stabilize grids and sustainably fuel our technological ambitions. However, the path to commercial fusion is fraught with technical and financial hurdles, necessitating global collaboration and sustained investment to bridge the gap between experimental breakthroughs and grid-ready solutions.
The message is clear: aligning technological innovation with environmental stewardship demands bold action. Governments must incentivize green practices, industries must prioritize sustainability, and individuals must advocate for responsible innovation to secure a liveable future for all.