The rapid rise of artificial intelligence (AI) is creating a growing energy burden, but the most intensive drain comes not from chatbots like ChatGPT, but from AI video generators. Tools such as OpenAI’s Sora and Google’s Veo are becoming viral sensations, yet their computational cost is drastically higher than text-based AI — a trend with significant implications for energy consumption and infrastructure.
The Energy Gap: Video vs. Text
Generative AI demands substantial power, but video creation dwarfs other applications. A recent study by Hugging Face found that generating a single 10-second AI video consumes approximately 90 Watt-hours. In comparison, image generation requires just 2.9Wh, while text generation needs a mere 0.047Wh.
This disparity is simple: video requires generating multiple high-resolution images per second. The process is computationally intensive, involving complex “denoising” to create fluid motion. To illustrate, creating one AI video uses roughly the same energy as running a modern 65-inch television for over half an hour.
Why This Matters: The Scale of the Problem
The energy gap isn’t just academic. AI video is exploding in popularity. OpenAI’s Sora reached over one million downloads within five days of launch, and Google’s Gemini has generated over 40 million videos in its first months. As usage expands, existing energy grids may struggle to keep up.
This has prompted massive investment in AI infrastructure. Nvidia is injecting $100 billion into OpenAI to build data centers capable of generating 10 gigawatts of power. Even more extreme, Microsoft is exploring the reopening of the Three Mile Island nuclear site — the location of the worst nuclear disaster in US history — to fuel its AI ambitions.
Transparency and Mitigation
The industry’s lack of transparency is a key concern. Companies are reluctant to disclose the energy footprint of their models, leaving users unable to make informed decisions.
“AI companies should be transparent about their environmental impacts… It’s unacceptable that for tools that we use each day, we don’t have the precise numbers,” says Sasha Luccioni, AI and climate lead at Hugging Face.
Consumers can mitigate their own usage by critically evaluating whether AI tools are truly necessary. But the core issue remains: AI video generation is a high-energy process, and its growth demands a serious, transparent discussion about sustainability.
The future of AI depends on finding solutions to its energy footprint, or risk creating an unsustainable technological boom.
