TLDR: Projections indicate that the energy consumption of OpenAI, particularly driven by advanced generative AI applications like text-to-video, could escalate to levels comparable to the total electricity consumption of an entire nation like India. This surge is attributed to the increasing complexity of AI models, the rapid expansion of data centers, and the growing volume of user requests, raising significant concerns about environmental impact and resource sustainability.
The rapid advancement and widespread adoption of generative artificial intelligence (GenAI), especially in sophisticated applications such as text-to-video models, are poised to dramatically increase the energy demands of leading AI developers like OpenAI. Recent analyses suggest that the power consumption required to fuel these cutting-edge AI operations could soon reach levels equivalent to the total electricity generation of a major country like India.
According to a report from Entrepreneur published on September 8, 2025, the massive growth of AI is driving soaring power and and water use in data centers. The International Energy Agency (IEA) projects that energy consumption from AI and digital technologies will surge from 415 TWh in 2024 to an estimated 945 TWh annually by 2030. For context, India generated a total of 1,949 TWh of electricity in the fiscal year 2023–24. This means that by 2030, the projected energy use from AI and digital technologies alone could amount to more than half of India’s current total power generation.
The energy intensity of AI operations is significant. A single ChatGPT request, for instance, requires approximately ten times more electricity than a standard Google search, as highlighted by the IEA. OpenAI CEO Sam Altman previously revealed that each request to its popular generative AI app ChatGPT consumes, on average, 0.34 Wh of electricity, which is between 10 and 70 times that of a Google search. This figure underscores the substantial energy footprint of even seemingly simple AI interactions, let alone the far more complex and resource-intensive text-to-video generation.
The surge in energy demand is multifaceted. Beyond the computational power for training and inference, the infrastructure supporting AI, primarily data centers, is expanding at an unprecedented rate. There are currently about eight million data centers worldwide, a stark increase from just 500,000 in 2012. This expansion, coupled with a projected 378 million AI users in 2025 (up from 116 million five years prior, according to KPMG), creates immense pressure on energy grids.
Furthermore, the manufacturing of the specialized semiconductors that power AI is highly water-intensive. A Morgan Stanley report indicates that semiconductor facilities can consume up to five million gallons of ultrapure water daily, equivalent to meeting the daily water needs of around 140,000 people. The report also projects an eleven-fold increase in annual water consumption for semiconductor manufacturing by 2028, reaching approximately 1,068 billion liters from 2024 levels.
Recognizing these environmental challenges, tech giants are exploring solutions. Many now offer miniature versions of their large language models with fewer parameters, such as Google’s Gemma, Microsoft’s Phi-3, and OpenAI’s GPT-4o mini. A UNESCO study, unveiled in July 2025, suggested that a combination of shorter queries and using more specific models could cut AI energy consumption by up to 90 percent without sacrificing performance. This highlights a growing awareness within the industry and among international bodies about the need to curb AI’s equally vast energy consumption.
Also Read:
- Novel Training Approaches for Diffusion Models Significantly Enhance Generative AI Efficiency
- AI Boom Fuels Nvidia’s Trillion-Dollar Vision and OpenAI’s Expansion
The implications of AI’s escalating energy demands are profound, necessitating a concerted effort from developers, policymakers, and consumers to foster more sustainable AI development and deployment practices.
