The official website of the World Economic Forum recently reported that in order for artificial intelligence (AI) to unleash its transformative potential, improve productivity and social well-being, humanity must ensure its sustainable development. The core challenge facing this vision is that as computing power and performance continue to improve, energy consumption is also rapidly increasing. The AI ecosystem, from hardware, training protocols to operational technology, consumes a significant amount of energy. In view of this, scientists are trying to make it more energy-efficient. The measures include changing the operational strategy of AI, developing more energy-efficient algorithms and chips, etc. Energy consuming AI is an energy intensive technology. How much energy does it consume? The data provides the answer. According to a report on the website of the French newspaper "Echo", Sasha Luccioni, a researcher and environmental director at the AI platform "Embrace Face", said that the electricity consumed by AI algorithms such as Midjournal or Dall-E to generate an image is equivalent to fully charging a smartphone. A Nvidia H100 graphics processing unit consumes more electricity in a year than a medium-sized American household. The Harvard Journal website suggests that the large language model performs better in generating text that is similar to humans, coherent, and contextually logical. But this improvement also comes at a cost, the energy consumed by training GPT-3 is equivalent to the energy consumed by 120 American households in a year. The report from The New York Times is even more astonishing, as ChatGPT responds to approximately 200 million requests per day, consuming over 500000 kilowatt hours of electricity during this process. According to data from the World Economic Forum, the annual growth rate of energy consumption required to run AI tasks ranges from 26% to 36%. This means that by 2027, the energy consumption of the AI industry will be comparable to the annual energy consumption of countries such as Iceland or the Netherlands. Changing operational strategies to make AI more energy-efficient is imperative. Firstly, adjust the AI operation strategy. The official website of the World Economic Forum points out that AI operation generally consists of two main stages: training stage and inference stage. During the training phase, the model learns and develops by digesting a large amount of data; After training, they enter the inference stage to solve the problems raised by the user. Limiting energy consumption in these two stages can reduce the total energy consumption of AI operation by 12% to 15%. Professor Thomas Dietrich from the School of Electrical Engineering and Computer Science at Oregon State University in the United States pointed out that another effective strategy is to optimize scheduling to save energy. For example, running lightweight tasks at night or larger projects in cold months can also save a lot of energy. In addition, transferring AI processing to data centers can also help reduce their carbon emissions, as the operational efficiency of data centers is very high, some of which use green energy. In the long run, promoting the synergy between AI and emerging quantum technologies is also an important strategy to guide AI towards sustainable development. The energy consumption of traditional computing increases exponentially with the increase of computing demand, while the relationship between quantum computing and energy consumption increases linearly. In addition, quantum technology can make AI models more compact, learn more efficiently, and improve their overall functionality. One driving force behind the competition between new models and new device AI companies is the belief that the bigger, the better