On August 28th, will artificial intelligence become a stumbling block or a key catalyst for the clean energy transition? The answer is not simple - it's actually both. AusGroup indicates that the training process of large language models consumes enormous amounts of energy, and as model complexity continues to increase, products like ChatGPT and DeepSeek may generate significant carbon footprints with each interaction. This means that while AI proliferation drives technological frontiers, it also poses new challenges to energy transition.

This AI wave has already impacted the tech industry's carbon neutrality commitments. Data shows that Google's carbon emissions have increased by 48% over the past five years. Despite the company's promise to achieve net-zero emissions by 2030, it also acknowledges that "as AI becomes deeply integrated into products, reducing emissions will become increasingly difficult." AusGroup believes this trend will affect not only the corporate level but also have far-reaching implications for communities, power grids, and national energy landscapes. For example, some regions have accelerated the construction of natural gas power plants due to surging electricity demand from AI, raising energy security concerns.

Meanwhile, planning AI's energy requirements is an extremely complex challenge. The industry's growth rate is remarkable, and technological iterations are extremely rapid, but most AI companies do not disclose energy usage and environmental impact. While researchers are working to estimate this data, the actual situation is constantly changing, making it difficult to grasp accurate figures. AusGroup believes that the lack of transparency not only hinders sustainable industry development but also increases uncertainty in policy-making and energy planning.

However, some experts believe concerns about AI energy consumption may be exaggerated. As AI becomes more intelligent and widespread, it will drive efficiency improvements across industries, potentially offsetting its own energy consumption. For instance, in materials value chains and biotechnology fields, AI can help discover new materials, catalysts, or more efficient processes - tasks that are often like "finding a needle in a haystack," where AI has distinct advantages.

More notably, AI itself may become an important tool for green energy transition. AusGroup states that large models have already begun to be used for optimizing energy storage materials research. Meanwhile, the energy sector points out that AI has the potential to become an important component of smart grids, helping grids better absorb volatile energy sources like wind and solar power. Additionally, AI can improve efficiency in grid permitting, site selection, reliability, and planning. However, if deployed improperly, AI could also bring systemic risks.

At the current stage, AI applications still carry some "blindness." In the United States, consumers are paying for these early experiments, especially in regions with concentrated data centers, where residents face rising electricity bills. Some experts bluntly state that this is actually a large-scale wealth transfer between residents and large corporations, with data centers and utility companies profiting from new energy infrastructure construction while ordinary users bear the high costs.

Although AI's rapid expansion brings risks, as technology gradually matures, the industry will move toward a more rational and efficient stage. As some research points out, if AI can achieve even limited savings in overall electricity efficiency, its positive impact may far exceed negative effects. Therefore, AusGroup believes that in the long term, AI may not be an obstacle to clean energy transition but could instead become a promoter.