Artificial intelligence (AI) technologies, particularly advanced models such as generative AI, are known for their substantial energy consumption. This is primarily due to the intensive computational processes required for training and operating these models. The high demand for computational resources, including powerful processors and extensive data storage, contributes to a significant carbon footprint. As AI continues to evolve and integrate into various sectors, addressing its environmental impact has become a critical challenge. In this presentation, we will explain how we have set up an AI provider as sustainably as possible, exploring strategies to efficiency, leverage renewable energy sources, and implement responsible AI practices. By adopting these approaches, we aim to mitigate the environmental impact of AI while still harnessing its transformative potential.
GreenPT is an open‑source AI platform hosted on proven sustainable providers across Europe and powered by renewable energy, which combines efficient model selection with per‑prompt energy‑tracking to minimise the carbon footprint of generative AI while delivering advanced reasoning, code generation, vision analysis and multilingual capabilities; the presentation also shows how vague or overly broad prompts generate extra tokens and clarification loops that increase emissions, and outlines future functionalities such as specialised task‑specific agents, an expanded Search API and deeper re‑search chat, illustrating a transparent, low‑impact approach to AI services.