Artificial intelligence today faces a paradox that defines a new era of technology. On the one hand, it promises to optimise resources; on the other, it is becoming one of the most energy-intensive sectors of the economy.
The ICT sector already accounts for 5-9% of global power consumption, and data centres, driving the AI arms race, could consume up to 8% of global energy by 2030. In the US, the scale of the problem is forcing radical steps: from the reactivation of old oil-fired power plants to the construction of private power grids by tech giants.
It becomes crucial to distinguish between two concepts: green AI and AI for Green. The first is about technological hygiene itself – designing algorithms to be energy efficient (which can reduce energy consumption by up to 90%) and conscious auditing of infrastructure. The second is about using AI as a tool to fix the world.
Examples are coming from the Polish startup ecosystem supported by PARP. Systems such as the Planter System from Agri 4 Zero optimise precision farming by reducing fertiliser and water consumption, while Waterly’s autonomous buoys monitor the quality of water bodies in real time.
The foundation for this transformation is data from IoT sensors. The use of energy harvesting technology, which derives energy from machine vibrations or temperature differences, allows building management systems (HVAC) to be implemented without generating additional operating costs.
At the scale of a large enterprise, optimising temperature and lighting with AI translates into savings of 20-30%, which means hundreds of thousands per year and real support for CSRD reporting.
However, the real challenge remains to move away from ‘AI fashion’ to substantive implementation. Krzysztof Gulda, President of the Polish Agency for Enterprise Development (PARP), emphasises that efficiency must not overshadow responsibility. There are a number of support programmes available to companies, such as ‘GOZ – it pays‘ or ‘Green Recommendations‘, which fund consultancy on the circular economy.
