A new assessment by the Basel Action Network warns that the rapid expansion of artificial intelligence could generate between 395 million and 617 million tonnes of electronic waste from 2025 to 2050 raising concerns over hazardous waste data security and unsafe recycling practices.
According to the assessment AI related electronic equipment could generate between 8.6 million and 13.1 million tonnes of waste annually by 2030. The estimate covers not only servers and graphics processing units (GPUs) but also the wider infrastructure of AI data centres, including networking equipment, power systems, cooling facilities and backup power infrastructure.
The report highlights that servers and GPUs account for only around 13 per cent of AI data centre equipment by weight. The remaining share consists of supporting infrastructure that is also likely to become waste as facilities expand and technologies evolve.
Rapid technological advances could shorten the useful life of electronic equipment, as companies replace older systems with newer, more powerful hardware. Even functional equipment may be discarded when it can no longer meet the demands of advanced AI applications, potentially increasing the volume of electronic waste.
The growing waste stream also presents environmental and public health risks. Electronic waste can contain hazardous substances such as lead and mercury, which may contaminate soil and water and harm human health if equipment is improperly dismantled or recycled.
Data security is another major concern. Retired servers, storage devices and networking systems may contain confidential corporate information, login credentials and sensitive technical details. The assessment stresses the importance of secure handling, proper data erasure and maintaining records throughout the equipment disposal and recycling process.
The international movement of discarded electronics could further complicate the issue. If obsolete AI equipment is exported to countries with weak recycling regulations or inadequate oversight, it could increase environmental pollution, expose workers to hazardous materials and create additional data security risks.
India could also face growing challenges as AI data centres and related infrastructure expand. Ensuring that retired servers, GPUs and other electronic components are reused, repaired or safely recycled will be crucial to preventing hazardous waste from entering informal disposal channels.
The assessment calls attention to the need to examine AI’s environmental footprint beyond electricity and water consumption. The entire life cycle of AI equipment from raw material extraction and manufacturing to operation, replacement, reuse and final disposal must be considered.
With hundreds of millions of tonnes of AI related equipment potentially reaching the end of its service life by 2050 effective recycling systems, stronger waste management safeguards and secure equipment disposal practices will be essential to reducing the environmental and social costs of the global AI boom.