Decarbonising compute: from the ground up

Getting the right compute architecture is critical.

Between 2022 and 2030, data traffic is projected to rise sixfold. One of the biggest implications is the need for a corresponding increase in compute resources. As 5G network deployments increase in volume and scale globally, many operators are also investing heavily in network automation to handle increasingly complex network management workloads. AI in the network is set to drive a wave of demand for computing resources. This will manifest across the board – from on-device to edge to central clouds. Increased compute requires improvements in semiconductor designs and manufacturing processes to keep pace with surging digital traffic.

While significant advances have been made in this field, there are constraints in terms of the supply and cost of energy. New chipset designs must be made with sustainability as a bedrock principle. It is not sufficient to only make improvements to existing processes to drive energy efficiencies. Rather, it must be done from the ground up – and this starts with the process of decarbonising compute.

This report examines the rationale and urgency behind the sustainability push, how compute efficiencies translate into energy reductions, and how competitive activity is set to accelerate among chipset designers and producers to improve power efficiency.