SambaNova at World Economic Forum Annual Meeting 2025

AI Inference Will Drive Power Demand Growth

While 70.0% of leaders recognize the energy-intensive nature of training large language models, only 59.7% are aware of the significant power demands of inference. This gap is critical as inference workloads are set to dominate AI usage with the scaling of Agentic AI.

Energy Efficiency is a Strategic Priority

Only 13.0% of organizations currently monitor AI power consumption, yet 56.5% acknowledge that energy efficiency will play a crucial role in future strategic planning. The need to address rising energy demands is being driven by both cost management imperatives and operational scalability concerns.

Scaling Agentic AI Brings New Challenges

The roll-out of Agentic AI is amplifying power concerns for enterprises. For 20.3% of companies, rising power costs are a pressing issue. Furthermore, 37.2% are experiencing increasing stakeholder pressure to improve AI energy efficiency, with a further 42.0% expecting these demands to emerge soon.

Few Enterprises Are Proactively Addressing AI’s Energy Impact

Among organizations that have widely deployed AI2, 77.4% are actively seeking to reduce power usage. Popular approaches include hardware and software optimization (40.4%), adopting energy-efficient processors (39.3%), and investing in renewable energy (34.9%). However, these measures remain insufficient compared to the rapid pace of AI adoption and scaling.