/Cloud%20Computing%20diagram%20Network%20Data%20Storage%20Technology%20Service%20by%20onephoto%20via%20Shutterstock.jpg)
The unprecedented demand for global data centers isn’t slowing down anytime soon, according to Global X, which now sees annual expansion of 14% all the way until 2030. This translates to an additional 100 GW of capacity, twice the current data center power load across the world.
A quarter of the existing demand is represented by AI workloads. Within this, the share of inference tasks (the ones that use AI-trained models to give real-time outputs like chatbots) has, for the first time, surpassed the share of AI training tasks (the ones that require compute to train AI models). That's not all; companies are now prioritizing key geographical locations for data center setups, not only to improve energy efficiency but also to reduce latency, which is instrumental in providing a real-time AI experience, such as through AI smart glasses or AI assistants.
