When people talk about AI, most of the attention still goes to chips, models, and technology companies.
But the real AI boom will increasingly take place beyond our screens.
Australian data centers could increase their annual electricity consumption from roughly 5 TWh to 34 TWh. That is nearly a sevenfold increase.
That number shows just how physical this story really is.
A data center is not simply a building filled with servers. It needs reliable electricity, new grid connections, transformers, switchgear, backup systems, cooling, and a vast amount of additional electrical equipment.
Chips may be the most visible part of the AI boom. But without sufficient energy infrastructure, their computing power cannot be fully utilized.
AI is increasingly becoming a story about power grids as well.
That could create pressure to build new generation capacity, strengthen transmission systems, and modernize infrastructure that, in many countries, was never designed for such rapid growth in electricity demand.
Silver is used in electrical contacts, power electronics, control systems, and other equipment needed to transmit and reliably manage electricity. That does not mean every additional TWh will generate a precisely predictable amount of silver demand. Technology is evolving, manufacturers are reducing material use, and some applications may shift toward alternative metals.
But the scale of the projected growth deserves attention.
And not only in Australia. A similar competition for electricity, grid capacity, and available transformers could gradually emerge wherever AI data centers are being built at scale.
The market is currently focused primarily on the companies designing the chips.
I would also watch who supplies the electricity, how quickly the grids can be expanded — and how much physical material this infrastructure will ultimately require.
Leave a Reply