Business · Companies · 3 hrs ago

AI compute shortages put SCX.ai’s inference strategy in focus

AI compute shortages put SCX.ai’s inference strategy in focus

Apollo Global Management says demand for AI computing is outpacing supply across areas including chips, memory and electricity.

The shortage matters because AI agents can use far more computing power than conventional chatbots, with some tasks consuming 100 to 1,000 times more tokens.

Tokens are the units of information AI models process and generate, and inference is the computing that lets trained models produce answers and complete tasks.

Australian-listed SCX.ai is focusing on AI inference, rather than the large computer clusters often used to train models.

It uses inference accelerators developed by US company SambaNova Systems in Australian data centres, and its first node began operating at Equinix’s SY5 centre in Sydney in March 2026.

SCX.ai raised $40 million in an initial public offering before listing on the ASX in August 2026.

The company plans to expand capacity as demand for AI inference grows.

1 source

Related