AI firm SCX sets market listing
SCX.ai is set to begin trading on the Australian Securities Exchange on Friday under the ticker SCX after a fully underwritten $40 million initial public offering.
IPO draws strong backing amid growing AI compute demand
The flotation attracted capital from several well‑known Australian fund managers, including Wilson Asset Management, Frazis Capital Partners and Ellerston Capital. Institutional interest reflects a broader trend of investors seeking exposure to the infrastructure that underpins rapid AI adoption.
Demand for artificial‑intelligence processing is outpacing the capacity of existing power, water and grid resources. In the United States, more than 500 local bans or restrictions on data‑centre expansion were reported by early August, up from roughly 300 in late June, according to the report.
Australia is responding with new standards that require data centres to support additional electricity generation while minimizing water use. Those constraints are creating a premium on technologies that can deliver more AI compute from the same physical footprint.
Related: Killi set for Lodestone drilling excitement
Purpose‑built inference hardware aims to lift efficiency
SCX.ai’s business model centers on installing purpose‑built inference processors inside commercial data centres already operating in the country. The firm uses reconfigurable dataflow units (RDUs) developed by US semiconductor firm SambaNova Systems. The vendor states these units can generate AI responses up to five times faster than conventional GPU‑based systems while consuming about 75 percent less electricity and 99 percent less cooling water.
RDUs differ from GPUs, which were originally designed for graphics rendering and later adapted for both AI training and inference. The architecture of the RDUs is tailored specifically for inference—the stage where a trained model produces an answer, analyses a document or runs an AI‑driven task.
Founder and chief executive David Keane said AI demand is growing at extraordinary speed and the infrastructure has to keep up. Purpose‑built inference technology allows more AI compute to be generated from the power and physical footprint already available in Australia’s commercial data centres.
As AI becomes embedded across businesses and government agencies, he expects inference to account for an increasingly large share of overall AI computing demand.
Related: Mount Ridley Study Backs Flotation Theory at Grass Patch
Comparing this approach to earlier attempts at scaling AI capacity, the shift toward inference‑only hardware mirrors a broader industry move away from generic GPU farms toward specialized chips that handle specific workloads more efficiently. That pattern has repeated in other sectors, where task‑specific processors have often delivered cost and energy savings over more versatile but less optimized hardware.
Cost considerations and data sovereignty
Running AI at scale is becoming a focal point for corporate budgeting. The Australian design platform Canva recently lowered its revenue growth outlook after discovering that the cost of deploying new AI products on frontier models was higher than expected. The company slowed the rollout of its Canva AI 2.0 suite while seeking to reduce unit costs and bring more of the technology in‑house.
SCX.ai’s platform is positioned to address the same cost pressures by offering on‑shore inference capability. Enterprises handling sensitive financial, health or government data can keep processing within Australia, reducing concerns about cross‑border data transfer.
