MAIN STORY
A small segment with an outsized warning signal
Neocloud companies lease powerful computing equipment, often built around AI accelerators, to developers and technology groups. Their rapid construction of data centres gives customers access to capacity without having to own every server. But the business model can involve substantial upfront borrowing, with repayment dependent on utilisation rates, financing costs and the durability of customer contracts.
At the Singapore conference, Kim Yong-kwan described this segment as approximately 8% of the overall market, according to Reuters. That estimate is Samsung's characterization, not proof that the wider system is immune. A comparatively small borrower can still matter when its purchasing and financing links touch larger chipmakers, lenders and customers. The key distinction is between vulnerability in a group of operators and weakness in total demand for AI computing.
Why Samsung sees durable demand for memory
Samsung's position is rooted in the supply chain. Advanced AI systems need processors, high-bandwidth memory and storage, while inference services generate recurring demand after initial model training.
















