Thursday’s slide in European and Korean chip stocks looks, on the surface, like a rates story. US inflation came in hotter, talk of Federal Reserve tightening grew louder, and technology valuations fell. For anyone building or buying AI, the more useful reading is different. The sector’s central question is shifting from “is there demand?” to “who pays for it, and at what cost?”
Rates matter because AI is capital-hungry. Higher borrowing costs shrink the present value of the distant earnings that justify tech valuations, and they make the debt behind data centres and processors dearer. September inflation was reported at 3.7%, driven by energy, and markets are pricing a tougher policy path. No new Fed decision has actually been announced.
The deeper shift is from equity to credit. For two years AI was told as a stock-market story, but data-centre-related debt issuance roughly doubled to around $180 billion in 2025. This year the pace has accelerated: investment-grade bonds from hyperscalers, data-centre developers and other AI-focused firms reached $218 billion by July 8, against $80.5 billion in all of 2025. Morningstar flagged signs of buy-side indigestion, and a Goldman Sachs trader was reported as saying AI bond supply had exhausted the market’s capacity to digest it. Reuters’ reporting on bondholder wariness toward heavily indebted AI builders fits the same picture.
Korea shows how fragile the chain can be. Samsung Electronics and SK Hynix together represent roughly half of the Kospi’s market capitalisation. The index was already in its fifth straight losing session on October 13, before the CPI release, as AI worries, chip competition and inflation fears combined. One trigger was Chinese memory maker CXMT saying it would introduce DDR5 server modules on a new DRAM cell architecture by year-end, though it has yet to demonstrate mass production. Leverage adds fuel: in July, the unwinding of heavily leveraged retail positions intensified a similar rout. Memory makers therefore carry three overlapping risks: customer financing, competition and speculative froth. For the AI industry itself, four implications stand out.
Chip suppliers depend on customers who can only keep buying if their AI products earn revenue. Expect investors to demand evidence of returns, staged builds and clearer financing structures. If credit tightens, cash-rich hyperscalers can keep expanding. Smaller labs and start-ups that rely on project debt or private credit may face delays or costlier capacity. One trade headline reported that Blue Owl was struggling to line up debt for a $4 billion CoreWeave data centre.
Memory, especially high-bandwidth memory, is a bottleneck. Models and inference methods that do more with less hardware gain value when capital is expensive. If Chinese memory capacity proves out, it would pressure prices and incumbents’ margins. That would hurt the suppliers but benefit AI developers.
Keep perspective, though. Nothing here shows chip orders reversing. The sector is weighing strong current demand against doubts over the long-term profitability of the capacity now under construction, and the same selloff is hitting companies at very different points in the production chain.
The practical lesson for business leaders is to treat AI capacity as a financed asset, not a free-flowing utility. Contracts, supplier diversity and realistic payback timelines will matter more as money gets pricier. The signals to watch are the Fed’s next move, hyperscaler capex guidance, corporate bond spreads and memory pricing.
Written by:
*Chloe Maluleke
Associate at BRICS+ Consulting Group
Russia & Middle East Specialist
**The Views expressed do not necessarily reflect the views of Independent Media or IOL.
** MORE ARTICLES ON OUR WEBSITE https://bricscg.com/ (https://bricscg.com/)
** Follow @ (https://x.com/brics_daily)brics_daily (https://x.com/brics_daily)on Twitter for daily BRICS+ updates and instagram @brics_daily (https://www.instagram.com/brics_daily?igsh=bmhvbTd0YzA4a2wx)
