The sell-off in artificial-intelligence shares spread through Asia on Thursday 15 October, with South Korea’s Kospi losing 7.6% and Japan’s Nikkei 225 falling 4.9%, according to Associated Press. Those are the day’s Asian session results, following Wednesday’s losses on Wall Street.
The pressure is about more than whether people will use AI. It is also about the price of financing the computers, chips and data centres needed to supply it. Reuters reported on Thursday that investors were reassessing crowded AI investments against high borrowing costs. The Nasdaq Composite had fallen 1.9% in the completed US session on Wednesday 14 October.
The immediate backdrop was US inflation. The Bureau of Labor Statistics said on Wednesday that consumer prices rose 3.7% over the year to September, up from 3.4% in August. September’s monthly increase was 0.6%. Reuters linked the market response to renewed expectations of higher interest rates.
Why does that matter to AI? There are two channels. First, businesses borrowing to buy equipment or build facilities must cover interest payments before shareholders receive the remaining profits. If new finance becomes more expensive, or existing debt must be refinanced at a higher rate, a project needs more cash to justify its cost.
Second, interest rates affect what investors will pay today for profits expected years from now. A Federal Reserve research note explains how share valuations depend on both expected earnings and the rate used to translate future earnings into today’s value. A higher required return can reduce that value even if the promised technology still works.
The financing question is particularly visible around data centres. Reuters’ Thursday morning analysis identified debt concerns around operator CoreWeave as a focus of the selling. That adds a timing question: can revenue from rented computing power arrive quickly enough to meet the commitments made to build the capacity?
These pressures do not establish that AI demand has disappeared, that every company faces the same risks, or that a bubble has definitively burst. Businesses with different debts, customers and cash reserves will face different tests. A falling share price is a market judgement about risk, rather than proof of a future default.
Our reading: watch the cash as closely as the computing power. The next test is whether businesses can turn AI demand into durable earnings while paying for the infrastructure behind it.
Archive context photographs: Wikimedia servers, 21 July 2015, Victor Grigas / Wikimedia Foundation, CC BY-SA 3.0; Tokyo Stock Exchange, October 2018, Kakidai / Wikimedia Commons, CC BY-SA 4.0. Neither image depicts today’s trading session. Photographs may be cropped; their licences apply to adapted photographs.
