The AI stock sell-off that began in late June has entered a second, deeper phase. On July 28, South Korea’s Kospi index fell 11.5% to its lowest level in three months, driven by double-digit drops in Samsung Electronics and SK Hynix — the two companies most exposed to the AI memory chip boom. US chip stocks extended losses when Wall Street opened, with the Nasdaq 100 briefly dipping into correction territory, more than 10% below its early June record high.
🔍 THE BOTTOM LINE
This is no longer a one-day blip or a single stock’s valuation correction. The sell-off is broad, global, and driven by a specific fear: that AI companies are borrowing too much to build datacenters that may not generate enough revenue to justify the cost. When the market reprices that risk, every link in the AI hardware chain gets hit at once.
What Happened on July 28
The Guardian reported that Samsung Electronics and SK Hynix both fell more than 10% on Tuesday, dragging the Kospi down with them. The sell-off spread to US markets when Wall Street opened: Intel, AMD, SanDisk, Western Digital, and Seagate Technology all dropped more than 4%. The Nasdaq 100 fell as much as 1.8% at one point, briefly crossing into correction territory — a 10% decline from its peak — before rebounding to roughly flat by the close.
The Financial Times described the move as part of a “deepening” AI sell-off, with investors rotating out of chip stocks ahead of Big Tech earnings reports. The concern is not that AI is failing. The concern is that the companies spending hundreds of billions on AI infrastructure are doing it with borrowed money, and the revenue to justify that spending has not materialised at the scale or speed the valuations require.
Why South Korea Is Ground Zero
South Korea’s semiconductor giants sit at the centre of the AI hardware supply chain. Samsung is the world’s largest memory chipmaker. SK Hynix is the dominant supplier of HBM (high-bandwidth memory), the specialised chips that pair with NVIDIA’s GPUs to train large AI models. When AI spending booms, these two companies are among the biggest beneficiaries. When the market questions whether that spending is sustainable, they are among the first hit.
Both companies have been covered in our reporting. Samsung’s 1,800% profit jump on AI chip demand was one of the standout financial results of the year. SK Hynix’s mega ADR listing was supposed to broaden its investor base beyond Seoul. Neither move has insulated the stocks from the broader repricing.
The Debt Question
The new factor in the July 28 sell-off is the focus on debt. The Guardian noted “rising concerns about the huge amount of borrowing among AI companies to fund their datacentre expansion plans.” This is the thread that connects the AI stock sell-off to the broader financial picture we have been tracking.
Oracle’s credit downgrade to BBB signalled that rating agencies see risk in AI infrastructure debt. IBM’s 25% share plunge showed what happens when the market loses confidence in the spending trajectory. Kioxia’s market value halving demonstrated that even record profits don’t protect a stock when the valuation was built on assumptions of perpetual growth.
The pattern is consistent: companies are borrowing to build, the market is questioning whether the build will pay for itself, and the stocks that rose fastest on the AI narrative are falling fastest on the doubt.
What Changed Since June
The first AI sell-off on June 23 was triggered by specific events: Alphabet researchers leaving, SpaceX’s bond offering spooking investors. The July 28 sell-off is different. There is no single catalyst. It is a gradual repricing that has accelerated as more data points accumulate — earnings misses, credit downgrades, delayed IPOs, and now the sheer scale of borrowing becoming visible across multiple companies.
The Nasdaq 100 briefly entering correction territory matters psychologically. A correction is not a crash, but it is the point where the narrative shifts from “buy the dip” to “is the dip the floor or the basement?” For AI stocks specifically, the question is whether this is a healthy valuation reset — the kind that makes the sector sustainable long-term — or the beginning of a more serious unwind.
NZ Angle
New Zealand investors are exposed to this sell-off through KiwiSaver funds, most of which hold international share portfolios dominated by US tech stocks. As we noted in our June coverage, seven tech companies now make up roughly 30% of the S&P 500. When those stocks fall, diversified funds fall with them. The deeper concern for NZ is on the infrastructure side: if global AI spending cools, the economics of sovereign AI compute change. The case for NZ-owned AI infrastructure should rest on sovereignty and data control, not on assumptions about perpetual demand growth that the market is now questioning.
❓ FAQ
Is this the same sell-off as June? It is the same trend, but a different phase. The June sell-off was triggered by specific events. The July sell-off is a broader repricing driven by accumulated concerns about debt-fueled AI infrastructure spending.
Why are Samsung and SK Hynix falling so hard? They are the most exposed companies in the AI memory chip supply chain. When investors question whether AI datacenter spending will continue at current levels, the companies that supply the memory chips are among the first to be repriced.
Does a Nasdaq correction mean AI is overhyped? Not necessarily. A correction means the index fell 10% from its peak. The AI technology is real and advancing. The question the market is asking is whether the valuations assigned to AI-exposed companies were justified by the revenue those companies can actually generate.
What does this mean for KiwiSaver funds? Most diversified KiwiSaver funds hold US tech stocks. When the Nasdaq falls, those funds feel it. Investors concerned about concentration risk should check what their fund actually holds — a “diversified” fund that is 30% US tech is less diversified than the label suggests.
Could this trigger a broader market crash? That is a question for financial analysts, not an AI news site. What can be said is that AI stocks have been the primary driver of market gains for two years. If they lead the market down, the broader index follows. But corrections are normal, and the market rebounded to roughly flat by the close on July 28.
🔍 THE BOTTOM LINE
The AI sell-off has gone from a June warning shot to a July global event. South Korea’s market is at a three-month low. The Nasdaq touched correction territory. The driver is not technology failure — it is financial structure. Companies borrowed heavily to build AI infrastructure, and the market is now asking whether that borrowing will be justified by revenue. The answer to that question will determine whether this is a healthy reset or something deeper.
