Updated July 29, 2026.
The AI stock selloff spread from US semiconductors to South Korea on July 29, yet the broader tape still looks more like a rotation than a wholesale market retreat. According to Associated Press on July 29, the Kospi fell 6% after plunging 10.8% in the previous session, while SK Hynix lost 9.6%.
On Wall Street, Nvidia dropped 2.6%, AMD fell 4.4% and Micron declined 3.7%. The key message is not simply that major indexes weakened. It is that heavily weighted AI winners were being sold while a wider group of companies continued to attract capital.
AI stock selloff: the rotation in numbers
| Market or stock | Move | What it signals |
|---|---|---|
| Kospi, July 29 | -6% | Second consecutive session of heavy selling |
| Kospi, July 28 | -10.8% | Losses concentrated in chip leaders |
| SK Hynix | -9.6% | Pressure on the AI memory supply chain |
| Nvidia | -2.6% | Large drag because of its index weight |
| AMD | -4.4% | Broad semiconductor weakness |
| Micron | -3.7% | Memory, pricing and investment concerns |
| Sherwin-Williams, July 28 | +8.3% | Reward for stronger-than-expected earnings |
| Illinois Tool Works, July 28 | +3.6% | Demand for solid results outside technology |
These are AP’s reported moves for July 28 and 29. They do not describe uniform risk aversion. Instead, they show investors reducing exposure to the most crowded winners and rewarding companies with better earnings, less demanding valuations or different economic sensitivities.
Market breadth improved beneath weak technology
Most US stocks rose on July 28 even as chipmakers continued to weaken. AP reported that the S&P 500 gained 0.2% and the Dow rose 1%, while the Nasdaq Composite slipped 0.2%. Sherwin-Williams rallied 8.3% and Illinois Tool Works advanced 3.6% after both delivered earnings above expectations.
That divergence is market breadth: the number and variety of stocks participating in a move. An index can look weak because a handful of giant constituents are falling even when the median stock is rising. Investors therefore need to separate index performance from participation below the surface.
CryptoRoad’s guide to inflation, interest rates and markets explains why no session should be reduced to one macro variable. Earnings expectations, sector weights and positioning can dominate the short-term result.
Concentration risk in the Nasdaq and Kospi
Concentration works in both directions. When Nvidia and other AI leaders rally, their market value can lift capitalization-weighted indexes despite narrow participation. When those names retreat together, the same arithmetic can make the Nasdaq look much weaker than the average US stock.
The effect is even clearer in South Korea, where major semiconductor producers command an unusually large share of the Kospi. A 6% decline following a 10.8% plunge signals severe pressure and possible technical selling, but it does not mean every Korean company suffered the same fundamental shock.
The portfolio lesson is straightforward. A diversified index by company count can still contain a substantial sector bet by market capitalization. Investors exposed through passive products may own more AI and semiconductor risk than the number of holdings suggests.
Valuation and capex are being repriced
The AI stock selloff is testing two assumptions. The first is that revenue and margins can expand fast enough to justify premium multiples. The second is that huge spending on data centers, GPUs, high-bandwidth memory and power infrastructure will generate adequate returns on capital.
AI demand does not need to disappear for valuations to fall. Investors only need to lower expected growth, apply a higher discount rate or demand firmer evidence of monetization. Excellent companies can then suffer sharp drawdowns because their share prices had already discounted near-perfect execution.
Monetary policy remains part of the background. Fed decisions affect the rate used to value long-duration earnings and the general cost of financing. CryptoRoad’s overview of central banks, the ECB and the Fed provides the wider framework.
Rotation is not proof the AI cycle is over
Two violent sessions do not prove the technology cycle has ended. A fundamental break would require persistent evidence: broad cuts to hyperscaler budgets, weaker orders, structural pressure on memory pricing, unused capacity or sustained downward revisions to sector earnings.
Until that evidence emerges, rotation and repricing are more precise descriptions. Capital is moving away from companies with the most stretched expectations and toward businesses delivering earnings surprises. The gains in Sherwin-Williams and Illinois Tool Works illustrate that mechanism, but do not guarantee the shift will continue smoothly.
The caveat cuts both ways. Labeling every correction a rotation can conceal genuine deterioration. Upcoming results, management capex guidance and semiconductor margins will provide a better test than any single closing price.
What it means for Nasdaq risk appetite and Bitcoin
For the Nasdaq, broader participation can coexist with elevated volatility. If the biggest constituents keep losing ground, the index may remain under pressure. If capital continues moving into industrials, cyclicals and profitable laggards, the overall US market may prove more resilient than the technology benchmark.
Bitcoin belongs in this discussion as an asset sensitive to liquidity and risk appetite, but there is no mechanical rule that a fall in Nvidia must cause BTC to decline. Broad deleveraging could hit both markets. A rotation contained within equities could leave crypto trading on its own catalysts.
Correlations vary by regime, as CryptoRoad’s analysis of gold, Bitcoin and stock correlations details. The dollar, real yields and financial conditions may matter more than one chip session; the strong and weak dollar framework adds another useful lens.
Signals to watch after the selloff
A constructive outcome would combine continued breadth with stabilizing semiconductor shares and earnings that support current investment plans. A more fragile outcome would involve contagion from chips to credit, small caps and other volatile assets, alongside rising volatility and shrinking liquidity.
The July 28-29 AI stock selloff is therefore a test of accumulated concentration, not a final verdict on artificial intelligence. Company results, capex discipline and market breadth will determine whether this becomes a healthy redistribution of leadership or a deeper reduction in growth expectations.
