AI Stocks Struggle to Rebound: Lacking Three Major Supports, July’s Rally Hard to Replicate

According to Woofun AI, JPMorgan (JPM.US) holds a pessimistic outlook regarding the strength of the AI stock rebound in October, presenting a stark contrast to its sentiment in July. Although stock prices have already pulled back, the path for market recovery is unlikely to replicate the previous swift and robust rally. Strategist Nikolaos Panigirtzoglou emphasized that the willingness of investors to “buy the dip” has weakened, and the combination of reduced corporate buybacks and insufficient short positions means that the current downturn lacks the underlying momentum to convert into a strong rebound.

A deeper analysis of capital flows and position structures reveals fundamental differences between this correction and that in July. Woofun AI’s collated data shows that, despite shares of Korean memory chip giants SK Hynix (SKHY.US) and Samsung Electronics having dropped by nearly one-third over the past month, these stocks have yet to display clear signs of being oversold, and the declines themselves do not constitute a sufficient basis for a rebound. Market concerns are centered around the possibility that the boom in AI capital expenditure may slow, as well as fears that lower-priced Chinese competitors could erode growth and profit margins. The recent sell-off has spread throughout the entire supply chain, with the Philadelphia Semiconductor Index dropping by about 10% over the past five trading days—Intel (INTC.US) experienced a similar drop during this period. Unlike July, ETF investors have not stepped in to “buy the dip” via leveraged and non-leveraged ETFs during the October correction; such buying has been rare. The funds that previously helped the market absorb selling pressures have yet to make an obvious return. The difference in short positions has further weakened potential support. In July, short positions in the largest US-listed Korean equity ETF—the iShares MSCI Korea ETF (EWY.US)—increased significantly, whereas this month, short positioning in this fund has seen little change.

At the same time, short positions in products like VanEck Semiconductor ETF (SMH.US) and Roundhill Memory Chip ETF (DRAM.US) also remain at low levels. As buying driven by short covering can at times fuel rebounds, this current low level means limited potential for such rebound-driven buying.

The most crucial variable is the reaccumulation of leverage risk. Panigirtzoglou focuses on the ratio of leveraged memory chip ETF assets under management to the total market cap of underlying equities. After hitting a low at the end of July, this ratio rebounded in August and September.

This indicates that some of the leverage risk unwound during the previous correction has now returned to the market, potentially amplifying volatility. The ongoing unprofitability of Anthropic and OpenAI, combined with the high interest rate environment, has prompted investors to re-evaluate the growth prospects for AI infrastructure expansion, resulting in an absence of “bottom-fishing” capital.

Changes in corporate behavior represent another key distinction. Following strong profits in the second quarter, SK Hynix and Samsung Electronics announced a combined $40 billion stock buyback program. The execution of these plans was faster than expected and had already been completed ahead of schedule by mid-October. For the market, buybacks mean companies are using their own funds to repurchase shares, providing relatively steady demand. When investors are reducing positions and market sentiment is weakening, such buying can absorb some of the selling pressure and help mitigate declines in share prices.

However, the early completion of buybacks means that this demand exited the market earlier than anticipated. The selling pressure previously absorbed by corporate funds must now be taken on by other investors. Even if the companies’ operating outlooks do not further deteriorate, this shift in buy and sell forces may make prices less likely to stabilize. At the end of July, the market did not face such a sharp drop-off in buyback funding. Therefore, although both corrections initially impacted Korean memory chip stocks, the funding conditions supporting subsequent recoveries now differ greatly. This also explains why JPMorgan emphasizes that even though stock prices have declined, position adjustment and capital flows have not changed enough to provide a foundation for a strong rebound.

Under multiple constraints, the recovery path for AI stocks has changed. In terms of buyer sources, ETF investors have not clearly entered the market to “buy the dip” as they did in July; low short interest implies limited demand from short covering; and the end of corporate buybacks has further reduced ongoing buying power.

Meanwhile, the relative size of leveraged memory chip ETFs rose again in August and September, so while market absorption capacity has weakened, leverage has increased volatility risk. All these factors combined make any rebound now more dependent on new funds entering the market. Falling stock prices can lower entry costs but cannot automatically generate buying interest. In the face of potentially slowing AI capital expenditures, intensified competition, and ongoing uncertainty around realizing profits in a high-rate environment, investors may need more evidence before increasing their positions again. Therefore, JPMorgan’s assessment mainly addresses the market’s recovery potential after this correction: The buying forces that drove the rapid recovery in July have now either weakened or failed to appear, while leverage risk has accumulated once more. This does not rule out short-term rebounds, but it does mean that the market cannot rely solely on previous experience to expect the same rapid and powerful recovery. According to this analytical framework, the following are worth watching: whether Korean and semiconductor-related ETFs can attract new capital, whether leveraged positions will be pared down, and whether companies will launch new buyback programs. Until these supports show improvement, this round of AI stock correction may take longer to re-establish a solid base of sustained buying interest.