Profit from Selling Chips, Pressure from Using Chips: JPMorgan Warns AI Stock Divergence Could Be a Key Turning Point for US Stocks in the Coming Weeks

AI concept stocks have shown clear divergence this year, with chip, optical communication, and AI infrastructure supplier stocks continuing to rise, while hyperscale cloud service providers (Hyperscalers) investing heavily in AI infrastructure have shown weak performance. JPMorgan (JPM-US) strategy team warns that current market trends are highly similar to the late 1990s internet bubble. Price changes in the coming weeks will determine whether this AI rally is just a healthy rotation or a precursor to a larger correction. Alphabet (GOOGL-US) and Tesla (TSLA-US) recently released their financial reports, both announcing further increases in AI-related capital expenditures. However, market reaction was lukewarm, and stock prices faced pressure, once again highlighting the core logic of AI investment this year. Market funds continue to flow towards the upstream of the AI industry chain, namely suppliers of GPUs, memory, optical communications, and data center equipment, the 'sellers.' Meanwhile, tech giants that purchase large amounts of AI computing resources and invest heavily in capital expenditures, the 'users,' face pressure from declining profitability and free cash flow, with investors adopting a more conservative attitude. Data shows that the Roundhill Magnificent Seven ETF (MAGS-US), which tracks the seven largest US tech stocks, has only risen about 1.5% this year. In contrast, the Philadelphia Semiconductor Index (SOX) has surged over 70% in the same period, reflecting the concentration of funds in the AI hardware supply chain. JPMorgan market technical analysis strategist Jason Hunter points out that if hyperscale cloud service provider stocks cannot break through important technical pressure zones, and semiconductor stocks continue to be constrained by key resistance levels, what was originally just a rotation of funds within the AI theme could evolve into a larger-scale correction for the entire market. JPMorgan Asset Management market and investment strategy chairman Michael Cembalest, in his latest podcast, uses the late 1990s internet bubble as a comparison. He points out that when the internet's heyday was nearing its end, communication service company stocks peaked first, but companies providing infrastructure continued to set new highs. The market mistakenly believed the bull run would continue, but ultimately faced a full correction. He believes that the current AI market is showing extremely similar phenomena. Hyperscale cloud service providers, due to massive AI investments, have continuously worsening free cash flow and stagnant stock performance; in contrast, chip, AI infrastructure, and optical communication supplier stocks continue to set new highs. Cembalest describes it as: 'Normally, the train should have the locomotive running faster than the carriages, but now the carriages are running ahead of the locomotive.' Hunter points out that since early July, the Philadelphia Semiconductor Index has fallen 20% from its high, while hyperscale cloud service providers have started a rebound within the large-scale consolidation zone in 2026, leading to different interpretations of the subsequent trend. Optimists believe that this is just a healthy rotation of AI concept stocks, with funds flowing from overvalued semiconductor stocks to large-cap tech stocks, helping to extend the AI bull market and make the AI investment theme more sustainable in the coming months or even quarters. Pessimists, on the other hand, believe that the current sector convergence trend is highly similar to the second quarter of 2000 before the internet bubble burst. At that time, there was also a reallocation of funds, but it ultimately led to a market-wide peak and decline. Hunter points out that from a technical analysis perspective, both scenarios are possible. Therefore, price changes in the coming weeks will be crucial in determining the market direction. Regarding semiconductor stocks, Hunter believes that the Philadelphia Semiconductor Index must break through the short-term pressure zone of 12,769 points to 13,333 points to confirm the continuation of the bullish trend. If it fails to break through, the index may retreat to the support zone of 9,975 points to 10,554 points, which would mean a correction range of approximately 28% to 32% compared to the high in June. However, he also points out that if the index retreats to the above support zone, it could be seen as a trading opportunity to buy. Hunter also proposes important technical thresholds for several large-cap tech stocks. Alphabet needs to break through the 50-day moving average of 368 dollars and the pressure level of 381 dollars, with the stock currently trading around 328 dollars pre-market on Thursday, still some distance away from the target. Amazon (AMZN-US) needs to break through the pressure zone of 251 to 258 dollars to confirm the continuation of a larger bullish trend, with the next target price being 278 dollars. Meta Platforms (META-US) has recently rebounded from the 525 to 553 dollar range but is currently constrained by the 669 to 694 dollar pressure band. If it can effectively break through 694 dollars, it will be seen as a strong bullish signal. As for Microsoft (MSFT), Hunter points out that 350 dollars is an important support level, and there is still some distance to the mid-term pressure zone of 465 dollars to 493 dollars. The stock price still needs to sequentially break through the 50-day moving average of 400 dollars, the 61.8% retracement position of 421 dollars since May, and the 200-day moving average of 438 dollars to have the opportunity to reconfirm the long-term bullish trend.