After the Philadelphia Semiconductor Index and Nasdaq plunged, Morgan Stanley called to "get ready to increase positions"

According to Zhitong Finance APP, top strategists from Wall Street financial giants such as Morgan Stanley believe that the logic of increasing positions and buying the dip in global AI computing power themes and broadly-defined technology stocks, which have recently suffered heavy sell-offs, is taking shape. However, whether tech stocks can achieve a bottom rebound and a sustained upward trend afterwards has entered a dual verification period: earnings season and AI financing conditions. Morgan Stanley's chief equity strategist Michael Wilson stated in his latest research report that, after months of valuation adjustments across most stocks, the correction is approaching its end. The market may complete this round of adjustment quickly, or it may decline slowly until November. Investors should be prepared to increase their exposure to stock market risk during this process.

Wilson stated that he still prefers companies with strong fundamentals, especially sectors such as capital goods where profit momentum continues to improve significantly. Therefore, this is a new judgment on the investment opportunities in the equity market after a significant downturn and adjustment. This view is largely consistent with the bullish stance of Bank of America—after a massive correction, valuations for most tech giants, including storage chip leaders and AI computing power theme champions, have approached historical cycle lows. Core chip prices related to AI infrastructure, rallies, orders, and earnings trends have not yet shown significant deterioration, suggesting that buy-the-dip opportunities after the pullback seem to have arrived.

Currently, new cloud leaders like CoreWeave reveal the most sensitive financing weaknesses in this adjustment. CoreWeave fell 25.21% last week, expanding its YTD loss to 14.27%; its 2031 bonds’ yield once reached about 16% last week, with 5-year CDS spreads topping 1,100 basis points. The implied cumulative 5-year default probability in related markets exceeds 60%. This probability is derived from market models that include risk premiums and are affected by recovery assumptions. Nvidia-backed Nscale also saw its IPO priced down—some investors are pushing for a $16B–$19B valuation versus the $30B target, a discount of about 37%–47%, though pricing is not final.

Thus, the market is now reassessing large expenditures on core AI infrastructure hardware components such as GPU/TPU/ASIC, data center CPUs, large server room construction, and related electricity costs, which occur upfront, while revenue lags until power-up, acceptance, and customer adoption. The more expensive financing becomes, the less able projects are to withstand delays or underutilization.

Just as storage chip sector stocks, which benefited from the unprecedented AI infrastructure boom, significantly declined recently, Bank of America analysts believe the market has already priced in a slowdown in future price increases, supply expansion, and AI financing pressures. However, spot price resilience, HBM iteration, and surging tiered storage demand driven by the AI agent inference boom—as well as overall explosive AI computing power resource needs—can still support strong profit expansion for AI computing theme stocks over a long period.

Notably, during the recent global semiconductor sector plunge, institutions have dramatically reduced directional exposure. However, leveraged ETFs favored by retail investors for the long term—such as the triple-leveraged SOXL ETF tracking the Philadelphia Semiconductor Index—saw enormous inflows. This potential demand for short covering combined with leveraged capital has jointly amplified market sensitivity to strong earnings season catalysts.

The latest bullish views from Morgan Stanley and Bank of America echo the latest market trends—technology stocks have already shown signs of breathing, but the wounds left by the recent sell-off remain significant: last week, the Nasdaq 100 fell by about 5.5%; the Philadelphia Semiconductor Index dropped for eight straight sessions, falling about 11.5% on the week—entering technical correction territory. As of October 16, Korea’s KOSPI Index has fallen 17.11% this month, with Samsung Electronics and SK Hynix dropping 19.74% and 28.38%, respectively. By the close on October 19 in Asia, KOSPI rebounded 1.76%, but Samsung and Hynix rose only 1.16% and 0.39%, respectively. In U.S. pre-market trade, Nasdaq futures rose about 1%, indicating tentative buying is returning, and storage chip leaders seem to be regaining leading AI momentum after several days of lagging.

Morgan Stanley’s Wilson calls “buy the dip”: Correction nears its end, prepare to increase risk exposure

Wilson is one of Wall Street’s most influential sell-side strategists: he joined Morgan Stanley in 1989, held positions in investment banking, institutional equities, and investment strategy, and is currently Chief U.S. Equity Strategist and Chief Investment Officer. He was named first place in Institutional Investor’s 2022 All-America Research Team for Portfolio Strategy, far ahead of other famous strategists.

His influence on the equity market comes mainly from integrating earnings, valuations, interest rates, and institutional positioning into allocation decisions. When he issues a positive “prepare to buy” signal, it typically forces large institutional investors to reassess the opportunity cost of being overly defensive.

Michael Wilson of Morgan Stanley stated that as the multi-month correction in segments of the tech-heavy stock market nears its end, investors should be ready to raise their equity risk exposure in their portfolios.

Although the S&P 500 is only about 2% below its all-time high, market valuations have already declined significantly over the past few months, while internal market breadth within the index has also deteriorated sharply.

This strategist believes that the current market regime—characterized by outperformance of so-called quality factors and sharp declines in equal-weighted indexes—is about to end, creating a favorable window for investors to buy stocks on weakness.

In a client note, Wilson wrote: “We recommend that investors be ready, as the market may complete this adjustment very quickly, or it may decline slowly until November. In our opinion, risk exposure should be increased at that time, not avoided.”

This strategist insists on a preference for quality stocks and emphasizes that financials, industrials, and consumer discretionary—sectors more sensitive to the economy—have borne the brunt of this round of large valuation adjustments. He states that valuation cuts in those areas have already created entry opportunities in capital goods, technology, and other sectors where per-share earnings momentum is improving.

As shown in the chart above, U.S. equity valuations have declined significantly over the past few months

Wilson’s view may reassure some investors. These investors are concerned that the recent surge in U.S. Treasury and high-yield corporate bond yields could undermine the historically unprecedented AI investment boom, which relies heavily on new debt financing. Given that the AI infrastructure frenzy has played a vital role in driving the ongoing four-year global equity market bull run, some participants say they are already seeing bubble signals.

Goldman Sachs strategists also noted Monday that investor positioning is becoming increasingly defensive. Meanwhile, strategists from JP Morgan and Morgan Stanley’s Wilson remain optimistic. They note that bond yields stabilizing at their current levels would “help a lot” in supporting equities.

Broader confidence in equities prospects mainly comes from the increasingly robust profit outlook driven by the AI computing power theme. According to the latest compilation by Bloomberg Intelligence of consensus Wall Street analyst forecasts, overall Q3 net profits for S&P 500 constituents are expected to soar by 24.6% year-over-year.

The Wall Street financial giant Jefferies recently projected that—powered by the dual engines of the AI investment boom and AI-related companies outperforming with earnings—the S&P 500 could surge to 8,000 by the end of 2026, and reach 9,000 in 2027. Jefferies’ core logic is clear and strong: In a cycle where AI-driven profit growth exceeds the historical average by more than twice, fighting the earnings trend is dangerous. Jefferies’ S&P 500 target of 8,000 for 2026 is based on earnings per share (EPS) reaching $373 (up 35% YoY, much higher than the market’s 29% consensus) and a price-to-earnings (PE) multiple of 21.5x.

JP Morgan’s recent research notes that stock prices may already be pre-trading the cooling of the memory chip cycle, while a prolonged earnings super-cycle for the global AI computing chain—including memory chip components—can still occur. Led by Jay Kwon, JP Morgan’s Asia-Pacific analyst team favors Samsung Electronics, SK Hynix, Kioxia, Samsung Electro-Mechanics, ISU, and ISC over the next six months. The bet is that the upcycle in memory will last longer—and that demand for substrates, PCBs, and testing will persist.

JP Morgan explains from the industry’s strong earnings track that why a storage correction could create opportunities: Market suspicions over the sustainability of AI capex may not have translated into equivalent worsening for memory orders and profitability. AI inference requires continuous model weight reading, context and KV cache saving, and meeting different levels of data needs via server memory and SSD. HBM upgrades take up wafer, packaging, and yield ramp resources, making it difficult for effective supply to keep up with demand quickly.

JP Morgan’s Korean equity semiconductor research report shows that macro doubts about AI investment continuity are clearly disconnected from bottom-up supply and demand data in the memory industry. Hence, they view Q3 as the “expectations reset quarter” and favor six-month overweight positions in Samsung Electronics, SK Hynix, Kioxia, extending to Samsung Electro-Mechanics, and ISU and ISC—leading companies for AI PCBs and test sockets, respectively. This “duration trade” is a bet on a longer upcycle and sustained high profitability: Even if the DRAM/NAND price ramp decelerates, as long as demand for HBM, server DRAM, and enterprise SSD continues to soak up new capacity and delivers strong growth, the previously priced-in rapid profit decline may need reassessment. This also explains why storage stocks can fall ahead of a profit peak, but may preemptively rebound once earnings duration is confirmed; the persistent gap between actual profit paths and stock price-implied expectations determines the trading opportunity.

Retail buying frenzy meets valuation at cycle lows; AI-driven technology stock rebound still needs cash flow confirmation

The latest judgment from legendary Wall Street investor Howard Marks, Co-Chairman of Oaktree Capital, offers a contrarian angle on today’s market pessimism: True bubbles are usually accompanied by a loss of risk awareness, while persistent bubble discussions mean investors have not abandoned restraint. The forward price-earnings ratios of top tech companies are clearly lower than during the Internet bubble, related to their now-scale profits.

His view intersects with Bank of America and Morgan Stanley’s latest: Whether low PE ratios provide opportunities depends on whether profits can be sustained. If the market underestimates the duration of high profits, valuations may recover; if earnings forecasts are too high, low multiples could just be a cycle trap.

Howard Marks emphasizes that true bubbles are accompanied by collective disregard for risk, whereas current ongoing bubble debates show skepticism and restraint remain. Fund flow data shows this split: Goldman Sachs reports around $9.5 billion flowed into leveraged ETFs last week, about $7 billion into triple-long semiconductor ETFs (SOXL); meanwhile, U.S. fundamental long/short fund net exposure has dropped to a record low percentile, with gross exposure still at 213.9%. This means directional bets have contracted, but long/short positions remain large: If earnings cut pessimism, increased longs and short covering could jointly lift a rebound. Notably, leveraged ETF inflows by retail symbolize strong bottom-fishing intent but do not in themselves prove institutional funds have fully repositioned for the long term.

Bank of America’s latest report provides more concrete profit and valuation evidence for a potential bottom rebound in AI-powered technology stocks. Based on consensus profit estimates for 2027–2028, Samsung and Hynix trade at only about 2-3x earnings, showing cyclical trough valuations—but the actual memory market has not seen the price collapse typical of past downturns. The report shows that 16Gb DDR5 spot prices rose 3% week-over-week to $60.6 and forecasts a 17% month-over-month increase in October contract prices; even if Bank of America expects Samsung and Hynix 2028 EPS to decline low single digits YoY, both revenues and operating profits should remain near record highs.

In Bank of America analysts’ view, the market has already over-discounted future profit declines. Underlying support factors cited for memory chips and wider AI computing resources include HBM4 rollout, and HBM, SOCAMM, and enterprise SSD demand driven by advanced GPU/ASIC upgrades: Model weights, long-context caches, and massively concurrent AI inference workloads require different levels of memory and storage, plus a broader AI computing ecosystem including CPU, GPU/TPU, and data center high-speed optical interconnects. Even changes in single GPU configs cannot be directly equated to widespread declines in data center storage demand.

The interest rate and U.S. Treasury yield environment are also marginally improving. U.S. Treasury data shows that from October 9–16, the yields on 10-year and 30-year Treasuries fell by 15 and 17 basis points respectively, to 5.09% and 5.43%; on October 19 in U.S. pre-market, Brent crude retreated to about $107, with the 10-year yield at roughly 5.085%. But JP Morgan notes that the recent rally in long-dated bonds also reflects risk-off demand triggered by the AI selloff, and should not be taken as confirmation that inflation pressures have disappeared.

For the AI computing supply chain and technology stocks in general, a sustainable rebound requires three reinforcing pieces of evidence: capex spending supported by AI revenues and orders from major cloud firms and new cloud players for the AI theme; improvement in new project delivery and accounts receivable recovery to enhance repayment capacity; and easing of both U.S. Treasury yields and corporate credit spreads. Only when earnings reports prove that computing power investment can be converted into profit and cash flow does valuation compression become a true buy opportunity as Wilson suggests, giving low net exposure institutions good reason to increase risk again.