Following a sharp selloff in US equities, Morgan Stanley warned that the investment logic for AI is undergoing a structural shift — the single narrative of betting on accelerated data center construction is no longer sustainable, and the market's focus will move from infrastructure suppliers toward the application layer that can directly improve profitability.

According to a Morgan Stanley research report, the S&P 500 recorded its largest single-day decline since "Tariff Day" on November 5, the Nasdaq Composite fell more than 10% from its October high to enter correction territory, and the Philadelphia Semiconductor Index (SOX) plunged more than 12% in a single day. Morgan Stanley believes the core driver of this selloff was hedge fund positioning adjustments spilling over into broad market indices rather than deteriorating economic fundamentals, but the process deeply exposed the market's excessive reliance on the single premise of "continuously accelerating AI capital expenditure."

Morgan Stanley maintained its strategy unchanged: reduce exposure to high-beta, order-driven suppliers, and after confirming the end of "capitulation selling," gradually rebuild portfolio beta through companies with earnings growth capability. The bank also noted that AI-related capex growth is expected to slow from about 100% in 2026 to about 60% in 2027, and further to about 14% in 2028, with the threshold for investment acceleration exceeding expectations now substantially raised.

Nature of the selloff: positioning adjustment, not fundamental collapse

Morgan Stanley characterized this sharp decline as a combined effect of hedge fund positioning adjustments and momentum strategy unwinding, rather than a signal of recession.

In terms of market performance, clear divergence had already emerged within US AI-related stocks on November 4: large technology companies on the investment side, such as Microsoft and Amazon, rose more than 6% in a single day, while data center operators fell sharply, with the SOX index hitting a six-month low. On November 5, selling spread further to the broad market, with the S&P 500 closing down about 4.36% and the VIX jumping from 16.6 to 28.2.

The bank also cautioned that supply-demand factors such as options hedging, leveraged ETF rebalancing, and CTA trend-following selling could amplify the decline, leading to repeated brief rebounds followed by new lows after sharp drops.

AI narrative shift: from "buildout" to "usage"

Morgan Stanley explicitly stated that the election outcome dealt the most profound blow to AI sentiment, and market pricing expectations for accelerated AI investment face a systematic reassessment.

On the capex front, Morgan Stanley's US equity strategy and GVAT teams expect AI-related capex growth, including leases, to slow from about 100% year-over-year in 2026 to about 60% in 2027 and about 14% in 2028. The bank noted that even before the election results, a scenario where investment acceleration exceeded what stock prices had already priced in was already difficult to achieve, and the election results further raised that bar.

Based on this judgment, Morgan Stanley's framework adjustments are as follows: reduce exposure to high-beta order-driven suppliers dependent on accelerated data center investment; retain companies that grow through their own profitability rather than borrowing, as well as areas like the application layer that can directly improve profitability.

Additionally, on earnings, Morgan Stanley noted that current earnings data is broadly solid, but AI-related profits are highly concentrated in a handful of stocks, warranting vigilance against structural risks. The bank also cautioned that demand is not uniformly strong — non-manufacturing recurring profits fell 1.7% year-over-year in the second quarter, and retail profits came in 10.9% below market consensus, suggesting selective exposure to domestic demand stocks, focusing on names with pricing power or profit growth support rather than indiscriminate buying.

Confirming the bottom: three key signals and follow-up focus

Morgan Stanley listed three key signals for judging a US equity bottom: first, a sharp rise in volatility and correlation, with a need to watch whether correlation rises further and whether selling escalates into indiscriminate "capitulation" liquidation; second, the bank's US equity strategy team's sentiment indicators and the S&P 500 tactical downside target range of 7,100 to 7,300 points (November 5 close was 7,354); third, retail investor behavior — if retail investors turn net sellers (as in December 2018), the adjustment may continue; if retail investors remain on the sidelines or keep buying, the adjustment may be nearing its end.

On timing, Morgan Stanley advised: if a brief rebound follows a sharp drop, use the rebound to reduce high-beta order-driven suppliers rather than rushing to buy back AI-related stocks on the initial bounce; after bottom signals are confirmed, gradually rebuild portfolio beta with stocks that meet selection criteria.

The bank also clarified conditions for reassessing its strategy: if AI investment plans once again substantially exceed market expectations and monetization makes substantive progress, or if the scope of Texas approval suspensions is limited, it will revisit its cautious stance on AI; if US retail investors turn net sellers and the adjustment spreads to credit markets and the economic outlook, it will reassess overall market beta.