Morgan Stanley: AI Investment Thesis Shifts from Infrastructure to Applications
nashnova research
Morgan Stanley warns that the AI investment thesis is undergoing a structural shift — the single narrative of accelerating data-center build-out is breaking down, and the market's focus is moving from infrastructure suppliers to application-layer companies that can directly boost earnings.
What exactly triggered this sell-off?
The S&P 500 posted its steepest one-day drop since "Tariff Day," falling roughly 4.36%; the Nasdaq pulled back more than 10% from its October high, entering correction territory.
The Philadelphia Semiconductor Index plunged over 12% in a single session; the VIX jumped from 16.6 to 28.2.
Morgan Stanley frames the sell-off as hedge-fund repositioning and momentum unwinds stacking on top of each other — not a recession signal. This means → the damage came from crowded trades, not collapsing fundamentals.
Options hedging, leveraged-ETF rebalancing, and trend-following sell programs may amplify swings, producing a pattern of brief rallies followed by fresh lows.
How much longer can AI capex keep growing?
Morgan Stanley projects AI-related capex (including leases) will decelerate from roughly 100% year-on-year growth in 2026, to about 60% in 2027, then to around 14% in 2028.
In plain terms = the most explosive phase of AI infrastructure spending is passing — growth drops from doubling to single digits in just two years.
The election outcome raised the bar for capex to beat expectations even further — that scenario was already hard to achieve before the vote.
This reflects a breakdown in the market's linear pricing logic that "AI investment only goes up."
Is the market already distinguishing "build-out" from "applications"?
On November 4, Microsoft, Amazon, and other big-tech investors in AI rose more than 6%, while data-center operators fell sharply and the Philadelphia Semiconductor Index hit a six-month low.
This means → money is already voting with its feet: companies spending on AI are favored; companies selling the picks and shovels are being dumped.
Morgan Stanley reads this divergence as a structural signal, not short-term noise.
How does Morgan Stanley suggest repositioning?
Core strategy: trim high-beta, order-driven suppliers that depend on accelerating data-center investment; keep companies that grow through their own earnings power rather than leverage, plus application-layer names that can directly improve profitability.
On timing, the bank advises: if a brief bounce follows the sell-off, use the rally to reduce supplier exposure — do not rush to buy back AI stocks on the first rebound.
Once bottom signals are confirmed, gradually rebuild portfolio beta with individual stocks that meet selection criteria.
Earnings look solid — so why stay cautious?
Headline earnings remain decent, but AI-related profits are heavily concentrated in a handful of names — a structural risk.
Demand is not uniformly strong: non-manufacturing recurring profits fell 1.7% year-on-year in Q2; retail profits came in 10.9% below consensus.
This means → the surface-level "earnings are fine" story masks deep internal divergence — focusing on stocks with pricing power or genuine profit-growth support matters more.
How to spot the bottom? Morgan Stanley offers three signals.
Signal one: a sharp spike in volatility and cross-asset correlation — watch whether selling turns into indiscriminate, "capitulation-style" liquidation.
Signal two: the bank's proprietary sentiment indicator and a tactical downside target range of 7,100–7,300 on the S&P 500 (November 5 close: 7,354).
Signal three: retail-investor behavior — if retail turns net seller (as in December 2018), the correction may extend; if retail stays on the sidelines or keeps buying, the drawdown may be near its end.
Morgan Stanley also flags reassessment triggers: if AI spending plans again substantially beat expectations and monetization shows real progress, the cautious stance will be revisited; if retail turns net seller and the correction spreads to credit markets, overall market beta will be reassessed.
市场有风险，内容仅供研究参考，不构成投资建议。