Morgan Stanley Private Meeting: AI Trade Overcrowded, HK Internet Stocks Enter Recovery Window
Claire Weston
A closed-door Morgan Stanley session warns that global AI hardware positioning has hit a historical extreme — structurally fragile — and capital is rotating toward Hong Kong internet stocks, signaling a shift from one-way AI bets to a more balanced allocation.
How crowded is the AI hardware trade?
Global hedge-fund exposure to semiconductors and memory is near a historical extreme, with heavy leveraged capital flowing in via 2× leveraged ETFs and similar instruments.
This means → the fundamentals may still hold, but the trade structure is deeply fragile — any shift in marginal rate expectations could trigger deleveraging or a stampede.
In plain terms = it looks like gold and crypto at their most frenzied peaks: the problem isn't the asset, it's that too many people are on the same side of the boat.
Where is the money coming from — and where is it going?
U.S. AI giants have already spent over $800 billion in capex this year; next year the figure may approach $1.2 trillion.
They still need to raise capital through bonds, equity, and IPOs over the coming year — AI is absorbing both industrial resources and financial-market liquidity at the same time.
This means → AI isn't just consuming chips and power; it's consuming the market's cash. Any mismatch between the funding pace and market confidence could send liquidity pressure back through the system.
Which three variables could break the balance?
The Fed's path: rate markets briefly priced in roughly a one-in-three chance of a hike at the late-July meeting. An inflation shock that turns the Fed hawkish would hit high-valuation AI assets directly.
Geopolitics and oil: Strait of Hormuz shipping capacity has not fully recovered; U.S. strategic petroleum reserves sit near historic lows. A renewed oil rally would re-ignite inflation expectations and feed through to global rates.
Big Tech earnings season: market trust in ever-higher AI capex is starting to waver. If the giants signal a slowdown — or if returns face skepticism — the AI narrative chain gets a stress test.
Why is the Hong Kong repair window opening now?
Morgan Stanley strategist Laura Wang judges that Hong Kong stocks entered a repair window around August, and the move is already under way.
Two capital flows are converging: institutions that shorted liquid Hong Kong names in H1 are unwinding, easing selling pressure; meanwhile, southbound capital flipped from net selling in May to accelerated buying in June–July.
Three catalysts lie ahead: Q2 earnings likely favour internet names; Tencent, Alibaba and peers will disclose AI progress more intensively from July onward; and after a July IPO lock-up expiry peak, August liquidity pressure should ease noticeably.
Is this a broad China bull market?
The session was explicit: no. The opportunity is structural repair in Hong Kong internet, quality consumer leaders, industrial capital goods, and select hard-tech — not a broad rally.
On policy, the late-July leadership meeting may deliver incremental support, but most likely by accelerating existing projects, not large-scale stimulus. Roughly RMB 2 trillion in fiscal and quasi-fiscal resources in H2 is expected to flow into AI computing-centre networks and ultra-high-voltage grids with energy storage.
Chief economist Xing Ziqiang stressed that the real lever for domestic demand is a third network — the social safety net — especially for 300-million-plus migrant workers and gig workers. In plain terms = without basic coverage, these workers keep saving defensively and consumer spending stays locked up.
Which specific plays does Morgan Stanley favour?
Consumer stocks: near-term fundamentals remain cautious, but average valuations have dropped from 14–15× to 12–13×, below the prior weak-cycle trough. Some quality leaders were sold off indiscriminately as "snowball" collateral; overseas long-only funds have started re-engaging over the past two to three weeks.
Weichai Power: driven by AI power-generation demand, AI-related profit may grow roughly 350% this year and over 150% next year, while the stock trades at about 18× earnings — a favourable growth-to-valuation balance.
This reflects Morgan Stanley's core logic: the second half is not about a single directional bet, but a gradual shift from the overcrowded one-way AI hardware trade to a more balanced allocation. Whether this rotation lands smoothly depends on the pace at which the Fed, oil prices, and Big Tech earnings play out.
Content is for reference only, not financial advice.