Morgan Stanley: Five Warning Signs of an AI Rally Inflection Point
Taylor Wilson
Morgan Stanley revisits the dot-com bubble playbook and lays out a five-signal framework for spotting an AI market top — the real red flag is not any single headline, but a structural shift in who is selling and who is buying.
Was the dot-com bubble driven by retail investors? Morgan Stanley says no
Academic research cited in the report shows institutional money led the rally — from January 1997 to March 2000, institutions accounted for 63.6% of active buying in tech stocks.
After the top, the structure flipped: institutional buying fell to 36.4%, retail direct buying rose to 49.0%, and another 14.6% flowed in through mutual funds.
This means → the defining top signal is "institutions exit first, retail catches the bag" — not a single-day crash.
Did hedge funds short the bubble early?
No. Research shows hedge funds stayed overweight tech and did not systematically short the bubble.
The key detail: hedge funds' excess exposure — the extra tech weight above the market — peaked roughly six months before the Nasdaq topped, then began to narrow.
In plain terms = the warning is not "smart money stopped buying." It is that smart money's rate of adding risk started to slow — marginal exposure began declining.
How are the five signals ranked? Which matters most?
Signal 1: Overseas capital flows (most important) — if foreign investors flip from net buyers to net sellers while retail, mutual funds, and margin traders become the main buyers, the market enters a fragile, late-bubble structure. Morgan Stanley notes Japan's AI and semiconductor trade has not yet shown this end-stage pattern.
Signal 2: Interest rates — Morgan Stanley's U.S. economics team expects no further rate hikes in 2026 and a return to cuts by March 2027. But if inflation re-accelerates and long-end rates rise sharply, high-multiple growth stocks remain vulnerable. This means → the more a stock's valuation depends on far-future cash flows, the more sensitive it is to discount-rate changes.
Signal 3: Equity supply pressure — concentrated IPO lock-up expirations were one trigger of the dot-com collapse, though academics debate this. Morgan Stanley's view: supply alone is not a "time bomb," but it amplifies vulnerability to other shocks. Equity supply pressure within the AI ecosystem is building.
Earnings are still growing — why could that be a warning too?
Morgan Stanley stresses the biggest difference from the dot-com era is fundamentals — many AI companies are already generating profits and cash flow, which was not the case then.
But the metric to watch is not "is EPS still rising?" — it is "are EPS expectations still being revised upward at an accelerating pace?"
In plain terms = even if earnings revisions stay positive (analysts are still raising), the moment the slope of those upgrades starts to flatten, share prices can come under pressure — the market trades the rate of change in expectations, not the absolute level.
What does the news flow tell us? Same headline, different reaction — what does that mean?
Morgan Stanley offers a practical test: for a negative headline of roughly the same severity, is the price impact getting larger and the recovery getting slower?
This reflects a shift from "bad news triggers a dip-then-rally" to "bad news keeps compounding" — a sign that the holder structure is already changing and fewer participants are willing to buy the dip.
Should you chase momentum or earnings revisions?
Morgan Stanley advises against simply chasing price momentum (buying what has already rallied) and favours the earnings-revision factor — the direction and pace of analyst EPS upgrades.
After the dot-com top, 12-month price momentum collapsed quickly, but the earnings-revision factor showed relatively more resilience.
This means → the earnings-revision factor is an "offence-and-defence" strategy — it can participate in further upside, yet may hold up better than pure momentum if the market reverses.
Content is for reference only, not financial advice.