The AI Crash is Coming (And It’s Worse than 2008) | Edward Dowd
Dowd presents the AI buildout, the U.S. housing market, and China’s economic and demographic weaknesses as interconnected risks, with tightening credit, weak end-customer demand, and infrastructure constraints potentially slowing AI investment. He also points to consumer stress, concentrated equity exposure, Taiwan conflict risk, and possible disruption to the yen carry trade. Dowd expects a 40% to 50% S&P 500 drawdown, a projected 0% 10-year SPY return relative to bonds, and a potential rally in long-term Treasuries if recession and an AI-bubble burst shift capital toward bonds.
Preview · 12 more points in the app
Read the Full SummaryKey Takeaways
Dowd portrays the AI buildout as vulnerable on both the demand and financing sides: Apollo leadership reportedly sees most AI-complex revenue coming from investors rather than end customers, while private credit is weakening and OpenAI has floated government financing. He also cites a survey in which 50% of AI-infrastructure respondents reported double ordering, potentially overstating chip backlogs.
Dowd expects AI capital expenditures to slow if credit becomes too expensive or the power grid cannot support planned data centers, with consequences for semiconductor companies and power generators that currently benefit from infrastructure orders. That view is reinforced by his observation that the semiconductor-led rally has been narrow and that some indexes carry 40% to 50% AI or AI-adjacent exposure.
Dowd identifies the AI bubble, an overvalued U.S. housing market, and China’s economic crisis as interconnected risk pillars that could be accelerated by an energy-price shock from the Iran war. He points to Walmart’s weakest same-store-sales performance in six years as evidence of consumer stress, saying roughly 80% of the country is not doing well.
Dowd says the U.S. housing market is effectively frozen because homes are about 30% overvalued and the gap between homes for sale and homes sold is the largest on record. Because housing represents 20% of the economy and supports construction, financial activity, and furniture purchases, he argues that prices must fall before activity can resume.
GEO-POLITICS
“It's a risk and you know, Warren Buffett got out of Taiwan Semiconductor…”—
TECHNOLOGY
“That's because the engines of financing were starting to crack a little bit…”—
“Recently the head of Apollo said that it appears that most of the revenues coming from the AI complex is coming from investors not end customers…”—
ECONOMY
“Three pillars of risk. the ai bubble the housing market in the US and China's economic acute crisis that's unfolding…”—
“Walmart Walmart just reported its quarter…”—
“We have the largest amount of homes for sale versus homes sold gap on record, meaning the market is frozen. So, there's a buyer strike…”—
MARKETS
“Trump announced anou and the market ripped and semiconductors led that rip and it was a very narrow rally. It took us to new highs…”—
“I always had the thesis that it would eventually get manifested in the currency markets. That's the that's the release valve and it's beginning in Japan…”—
FORECASTS
“If you look forward on 10-year projected returns versus bonds, you're going to if if you took a million dollars and put it in the spy ETF, you'd ear you're…”—
“Well, look, you know, the.com bust produced a 50% draw down on the S&P 500 over two years…”—
Introduction
Anthony Faties welcomes Edward Dowd back to WTFinance, and Dowd outlines three interconnected risks: the AI bubble, U.S. housing, and China’s escalating economic crisis.
Overview of the economy?
Dowd argues that the AI buildout is weakening as private credit freezes, end-customer demand slows, token prices fall, and financing spreads widen across heavily indebted companies.
AI struggles
OpenAI’s government-financing discussions, insider departures, investor-funded revenues, semiconductor weakness, and power shortages suggest the AI investment cycle may be approaching a forced slowdown.
A house of cards
Dowd describes AI infrastructure as a house of cards vulnerable to double ordering, shadow inventories of NVIDIA chips, collateralized hardware, and nonbinding financing announcements.
How bad would crash be?
Dowd expects a potential 40% to 50% S&P 500 drawdown, arguing that weak consumers, deteriorating housing, extreme valuations, and concentrated AI exposure could combine into a severe recession.
Hyperscaler capex
Hyperscaler spending need not stop entirely to damage markets; Dowd says slowing incremental growth would pressure semiconductor valuations, especially if South Korean technology indexes continue making new lows.
US Real Estate
Dowd compares the prospective decline with the dot-com bust and the Global Financial Crisis, expecting substantial losses before monetary and fiscal authorities respond with renewed money printing.
China risks
U.S. housing is slowly rolling over, with affordability problems, a record gap between homes listed and sold, a buyer strike, and prices estimated to be roughly 30% overvalued.
Demographic issues
Dowd says housing’s demographic outlook has permanently weakened as boomers age, meaning homes should primarily be viewed as places to live rather than investments with strong long-term appreciation.
Plaza Accord 2.0
China faces excessive housing inventory, collapsing internal consumption, demographic decline, and falling fixed investment, while export expansion and trade conflict provide increasingly unreliable support.
Government debt crisis
China’s aging population, high savings, absent social safety net, and reluctance to redistribute wealth prevent the country from successfully shifting toward a consumption-led economy.
Bonds
A Plaza Accord-style restructuring may be necessary, but Dowd says China cannot replicate Japan’s export model because its economy is vastly larger, increasing trade tensions and Taiwan conflict risks.
One message to takeaway?
Dowd argues that recession and an AI-bubble burst should eventually pull down long-term yields, making bonds an asymmetric opportunity while investors preserve cash and avoid trying to short markets.
Related Episodes
Important Disclaimers
Not Investment Advice: For education only — not financial, tax or legal advice.
AI-Generated: Made by AI from source transcripts and may contain errors.
No Guarantees: Past performance does not predict future results; investing carries risk.
Our Commentary · Request Removal