Summaries > Stocks > China > China's Tech Stocks Crash 40%: Is Wall Street Next? | David Woo...

China's Tech Stocks Crash 40%: Is Wall Street Next? | David Woo

https://www.youtube.com/watch?v=Gn1XvAP5jCw

TLDR AI stocks face growing downside risk as slowing revenue growth, falling token prices, rising infrastructure costs, heavy debt, and crowded valuations undermine the case for winner-take-all profits. David Woo sees China’s AI selloff as a warning for U.S. markets, doubts Anthropic can justify a $2 trillion valuation, and expects political and financing pressures to constrain data-center spending. He also warns that possible U.S.–Israeli action against Iran could spike oil prices and deepen an AI-led market correction, while flagging France’s political crisis and China’s investment risks; he remains interested in gold but is less certain about its near-term outlook.

Key Insights

1. Test AI valuations against realistic growth and margins

Do not value AI companies on the assumption that one provider will capture the entire market. Assess whether revenue can keep growing as models and computing become more commoditized, and compare expected profit margins with the costs of developing and running AI systems. The discussion argues that slowing growth at major AI firms and limited increases in total token spending make very ambitious forecasts harder to support. For any proposed valuation, work backward to identify the revenue and profitability it requires, then decide whether those assumptions seem achievable.

2. Track both AI demand and the cost of supplying it

Rising AI usage does not automatically translate into rising revenue or stronger margins. Falling prices per token can offset growth in demand, while GPU rental costs, data-center construction, grid capacity, and permitting can add expense or delay the start of operations. Monitor indicators such as token spending, compute prices, infrastructure utilization, and reported revenue growth. Pay particular attention to whether companies can bring new capacity online before their hardware loses value.

3. Evaluate hyperscaler spending through cash flow and financing

Large technology companies are investing heavily in AI infrastructure, which can make their businesses more capital-intensive and reduce free cash flow. Separate spending justified by clear customer demand from investment driven mainly by fear of missing out. Review how companies plan to fund capital expenditure, including borrowing, and watch for rising borrowing costs or weakening demand from bond investors. The interview highlights debt-market pressure on AI infrastructure firms as a reason to scrutinize financing risk across the sector.

4. Avoid crowded trades without checking for downside catalysts

A popular investment theme can become vulnerable when growth slows, financing tightens, or market conditions change. The speaker points to weakening technical signals, falling AI-related stocks, and rising bond yields as factors that could intensify losses in an overcrowded AI trade. Political decisions about AI regulation and data-center construction may also affect expected growth. Before investing, consider how a position could perform if several of these risks occur together, rather than relying only on the long-term AI narrative.

5. Include oil and geopolitics in portfolio risk assessments

Oil prices can affect technology stocks, inflation expectations, and government bond yields, so energy and geopolitical risks are relevant even to investors focused on AI. Shipping costs, insurance premiums, and uncertainty around the Strait of Hormuz can lift the cost of oil independently of changes in supply. The discussion warns that military action involving Iran could create a sudden price shock, while emphasizing that such outcomes are uncertain. Use scenarios—not predictions—to assess how higher oil prices might affect equities, bonds, and portfolio liquidity.

6. Treat regional opportunities and forecasts with caution

The interview also raises potential risks and opportunities in China, Brazil, France, and gold, but these depend on political and economic conditions that can change quickly. Before acting on a regional investment idea, check the underlying evidence, current market prices, and your portfolio’s exposure to the relevant risks. For example, claims about oil flows through the Strait of Hormuz varied by period, and gold’s recent decline challenged a simple safe-haven thesis. Keep forecasts separate from verified facts and size positions to reflect uncertainty.

Questions & Answers

Why does David Woo think the AI trade may be overvalued?

He argues that markets assume a winner-take-all outcome, but AI is more likely to become commoditized. Meanwhile, revenue growth appears to be slowing, token spending has stalled, and falling prices may be offsetting rising demand. These trends make ambitious revenue forecasts and current valuations difficult to justify.

What does Woo say Anthropic would need to justify a hypothetical $2 trillion valuation?

Anthropic would need to reach about $1.2 trillion in revenue by 2030 to compare favorably with companies such as TSMC or Meta. Woo says its losses, reported slowdown in third-quarter growth, and difficult path to that level of revenue weaken the investment case. He expects its valuation to be closer to $1.4 trillion or less.

What pressures are squeezing AI companies’ margins?

AI token prices are falling while compute costs are rising. GPU rental rates are increasing, and data-center construction is delayed by grid capacity, permitting, and other constraints. Woo also suspects some newly purchased Nvidia Rubin chips may remain unused while depreciating before installation.

Why are investors questioning hyperscalers’ AI spending?

Companies such as Meta, Amazon, Microsoft, and Google are shifting toward capital-intensive, infrastructure-like businesses. Their heavy AI investment has pushed some toward negative free cash flow, and Woo says fear of missing out is helping drive spending. Investors and bondholders are increasingly concerned about how companies will finance continued capital expenditure.

What risks does Woo see for AI infrastructure financing and the broader AI trade?

He says debt is tightening for AI infrastructure firms, citing CoreWeave borrowing at 15% and a report that 200 neoclouds had lost access to debt markets. He also points to slowing revenue, rising yields on AI-related bonds, technical weakness in the Nasdaq 100, and possible regulation or limits on data-center construction. Despite losses among some hedge-fund strategies, he sees little evidence of investor capitulation.

How could higher oil prices affect the AI trade and bond yields?

Woo says an oil-price spike could undermine the AI trade’s recovery and hurt the global economy. Elevated oil could weigh on AI-heavy Nasdaq stocks; in the near term, an oil-driven correction could also lift Treasury bonds and lower 10-year yields by around 20 basis points, given the recent correlation between yields and the Nasdaq.

Why does the speaker think oil prices are high, and what is the risk around the Strait of Hormuz?

Shipping oil from the Gulf to China reportedly costs about $40 per barrel because shipowners demand large risk premiums. Higher freight costs, low inventories, and resilient demand are pushing refiners to bid up oil prices. The speaker says Iran benefits from higher prices without closing the Strait of Hormuz, but could potentially cut Gulf oil flows to zero; reported throughput figures vary, with Kepler data showing a recent fall to 9 million barrels per day.

How likely does the speaker think U.S. military action against Iran is in the near term?

The speaker estimates the chance of U.S. action in the next week or two at greater than 50% and says oil markets have not priced in the risk. They cite reports that U.S. forces could be ready to strike on 96 hours’ notice and suggest Donald Trump may act before or after the U.S. midterms.

How might Israel’s election affect the prospect of action against Iran?

The speakers expect heightened U.S.–Israeli action before the U.S. midterms, potentially while Benjamin Netanyahu remains prime minister. Even if Netanyahu loses, coalition negotiations could take weeks, creating a transition period that Trump might view as an opportunity to act. One speaker also suggests polls may underestimate right-wing support among younger voters with recent IDF service, though Netanyahu’s victory remains uncertain.

What is the speaker’s view on buying Brazilian assets after Lula’s reelection?

One speaker says the market’s negative reaction to Luiz Inácio Lula da Silva’s reelection may create an opportunity to cautiously buy Brazilian assets. They note that Jair Bolsonaro accepted the result and that Brazil could help diversify a portfolio.

What are David Wu’s main concerns about Europe and China?

Wu is increasingly bearish on Europe and warns that unrest in France could escalate into a major crisis before the next April election, making the coming year difficult. He also sees China as vulnerable: its expected AI-driven economic boost has weakened, and geopolitical uncertainty under Donald Trump may deter foreign investors.

What is Wu’s view on gold as a 2027 investment?

Wu is interested in gold as a 2027 trade against a basket of currencies, but its sharp recent decline—even as 10-year real yields fell—has made him less certain it would immediately benefit from an AI bubble bursting. He suggests hedge funds may have sold gold to raise liquidity or meet margin calls.

Summary of Timestamps

David Woo argues that U.S. AI valuations assume one company will dominate the market, even though AI may become a commodity with many competing providers. He points to China’s crowded AI sector and the roughly 40% decline in the STAR 50 index as a possible warning for U.S. AI stocks, despite China’s improving AI capabilities.

Discussing Anthropic’s IPO filing, Woo says its reported revenue is substantial, but its losses and proposed valuation make the investment case challenging. At a hypothetical $2 trillion valuation, Anthropic would need about $1.2 trillion in revenue by 2030 to compare favorably with companies such as TSMC or Meta. He also notes that reported growth slowed sharply in the third quarter.

The speakers question whether AI companies can meet ambitious revenue forecasts. OpenAI and Anthropic growth is reportedly slowing, and Ramp data showed no increase in total AI-token spending in August or September. Lower token prices may be offsetting increased usage, weakening the case for rapid revenue growth.

AI companies are also facing rising costs. Rental prices for older Nvidia H100 GPUs are increasing, while data-center projects encounter grid-capacity, permitting, and construction delays. Woo suspects some newly purchased Nvidia Rubin chips may remain unused, risking depreciation before installation; together, falling prices and rising expenses could squeeze margins.

Woo says long-term revenue growth and profit margins are crucial to valuing Anthropic and OpenAI, and he doubts current market assumptions will hold over the next two to four years. Anthropic may be able to raise its targeted $100 billion, but Woo expects its valuation to remain closer to $1.4 trillion or less rather than reach $2 trillion.

The discussion broadens to hyperscalers such as Meta, Amazon, Microsoft, and Google. Heavy AI investment is turning these formerly capital-light businesses into infrastructure-heavy, utility-like operations, and some are approaching or entering negative free cash flow. Woo says fear of missing out is helping drive spending alongside business fundamentals.

Investors are increasingly asking how technology companies will finance continued capital expenditure, while bond investors are growing wary after purchasing substantial AI-related debt. The excerpt begins to raise the prospect of another $500 billion in such borrowing in 2027. A sponsored segment then introduces Monetary Metals, a platform offering monthly yield in physical gold through gold leasing.

Woo sees mounting downside risks for the AI trade. He cites tighter debt access for infrastructure firms, CoreWeave borrowing at 15%, and a reported comment from Elon Musk that 200 neocloud companies had lost access to debt markets. Although reduced competition could help hyperscalers raise prices, Woo says the AI trade remains overcrowded and investors have not clearly capitulated.

Slowing revenue, higher yields on AI-related bonds, technical weakness in the Nasdaq 100, and possible political pressure to regulate AI and restrict data-center construction are making the risk balance more bearish. Woo also warns that an oil-price spike could abruptly reverse an AI-market recovery and damage the broader global economy.

The speakers discuss the possibility of U.S.–Israeli military action against Iran. They say an attack may have been considered for the weekend but was reportedly blocked by Israel’s military chief because of the election timetable and lack of immediate urgency. With that objection potentially removed, and U.S. forces reportedly able to act on 96 hours’ notice, they see a greater chance of action later in the week.

The speakers expect the possibility of heightened U.S.–Israeli action before the U.S. midterms, potentially while Benjamin Netanyahu remains prime minister. Even if Netanyahu loses Israel’s close election, coalition talks could take weeks, creating a transition period in which Donald Trump might see an opportunity to act. One speaker also thinks polls may underestimate right-wing support among younger Israelis.

On Brazil, one speaker says the negative market reaction to Luiz Inácio Lula da Silva’s reelection may create an opportunity to buy Brazilian assets cautiously and diversify a portfolio. The conversation then turns to Strait of Hormuz oil flows: Kepler data reportedly showed throughput falling to 9 million barrels per day the previous week, complicating claims that shipping had returned to pre-war levels.

Oil prices are being supported by the high cost and risk of shipping Gulf crude to Asia. The speakers estimate that transporting a barrel to China now costs about $40, as shipowners demand risk premiums for operating in a dangerous region. Higher freight costs prompt Asian refiners to bid up Brent and WTI, while low inventories and resilient demand have meant Saudi Arabia has not needed to discount its oil.

Woo believes the chance of U.S. military action against Iran in the next week or two is greater than 50%, and that oil markets have not fully priced in the risk. He argues Trump may be willing to act before or after the midterms, particularly if he believes there is little political downside. In Woo’s view, Trump’s decisions have helped drive both this year’s higher-oil and higher-AI trades.

For equities, elevated oil prices could further pressure the crowded AI trade. The speakers describe Nasdaq technicals and fundamentals as weak, noting that last week’s oil rise coincided with a sharp Nasdaq decline. In the short term, an oil-driven equity correction could even push 10-year Treasury yields down by around 20 basis points, given the recent relationship between bond yields and the Nasdaq.

Woo turns increasingly bearish on Europe, warning that unrest in France could become a major crisis before next April’s election. He also sees China as vulnerable: hopes for an AI-driven economic boost have weakened, and geopolitical uncertainty under Trump may discourage foreign investment. He is interested in gold as a 2027 trade against a currency basket, but its recent decline despite falling real yields has made him less certain.

Woo suggests that hedge funds may have sold gold to raise cash or meet margin calls, rather than because its longer-term case has disappeared. The host closes by thanking Woo, congratulating him on his channel’s growth, and encouraging viewers to watch his work, like the video, and subscribe.