The Nam Nam Self-Talk attempts to deconstruct the recent global AI stock selloff from several angles.
1) The core issue isn't a sudden drop in demand, but rather the market's growing fear that computing power and infrastructure companies can't sustain financing. Since Broadcom (US.AVGO)'s Hock Tan and Intel (US.INTC)'s Lip-Bu Tan publicly raised concerns about AI financing, market focus has shifted from "buying growth" to scrutinizing "who can borrow and repay" like a post-mortem.
2) Neocloud borrowing costs have become absurdly high. Take CoreWeave (US.CRWV) as an example: its public senior debt coupon rate is around 9.6% to 9.8%, while secondary market yields一度 surged to nearly 15%, completely disconnecting coupon rates from actual refinancing costs.
3) The AI infrastructure sector's frenzied bond issuance has drained funds from the credit market, directly pushing up credit spreads and risk pricing. Risk assets must now offer higher premiums, while refinancing windows continue to shrink.
4) Although reported earnings growth remains strong and backlogs are substantial, once financing channels narrow and interest rates remain elevated, companies may have to cut CapEx guidance, consequently delaying contract fulfillment. Growth achieved by burning cash becomes the biggest variable when "money becomes too expensive."
5) The market's greatest fear is a funding chain gap. If hardware orders, data center progress, and customer payment timelines become misaligned, AI hardware stocks could easily become value traps—appealing on paper, but suffering a double blow to cash flow and valuation.
6) In a high-interest environment, capital traditionally favors clustering into themes like AI hardware and cybersecurity, which offer "high earnings certainty." However, positions have become excessively crowded. Once risk appetite weakens, the market lacks consensus on where safe-haven funds might flow next.
7) Fund flows already show money moving into defensive sectors such as Staples, Utilities, Energy, and Healthcare, which theoretically could outperform in the short term. However, these stocks aren't cheap in valuation, have low volatility, and considering the 10-year U.S. Treasury yields around 5%, buying U.S. Staples stocks offering only 2% dividends hardly justifies the opportunity cost in terms of risk/reward.
8) Until financing costs decline and CapEx realization becomes clearer, it's better to increase cash holdings and preserve ammunition, waiting for both valuations and financing conditions to improve before re-entering. This approach is preferable to stubbornly holding corrections within highly concentrated, highly leveraged themes—drawdowns and stop-losses are painful and unpleasant, but risk management necessitates forced reductions. Mo Ho Nam, Fund Manager (Equities), Yuan Universe Securities
*Articles published in ETNet, whether signed or unsigned, represent the authors' personal opinions and do not reflect ETNet's stance. ETNet serves as a platform providing free speech.
1) The core issue isn't a sudden drop in demand, but rather the market's growing fear that computing power and infrastructure companies can't sustain financing. Since Broadcom (US.AVGO)'s Hock Tan and Intel (US.INTC)'s Lip-Bu Tan publicly raised concerns about AI financing, market focus has shifted from "buying growth" to scrutinizing "who can borrow and repay" like a post-mortem.
2) Neocloud borrowing costs have become absurdly high. Take CoreWeave (US.CRWV) as an example: its public senior debt coupon rate is around 9.6% to 9.8%, while secondary market yields一度 surged to nearly 15%, completely disconnecting coupon rates from actual refinancing costs.
3) The AI infrastructure sector's frenzied bond issuance has drained funds from the credit market, directly pushing up credit spreads and risk pricing. Risk assets must now offer higher premiums, while refinancing windows continue to shrink.
4) Although reported earnings growth remains strong and backlogs are substantial, once financing channels narrow and interest rates remain elevated, companies may have to cut CapEx guidance, consequently delaying contract fulfillment. Growth achieved by burning cash becomes the biggest variable when "money becomes too expensive."
5) The market's greatest fear is a funding chain gap. If hardware orders, data center progress, and customer payment timelines become misaligned, AI hardware stocks could easily become value traps—appealing on paper, but suffering a double blow to cash flow and valuation.
6) In a high-interest environment, capital traditionally favors clustering into themes like AI hardware and cybersecurity, which offer "high earnings certainty." However, positions have become excessively crowded. Once risk appetite weakens, the market lacks consensus on where safe-haven funds might flow next.
7) Fund flows already show money moving into defensive sectors such as Staples, Utilities, Energy, and Healthcare, which theoretically could outperform in the short term. However, these stocks aren't cheap in valuation, have low volatility, and considering the 10-year U.S. Treasury yields around 5%, buying U.S. Staples stocks offering only 2% dividends hardly justifies the opportunity cost in terms of risk/reward.
8) Until financing costs decline and CapEx realization becomes clearer, it's better to increase cash holdings and preserve ammunition, waiting for both valuations and financing conditions to improve before re-entering. This approach is preferable to stubbornly holding corrections within highly concentrated, highly leveraged themes—drawdowns and stop-losses are painful and unpleasant, but risk management necessitates forced reductions. Mo Ho Nam, Fund Manager (Equities), Yuan Universe Securities
*Articles published in ETNet, whether signed or unsigned, represent the authors' personal opinions and do not reflect ETNet's stance. ETNet serves as a platform providing free speech.
