A rerun of the 1997 Asian Financial Crisis? HSBC's chief economist sees significant similarities

HSBC’s chief economist warns that Asia is showing similarities to the pre-1997 crisis, mainly due to surging U.S. Treasury yields: the 10-year climbed from ~0.5% (Aug 2020) to ~4.79% recently, and rose ~80 bps since 3.9% in February. He also flags yen weakness of ~57% from ~103 (Jan 2021) to ~163 (Jul 2026), with markets watching for renewed intervention. While Asia is less exposed to capital-flight-style financial stress, Neumann says the key risk is “demand vulnerability” from Asia’s dependence on U.S. AI hardware demand—if higher USD funding costs or yen volatility dent that demand, regional growth could “fizzle.”

Analysis

This is not a 1997-style balance-of-payments stress event; it is a concentration-risk story. The market’s real fragility sits in the narrowness of the AI hardware spend cycle: if U.S. real yields stay elevated, the first transmission is not Asian FX, but order-book air pockets in semis, test equipment, substrates, and packaging across Taiwan, Korea, and Japan. That creates a delayed earnings risk for names tied to capex intensity rather than spot demand, with multiple compression likely to show up before any hard-data recession signal.

The second-order effect is that Asia’s exporters are now more exposed to U.S. corporate capex than to regional funding stress. That makes the vulnerable basket broader than just chipmakers: logistics, industrial automation, and selected banks with lender exposure to hardware supply chains can see weaker fee growth and slower credit demand if the AI cycle pauses. HSBC is less a direct beneficiary/loser than a barometer of Asia-linked corporate activity; the real pressure would come through slower trade finance, weaker wealth flows, and later-stage credit normalization, not an immediate capital flight shock.

Contrarian view: consensus is overpricing the analogy to 1997 and underpricing how binary AI demand has become for several Asian economies. The thesis breaks if U.S. 10-year yields roll over, if yen intervention stabilizes global funding sentiment, or if the next round of hyperscaler capex guidance re-accelerates. In that case, the market can quickly re-rate this from macro scare to temporary volatility, and the best shorts would cover sharply on relief rallies.

Key Decisions for Investors

• Short SOXX or SMH for 1-3 months as the cleanest expression of 'AI demand vulnerability' if U.S. yields remain near recent highs; target a 10-15% drawdown on multiple compression, stop if NVDA/TSM guide capex materially higher.
• Pair trade: short EWT or EWY against long SPY for 1-3 months. Taiwan/Korea have the highest beta to the AI hardware cycle, while the U.S. index is more diversified; this should work if export orders decelerate before macro data confirm it.
• Watchlist, not a trade yet: HSBC. A better entry would be on evidence of slowing Asia trade-finance growth or softer fee income in the next earnings cycle; higher rates alone are not enough to justify a short because NII can offset part of the slowdown.
• If the 10-year Treasury remains above ~4.5% and the yen fails to recover after intervention, consider adding downside via semicap names with the most AI-capex sensitivity rather than broad Asia ETFs; the thesis is strongest in suppliers closest to hyperscaler spend.
• Falsifier trigger: if U.S. yields fall 50-75bp from current levels or hyperscaler capex guidance is revised up again, cover semis shorts and rotate to a neutral stance on Asia exporters.