Key Takeaways
- • Jean-Louis Nakamura, head of conviction equities at Vontobel Asset Management, told Fund Selector Asia that the AI capital expenditure cycle has not yet peaked and will continue for at least the next two years, with the capex-to-sales ratio expected to peak at the end of next year.
- • Nakamura argues AI hardware and semiconductor companies retain strong earnings visibility over the next 12 to 18 months, and that recent market fears about AI capex slowing are excessive.
- • High bandwidth memory and substrates have seen lead times extend from three months to one year, with no clear easing expected before 2030, which he cites as evidence that this cycle differs from past semiconductor boom-bust patterns.
- • For diversification away from AI, Nakamura points to parts of healthcare (hospital services and certain biotech segments), with financials and automation as secondary options.
- • Emerging markets have been resilient to recent tariff, oil and long-term yield shocks, and Nakamura says they should now be viewed as a diversifier of global equity exposure rather than a high-risk satellite allocation.
Summary
Despite recent volatility in AI-related equities, Jean-Louis Nakamura, head of conviction equities at Vontobel Asset Management, told Fund Selector Asia that the current capital expenditure cycle has not yet peaked and that companies supplying AI hardware still have strong earnings visibility over the next year. He said that investor fears around the AI trade have become excessive.
Nakamura attributed the capex cycle to a competitive race among developers of frontier AI models, each seeking to lock in clients and build ecosystems that strengthen pricing power. He drew a parallel to Microsoft and Amazon's earlier cloud strategies. He expects capex to keep rising for at least two years, with the ratio of capex to sales peaking at the end of next year and remaining elevated relative to history afterward.
On semiconductors and memory, Nakamura said investors are applying the lens of a traditional super-cyclical market, in which oversupply and price collapses follow roughly two-year upswings. He argued that this cycle is different because demand is unusually strong and capacity is tight and difficult to expand within two and a half years. He cited high bandwidth memory and substrates, where lead times have moved from three months to a year with no clear easing before 2030. He said semiconductor and memory makers had been punished by the market over recent months on fears that capex would slow, and that they could again surprise positively versus expectations.
Nakamura acknowledged that diversifying away from AI exposure is not straightforward. He identified parts of healthcare, notably hospital services and some biotech segments, as potentially interesting. He said financials have benefited from rising rates but that long-term interest rates are probably near a peak. Other industrial sectors could perform if manufacturing growth broadens, and automation offers pockets of good earnings visibility, though AI-related hardware and semiconductors remain the segment with the best visibility over 12 to 18 months.
Nakamura also made the case for emerging markets as a diversifier. He noted that these markets have been resilient through U.S. tariff increases, a spring oil price shock tied to the Middle East, and a recent sudden rise in long-term yields. He attributed this to structural changes over the past 15 years, including more mature domestic demand, lower and more stable inflation, more balanced external accounts, and reduced dependence on trade with developed economies. He said emerging markets, with the exception of the last few weeks, have persistently outperformed developed markets and should now be viewed as a way to diversify global equity exposure rather than a high-risk, high-volatility allocation.
Nakamura attributed the capex cycle to a competitive race among developers of frontier AI models, each seeking to lock in clients and build ecosystems that strengthen pricing power. He drew a parallel to Microsoft and Amazon's earlier cloud strategies. He expects capex to keep rising for at least two years, with the ratio of capex to sales peaking at the end of next year and remaining elevated relative to history afterward.
On semiconductors and memory, Nakamura said investors are applying the lens of a traditional super-cyclical market, in which oversupply and price collapses follow roughly two-year upswings. He argued that this cycle is different because demand is unusually strong and capacity is tight and difficult to expand within two and a half years. He cited high bandwidth memory and substrates, where lead times have moved from three months to a year with no clear easing before 2030. He said semiconductor and memory makers had been punished by the market over recent months on fears that capex would slow, and that they could again surprise positively versus expectations.
Nakamura acknowledged that diversifying away from AI exposure is not straightforward. He identified parts of healthcare, notably hospital services and some biotech segments, as potentially interesting. He said financials have benefited from rising rates but that long-term interest rates are probably near a peak. Other industrial sectors could perform if manufacturing growth broadens, and automation offers pockets of good earnings visibility, though AI-related hardware and semiconductors remain the segment with the best visibility over 12 to 18 months.
Nakamura also made the case for emerging markets as a diversifier. He noted that these markets have been resilient through U.S. tariff increases, a spring oil price shock tied to the Middle East, and a recent sudden rise in long-term yields. He attributed this to structural changes over the past 15 years, including more mature domestic demand, lower and more stable inflation, more balanced external accounts, and reduced dependence on trade with developed economies. He said emerging markets, with the exception of the last few weeks, have persistently outperformed developed markets and should now be viewed as a way to diversify global equity exposure rather than a high-risk, high-volatility allocation.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.