Capital is tilting East: Inside Asia’s AI gold rush

Joyce Chang · J.P. Morgan

Debt markets may struggle to absorb the scale of AI and hyperscaler financing as yields rise.

There is more concern about the debt market's capacity to absorb the scale of AI and hyperscaler financing. The main reason is the surge in developed-market sovereign bond yields, which adds competition for capital and raises its cost. This partly offsets the view that hyperscaler spending is insensitive to rates because of strong cash generation.

“But there's more concern about the market's capacity to absorb the scale of AI and hyperscaling financing, particularly in the debt markets. And this is in large part because we've seen the surge in sovereign bond yields in developed markets”at 6:50

Sjoerd Leenart · J.P. Morgan

Asia will host most data center investment after the US, though some governments are pausing projects.

Asia is a rapidly emerging theme for data centers, and after the US most data center investment and activity is expected there. Chinese companies and global hyperscalers need computing power in the region and are looking for places where land, electricity and grids can support it. Governments in places such as Thailand and Australia have paused some projects to check whether energy demand affects other industries and consumers. Financing remains a question, so the bank is seeking alternative sources of liquidity.

“it's certainly a rapidly emerging theme in Asia. And after The US, we expect actually most data center investments and activity to happen in Asia.”at 8:47

Joyce Chang · J.P. Morgan

AI capex rising from about $900B to $1.5T or more; compute demand exceeds supply for two to three years.

AI capex is expected to rise from about $900B this year to around $1.5T or more. Demand for computing capacity still exceeds available supply, so infrastructure bottlenecks will likely persist for another two or three years. Hyperscaler spending plans look relatively insensitive to interest rates because of strong cash generation and strategic priorities, and the AI build-out is still a hyper-growth regime given earnings visibility.

“and we actually think that AI CapEx is going to increase to somewhere around $1,500,000,000,000 or higher from about $900,000,000,000 this year. We see that the demand for computing capacity still exceeds available supply”at 10:06

Sjoerd Leenart · J.P. Morgan

Asia's AI-driven manufacturing renaissance and domestic demand push regional growth to 4.5%, double-triple the US and Europe.

There is a manufacturing renaissance driven by AI data center demand and by the rewiring of global trade and manufacturing capacity. Korea, Taiwan, Japan and China benefit from it. Policy actions such as China's stimulus and Japan and Korea's market reforms add to growth. India, Vietnam and Malaysia gain from China-plus-one and from domestic demand and a growing middle class. Together this produces regional GDP growth of 4.5%, two to three times the US and Europe, with active corporates and liquid capital markets.

“There's what I call the manufacturing renaissance driven by AI data center demand ... regional GDP growth of 4.5%. And as you know, that's two, three times what we see in The U. S. And Europe.”at 2:20

Joyce Chang · J.P. Morgan

EM fundamentals are stronger than developed markets; portfolio flows are returning to multi-year highs.

Emerging markets are in a stronger position in this cycle, with better external balances, more proactive inflation management, and better fiscal and debt dynamics than developed markets. Central banks and finance ministries have anchored fiscal balances. Strong growth has attracted portfolio flows back to levels not seen in a number of years.

“Emerging markets fundamentals are stronger in the cycle. Better external balances, more proactive inflation management, also better fiscal and debt dynamics when you compare it to the developed markets.”at 7:46

Joyce Chang · J.P. Morgan

China is shifting manufacturing to ASEAN, which gains from US tariff differentials and supports ASEAN exports.

ASEAN countries are a key beneficiary of regional supply chain reconfigurations in the AI tech cycle. China has steadily moved manufacturing to ASEAN, first to cut labor costs and later to adapt to US trade policy, which favors ASEAN over China on tariff differentials. This is an extra tailwind to ASEAN exports, though unevenly. Vietnam and Thailand receive Chinese FDI, and Malaysia and Singapore have seen tech exports soar, while Indonesia and the Philippines have lagged because of smaller manufacturing bases.

“the ASEAN countries are a key beneficiary of regional supply chain reconfigurations in the AI tech cycle. And what we've seen is that China has steadily shifted manufacturing activities to the ASEAN region.”at 13:28

Joyce Chang · J.P. Morgan

China-EU engagement suggests both sides have incentives to negotiate rather than enter an uncontrolled trade fight.

Europe is considering more trade barriers, particularly in autos, after facing setbacks from China's export gains. From the team's discussions, recent China-EU engagement still shows enough incentive on both sides to negotiate rather than move toward an uncontrolled trade confrontation.

“But I still feel that from the discussions we had that the recent China and European Union engagement still suggests that there's enough incentives there for both sides to negotiate rather than to enter an uncontrolled trade [confrontation].”at 15:46

Joyce Chang · J.P. Morgan

Investors shifted from FOMO to selective, quality-focused risk-on; AI capex still lifts markets near term.

Sentiment on equities stayed positive despite higher-for-longer yields, but the tone is now selectively risk-on, with more emphasis on quality and diversification than on a universal bet on AI capex. Fear of missing out drove earlier conferences, while this round raised direct questions about return on investment and productivity gains. The team still expects AI capex plans to keep driving markets higher over the near term.

“There was more of an emphasis on quality and diversification rather than a universal stance that's tied to the AI CapEx boom. So we think that the AI CapEx plans will continue to drive the market higher over the near term”at 6:35

Joyce Chang · J.P. Morgan

Semiconductor capacity growth lags AI demand; Korea and Taiwan lead, and memory is critical.

Semiconductor capacity growth remains well below AI-related demand growth, which makes North Asian hardware suppliers the most profitable part of the chain. Korea and Taiwan keep their leadership in semiconductors and hardware, which puts them at the center of the AI investment cycle. Memory chips have also become a critical component of AI infrastructure, with their share of data center spending growing to nearly 50%.

“This is where we see semiconductor capacity growth that remains well below some of the AI related demand growth. And the leadership is still in place in semiconductors and hardware in Korea and Taiwan, which place them at the center of the AI investment cycle.”at 10:32

Sjoerd Leenart · J.P. Morgan

The US–China summits bring stability but no long-term solution yet; China is too big for investors to ignore.

The US–China meetings were welcome, and two have happened with two more likely this year. They bring a degree of stability but no long-term solution yet. China is about 18% of global GDP, so corporates and investors must decide how to position themselves. Corporates invest for decades and are shifting to more localized operations and local brands, while financial investors can move quickly. Tariffs remain a significant uncertainty for corporates.

“And it brings a degree of stability without really a long term solution. So that's still to be seen.”at 11:38

Sjoerd Leenart · J.P. Morgan

Chinese firms are expanding abroad more than ever, targeting Southeast Asia, Europe and LatAm over the US.

Chinese companies keep expanding abroad without pause, more than ever, because they can no longer find the growth they need at home. Southeast Asia, Europe and Latin America are preferred destinations, while the US sees less Chinese investment. Hong Kong is central because most Chinese companies run their regional and global operations through it. Long-only investors are still cautious because it is unclear whether the Chinese consumer and property market have turned the corner, while hedge funds are active given the liquidity of China and Hong Kong markets.

“When we look at the behavior of Chinese companies, they continue to go abroad without any pause. And more so than ever, looking for growth that they can no longer find domestically.”at 12:59

Joyce Chang · J.P. Morgan

China's export surge hits Europe harder than the US, with Germany most exposed; Europe's growth is also hampered by structural issues.

China's exports are absorbing manufacturing output that domestic demand cannot absorb. Germany has greater direct competitive overlap with China than the US does across autos, manufacturing, machinery, chemicals and other sectors. That makes China's export gains more problematic for Europe and a larger drag on German growth. Europe also faces aging populations, high energy costs and rising debt, so it needs faster reforms to lift productivity and growth, and it is considering more trade barriers, especially in autos.

“raising concerns that Europe could be the main casualty from the China shock. We are seeing that [Germany] has greater direct competitive overlap with China than The US does across autos, manufacturing, machinery, chemicals”at 15:09

Sjoerd Leenart · J.P. Morgan

Money is gradually reallocating to Asia, but flows are uneven and move in and out faster than in the US.

A gradual reallocation of money towards Asia is happening, but it is very uneven. Flows move in and out much more than they do in the US. Korea and Taiwan were early darlings this year but have come off significantly. China and Hong Kong have struggled after a strong 2025. Global flows into India have restarted cautiously after about €60B of outflows over two years. Southeast Asian markets show wide dispersion, so investors must reallocate with agility.

“So I do think a gradual reallocation of money towards that region is happening. But it is, as you alluded to earlier, very uneven. The flows tend to move in and out much more so than what I observe in The U. S.”at 16:18

Joyce Chang · J.P. Morgan

AI models can find and exploit vulnerabilities; time-to-exploit has shrunk from months to hours.

Cyber attacks are growing, and AI models can both discover and exploit vulnerabilities. The time needed to exploit a vulnerability has gone from months to hours, and may soon be minutes. Because global AI systems are only as strong as their weakest link, cyber resilience, crisis preparedness and oversight of critical third-party providers will need more focus.

“we see that these models can both discover and exploit the vulnerabilities and that what used to be a matter of months has come into hours and soon to minutes as far as the days to exploit these cyber vulnerabilities.”at 17:34