Asia: the chip exposure you own, and the Japan you may be missing

We see a stronger long-term case for adding Japanese equities than for adding more exposure to the Asian chip boom. Japan's companies are under pressure to use their cash better, sell unproductive holdings and return money to shareholders. In Korea and Taiwan, much of the investment case already depends on exceptionally strong chip profits continuing.

Korean shares could still rebound sharply after their recent fall. That prospect deserves a separate decision from how much of your savings you want tied to semiconductors for the next ten years.

Key points

• Taiwan and South Korea make up almost half of the emerging-markets index. An emerging-markets ETF can add substantial chip exposure to a portfolio already heavy in US technology.
• Internet-era suppliers grew rapidly, then faced falling margins and share prices when the investment boom slowed. Several recovered their profits years before their shareholders recovered their money.
• Japan has the highest estimated return among the major equity markets in copio's assumptions: 8.6% a year in euros, with currency exposure included.

Look inside the emerging-markets fund

At the end of October 2026, Taiwan made up 28.9% of MSCI Emerging Markets and South Korea 19.7%. Information technology accounted for 41.8% of the index. Three chipmakers, TSMC, Samsung Electronics and SK hynix, represented 28.1% of its value, excluding Samsung's preferred shares. 1

These are large positions inside their home markets too. TSMC made up about 55% of MSCI Taiwan. 1 Samsung and SK hynix together accounted for roughly 67% of the uncapped MSCI Korea index, including Samsung's preferred shares. 1

You may already own other parts of the same supply chain. Nvidia designs chips, TSMC manufactures them, and memory producers supply the high-bandwidth memory used alongside them. Their businesses differ, but they depend on many of the same customers continuing to spend on AI infrastructure.

Adding an emerging-markets ETF can therefore increase the AI exposure you already have through a world or Nasdaq fund. The country names change while the underlying spending cycle remains closely connected.

What the internet build-out tells us about chips

The internet needed processors, routers and fibre networks before most of its commercial uses had developed. Suppliers sold into a rush to build capacity, and investors priced their shares for years of rapid growth. When customers cut investment, suppliers were left with excess stock and factories that were expensive to run below capacity.

Intel's results show the cycle. 13

Revenue more than doubled between 1995 and 2000, while gross margin expanded by almost eleven percentage points. In 2001, revenue fell 21% and the margin dropped by thirteen points. New capacity, lower demand and price competition squeezed the return on each chip sold.

The business recovered. By 2005, Intel's revenue exceeded its 2000 level and its margin was close to the earlier peak. Yet its December 2005 share price remained 67% below its highest month-end price in 2000. 14 The price paid during the boom still weighed on the shareholder's result after the company returned to growth.

Cisco shows the same risk in network equipment. It recorded a $2.2bn excess-inventory charge in 2001. By 2005, revenue had risen above its 2001 level and gross margin reached 67%, while its share price remained 78% below its highest month-end price in 2000. 14 15 Strong technology demand did not ensure a good return from the peak valuation.

AI chip suppliers face a similar investment cycle. Demand initially grows faster than capacity, giving scarce products high prices and margins. Those profits finance new factories and attract competitors. If supply catches up while customers slow their spending, both margins and the valuation investors give those profits can fall.

SK hynix reported an 80% operating margin in Q3 2026, up from 76% in Q2, and approved ₩54.3tn for two new factories in August. 3 3 4 China's CXMT increased its share of global DRAM revenue from 4% to 10% in the year to the second quarter of 2026, according to Counterpoint Research, and plans further expansion. 5 5 Capacity takes years to arrive, so the current profits can last longer than a short-term seller expects.

The history supports testing today's share prices against lower margins once that supply arrives. It gives us no fixed date for the turn. Advanced AI memory has manufacturing barriers and longer customer commitments, while the largest buyers have substantial cash-generating businesses. We would still require an investment case that works after exceptional chip profits ease.

Taiwan needs its own analysis

TSMC manufactures chips designed by other companies. Its leading-edge process technology and relationships with customers create a different business from commodity memory production. It should be assessed on its own margins, capacity use and investment needs.

Its gross margin rose from 59.5% to 66.4% over the year to the third quarter of 2026. Management expects overseas factories to reduce gross margin by about 2 percentage points in 2026, widening to 3-4 points in the next few years. 6 6 High demand for advanced chips supports the business, while expansion outside Taiwan raises costs.

Because TSMC represents more than half of the Taiwan index, buying that index gives one company's economics an unusually large role in your return. We would account for that concentration explicitly when combining it with American technology holdings.

Why Korea can rebound from here

The KOSPI fell from 9,114.55 on 22 June 2026 to 5,554.22 on 15 October, a decline of 39%. It remained about 32% above its end-2025 level. 2 2 Taiwan had fallen much less, sitting about 5% below its 6 October record close on 16 October. 7 7

Korean valuations reflect a very different starting point after that fall. MSCI Korea traded at 4.0 times expected earnings at the end of October, down from 4.95 a month earlier. 1 If chip profits stay high for longer than investors expect, or fall less than feared, share prices could recover quickly.

We see scope for a Korean rebound, with limited evidence that today's chip margins can be sustained across a full decade. A short-term position would need its own exit conditions and loss limit. For a long-term allocation, we would use a range of earnings that includes a weaker part of the cycle.

Japan is changing how companies use their money

Japan spent more than three decades recovering from its 1989 equity bubble. The Nikkei 225 price index regained that peak in February 2024, after deep falls and several recoveries. Dividends and currency moves change an investor's return; the chart shows the yen price index. 16

Nikkei 225 prices exclude dividends. OECD monthly yields through September 2026; World Bank / IMF annual CPI inflation through 2025. Market data · Inflation data.

Years of weak inflation and falling yields left companies operating in an economy with little nominal growth. That environment is changing alongside the reforms to corporate governance. In March 2024, the Bank of Japan ended negative policy rates and its yield-curve-control framework as it judged the wage-price cycle stronger. 17 Higher wages and prices create room for revenue growth, while rising financing costs make unproductive uses of capital harder to justify.

Japanese companies have historically kept large cash balances and held shares in business partners. Those arrangements helped preserve relationships and management control, often at the expense of returns to outside shareholders. Listed companies still held about ¥118tn in cash and deposits in the 2026 research cited here. 10

In March 2023, the Tokyo Stock Exchange asked companies to explain how they would earn enough to cover their cost of capital. In everyday terms, managers had to explain why the money tied up in their business could earn a competitive return.

By February 2026, 93% of Prime-market companies had disclosed a response, including companies still considering measures. The share trading below book value fell from 50% in July 2022 to 27% in March 2026. 8 Book value is the accounting value of assets after liabilities; a price below it can signal that investors expect poor returns from those assets.

The changes reach shareholders in several ways.

Companies are returning more cash. Announced buybacks reached a record ¥22.3tn in fiscal 2025. 9 When shares are repurchased at a sensible price, remaining shareholders own a larger share of future profits. The company needs to complete the purchases for that benefit to arrive.

Long-standing shareholdings are being unwound. Business corporations owned 17.7% of the listed market in fiscal 2025, down from 30.1% in 1990. City and regional banks' holdings fell from 15.7% to 1.7% over the same period. 11 Selling those stakes releases money for investment, debt repayment or shareholder distributions. It also reduces the protection that friendly shareholders give management.

Outside owners have more influence. In March 2026, Toyota's group raised its offer for Toyota Industries from ¥18,800 to ¥20,600 a share after pressure from an activist investor. 12 That is a concrete example of outside shareholders seeking a better price for an asset.

We expect continued pressure to improve capital use to support Japanese equity returns. It gives investors a source of earnings and cash distributions that is broader than the AI spending cycle.

Profitability and the yen still need attention

Higher valuations have arrived faster than some operating improvements. The share of Prime-market companies earning less than 8% on equity fell only from 47% to 43% in the exchange's comparison. 8 Managers still have work to do on the underlying businesses. Returning excess cash helps, but it cannot indefinitely compensate for weak operations.

For a euro investor, the currency can also dominate several years of progress. A euro bought about ¥114 in October 2016 and around ¥177 in October 2026. MSCI Japan returned 8.76% a year in euros over the ten years to October 2026, compared with 12.74% for MSCI World. 1 Yen depreciation reduced the euro value of Japanese holdings, alongside the effect of differences in companies and valuations.

A recovery in the yen would help an unhedged euro investor. Further weakness would hurt. Currency hedging changes that exposure and has its own cost, which belongs in the decision about how to hold Japan.

What copio's assumptions show

Japan has the highest return estimate among these markets. Asia ex-Japan sits below world equities on expected return and above it on volatility. The assumed correlation between Asia ex-Japan and emerging-market equities is 0.97, so combining those two broad funds offers little additional diversification in this model.

We have no standalone Korean or Taiwanese return estimate. The regional figures include other countries and cannot establish a return ceiling for either chip market.

Why we favour keeping the yen exposure

Our unhedged estimate includes a recovery in the yen over the long term. For a euro investor, a stronger yen raises the euro value of Japanese shares. Hedging the position to euros removes that potential gain as well as the loss from a weaker yen.

Hedging also includes the interest-rate difference between the two currencies. When euro short-term rates exceed Japanese rates, a euro investor hedging yen can receive a return from that difference, before fees and trading costs. Our assumptions still favour the unhedged position: expected currency recovery and the equity-currency relationship more than offset the modelled benefit of hedging. The one-percentage-point return gap includes these effects together.

The currency can also offset part of the equity move. A weaker yen can raise exporters' yen earnings while reducing the euro value of the holding; a stronger yen can work in the opposite direction. The model assumes enough of this offset to give unhedged Japan lower volatility than hedged Japan. That is a long-term relationship, and individual sell-offs can behave differently.

We favour keeping yen exposure within a diversified long-term allocation. Money needed for a near-term euro expense has a different currency constraint. All figures are estimates before product costs and personal taxes.

Choose the exposures separately

Add up semiconductor exposure across your world, US, emerging-market and specialist funds. Then decide how much of your portfolio should depend on that industry. A broad fund can still leave you with a concentrated economic bet.

Give Japan a separate weight and make a conscious decision about the yen. The higher return estimate supports allocating to Japan, while the uncertainty around reform, currency and valuation argues for holding it alongside other markets.

For Korea, distinguish money committed to a possible rebound from money allocated for ten years. The price you pay, the profits you assume and the time you can wait all affect the outcome.

Information only; no personal investment advice or recommendation of individual companies. Long-term estimates can differ substantially from realised returns. Past performance does not guarantee future results.