Generated by Codex with GPT-5

A boom with a hollow center

The article argues that north-east Asia’s most advanced economies are enjoying a headline boom that hides a deeper industrial problem. Taiwan, South Korea, and Japan look strong because artificial intelligence has created huge demand for chips, servers, and the machinery and materials needed to make them. Taiwan’s economy is growing at a pace more typical of a fast-developing country than a rich one. South Korean corporate profits have surged. Japanese exporters are also doing better than the broader economy.

But this success is narrow. Once semiconductors, AI servers, and related supply-chain goods are stripped out, Taiwan’s exports have fallen sharply since 2022, South Korea’s non-AI exports have stagnated, and Japanese manufacturing is weakening. The article’s core point is that the AI boom is not proof that the old export model is healthy. It is concealing the fact that much of the region’s wider industrial base is losing ground.

China is the main reason. For decades, Japan, South Korea, and Taiwan supplied high-value parts and capital goods to Chinese factories, which assembled finished products for the world. That relationship has changed. China now competes directly in cars, chemicals, batteries, machinery, and many other sectors that once sustained its richer neighbors. It is not merely taking over low-end manufacturing. It is moving across the value chain and crowding out firms that used to sell into, or around, China.

The pressure shows up in trade balances and factory floors. Taiwan’s long-running goods surplus with mainland China has flipped into deficit. Japan’s deficit with China has reached record levels. South Korea has been rescued recently by chip exports, but its broader industrial position has weakened. Behind the AI-driven glow are furloughed machinery workers, struggling battery makers, and chemical producers hit by Chinese overcapacity.

The risks of narrow strength

Specialization is not automatically bad. Rich economies often shed lower-value manufacturing and focus on sectors where they are globally excellent. The problem is that north-east Asia’s specialization is becoming too concentrated, too export-dependent, and too exposed to the decisions of China and America.

Chips and other AI-linked goods now make up more than 40% of South Korean exports and roughly 80% of Taiwan’s. The Economist calculates that the entire rise in the region’s industrial output since 2019 can be attributed to AI-related production. That makes the region vulnerable to any slowdown in AI spending, any downturn in the chip cycle, or any geopolitical disruption that affects the technology supply chain.

Governments are responding by doubling down. South Korea plans \$530bn in support for chipmaking over two decades. Taiwan offers tax breaks to its chip giants. Japan is pouring public money into strategic industries, including a \$16bn bet on Rapidus, a project meant to revive cutting-edge domestic semiconductor production. These policies may strengthen national champions, but they also risk making already concentrated economies even more dependent on a single global technology cycle.

The external vulnerability is just as serious. Taiwan and South Korea rely heavily on exports as a share of GDP, and Japan has become steadily more export-oriented since its late-1980s bubble burst. Their customers are concentrated too. Taiwan’s foreign sales, dominated by chips, are heavily tied to America and China. That leaves the region squeezed between China’s industrial rise and America’s tariffs, subsidies, and pressure to move production onshore.

The domestic weakness underneath

The article argues that the best answer is not more industrial policy but stronger domestic economies. Taiwan, South Korea, and Japan consume too little of what they produce. Private consumption is about 53% of output in Japan and closer to 40% in South Korea and Taiwan, far below the rich-country average. This is especially striking because aging societies should normally consume more, not less.

Low consumption is not a cultural mystery. It is the legacy of development models built to favor production, exports, savings, and large manufacturers. Governments steered capital toward exporters and protected favored workers inside big firms. That helped these countries catch up when they were poorer. Now it produces warped labor markets and weak demand.

The result is a split workforce. Employees in elite exporting firms enjoy relatively secure jobs, high wages, and better benefits. Workers in small firms, temporary jobs, or irregular employment earn much less and face greater insecurity. In South Korea and Japan, irregular workers make far less than regular workers. In Taiwan, the electronics sector pays a large premium, while many graduates struggle to find attractive jobs at home.

Welfare systems reinforce the imbalance. Taiwan and South Korea spend far less on pensions than the OECD average, and pension benefits in Japan and South Korea depend heavily on the kind of job a person held. Elderly poverty is high across the region. When households cannot count on stable wages, decent pensions, or reliable support, they save more and spend less. That deprives domestic firms of demand and leaves the economy leaning harder on exports.

The article’s warning is that the AI boom gives north-east Asia a chance to reform, but also an excuse not to. If governments use today’s chip profits to preserve old structures, they will make their economies narrower and more fragile. A better path would loosen labor markets, strengthen safety nets, raise household incomes, let weak firms fail, and reduce barriers to trade beyond the China-America axis. The takeaway is not that the AI boom is fake. It is that real technological success can still be dangerous if it hides the need for broader renewal.