An estimate issued by Taiwan's General Accounting Office on 31 July shows that real GDP grew by 13.72 per cent in the first half of 2026, the strongest performance in the first half of 50 years, and by 12.92 per cent in the second quarter, the fastest increase in the second quarter in 39 years. The second quarter results were 2.09 percentage points higher than projected in May, not only because of strong AI-related exports, but also because private investment and consumption were better than expected.
This set of figures continues the rapid growth in the first quarter. Preliminary statistics released at the end of May by the General Directorate show that real GDP increased by 14.55 per cent over the same period during the quarter, real exports of goods and services increased by 35.76 per cent and manufacturing value added by 26.18 per cent, driven mainly by semiconductors, computers, electronics and optical products. Global cloud manufacturers have expanded the AI infrastructure to channel demand for advanced chips, servers, networks and related parts to the Taiwan production system.
A broader base of growth emerged in the second quarter. Through CS, CS officials indicated that exports of goods in United States dollars increased 43.73 per cent over the same period; private consumption grew by 5.88 per cent, the fastest in 11 quarters. An active stock market has brought demand for financial services and wealth effects, as well as the resumption of car sales and increased out-of-country tourism, which together boost consumption. This means that growth in the current round has not yet been completely confined to export statements, and that part of the income and asset price effect has been passed on to domestic demand.
However, 13.72 per cent also contained clear base and periodicity factors that could not be mechanically extrapolated. Compared to the same period last year, the time of delivery, inventory and export of AI equipment could cause quarterly fluctuations. Taiwan’s small economy, with a high concentration of technological and technological manufacturing, is likely to increase significantly when external orders are growing rapidly; and the reverse impact is even greater when customers absorb inventory or capital expenditure.
How does AI demand drive exports, investment and consumption simultaneously?
Taiwan is at several critical points in the AI hardware supply chain. The Crystal Circle Generation is responsible for the advanced process, the containment test connects the computing chip to high bandwidth storage, the electronic manufacturing service enterprise assembles servers, and the spare parts plant provides dissipation, power, housing and network equipment. The increase in the budget of a data centre by cloud manufacturers is not the creation of a chip order, but it spreads along this industry network.
With the increase in orders, enterprises need to expand their plants, purchase equipment and upgrade their R&D inputs to form fixed capital. Production and exports are accompanied by profits, wages and bonuses, and improved performance of listed companies pushes up stock markets, further affecting household consumption. The increase of 5.88 per cent in private consumption in the second quarter, which indicates the need to channel domestic demand through such channels as employment, asset and service transactions, is a more solid part of the current round of data than “single-to-export”.
Import growth should not simply be viewed as negative. Taiwan ' s production of high-end electronics requires the import of energy, raw materials and equipment, and export booms often accompany rising imports. The quality of growth should be judged by whether imports are consumer goods or capital equipment, how much added value is added to final exports, and whether investment by enterprises improves future capacity. Only a single-month comparison of exports and imports can easily miss the true way in which the supply chain economy operates.
AI has also changed the traditional electronic cycle. The demand for smartphones and personal computers is more dependent on switchovers, while the AI data centre is dominated by capital expenditure by a small number of large clients, with larger individual projects and faster technological overlaps. As a result of this higher growth, Taiwan enterprises are also exposed to the risk of customer concentration and product switching. Changes in a generation of chips or server structures may quickly benefit some suppliers and may result in a faster depreciation of the old production line.
The most important thing behind high growth is concentration, not numbers. Look.
The first risk is a slowdown in global AI capital spending. Current demand is based on the belief of cloud producers that future models and applications will generate sufficient income. If the data centre returns less than expected, the client may delay the order. As Taiwan ' s export and manufacturing growth is highly exposed to this investment cycle, the high rate of growth of 13.72 per cent itself means that next year it will also face a higher comparative base.
The second risk is energy and energy constraints. Advanced manufacturing requires stable power, water, land and engineering. When demand rises rapidly, firms can expand their production; if infrastructure fails, costs and delivery times rise. In order to serve overseas clients, some enterprises have also increased offshore investment, which has helped to move closer to markets and spread geographical risks, but has the potential to change capital formation and supply chain division within the island.
The third risk arises from trade and geo-policy. Advanced chips are both commercial products and subject to export controls and national security policies. Changes in tariffs, technical licences or customer area restrictions may affect the order path. The more the enterprise concentrates on high-end AI hardware, the more it needs to prepare capacity and compliance programmes for different policy scenarios.
Thus, while Taiwan ' s growth round is positive, the more important question is whether the AI order can be translated into sustained productivity gains. If prosperity leads to more R & D, talent and local service innovation, the economy gains beyond the hardware cycle; and if growth depends mainly on a few export and asset prices, volatility is magnified. 13.72 per cent is an extremely strong periodic report card and a stress test of industrial diversification, energy supply and risk management capacity.