The AI Capex Boom Significantly Spills Over to Asia’s Technology-Export Economies

Overview

The report analyzes how the AI investment boom spills over to Asian economies through data-center equipment, servers, semiconductors, storage, networking equipment, and packaging and testing links in the value chain. Goldman Sachs argues that global AI capital expenditure could exceed US$1 trillion in 2026, with hardware spending above US$600 billion, and that Asia’s manufacturing-and-assembly system is the main beneficiary. Taiwan, with advanced logic-chip and AI-server manufacturing capabilities, is the most prominent beneficiary at both regional and global levels. South Korea benefits through high-bandwidth memory and high-capacity NAND, while Malaysia, Singapore, Vietnam, and Thailand also contribute via packaging and testing, electronics manufacturing, and data-center investment.

Core views

The core thesis is: first, AI-related exports are already and will continue to materially lift growth in Asian technology-export economies, with Taiwan contributing the most, followed by South Korea, while several Southeast Asian economies receive moderate benefits. Second, AI demand amplifies growth divergence across the region, with technology-intensive economies and sectors leading while non-technology sectors, energy-importing economies, and some South Asian markets face relative headwinds. Third, technology trade surpluses improve Taiwan and South Korea current accounts and could put upward pressure on their currencies. Fourth, North Asian markets may continue to outperform in the near term; the report mentions market strategy commentary calling for long positions in MSCI Korea and Taiwan and short positions in MSCI India, Philippines, and Thailand.

Analysis framework

The report uses multiple methods to estimate the contribution of AI-related technology exports to Asian real GDP growth: first, using industry GDP or gross value added data to directly assess the contribution of manufacturing subsectors such as electronics, computers, and optical products to growth; second, estimating contribution of domestic value added in technology exports using the value using technology exports and 2022 OECD TIVA domestic VA share along with export price indices; third, using net real technology exports as a cross-check; fourth, for economies with richer data such as Taiwan and South Korea, using actual tech value-added or chip shipment proxy indicators to supplement the estimate. For the 2026 forecast, the report runs regressions between each economy’s technology exports and U.S. hyperscaler capex, then uses 2026 consensus capex estimates to infer export contribution.

Methodology notes

• Macroeconomic Growth EstimationIndustry GDP Contribution Method

  Measures spillover from technology production to real GDP growth using manufacturing subindustry contributions.

  The advantage is that it uses official national real GDP or gross value added data. The downside is that the industry scope is often broader than AI-related products, and data may be revised in later periods.

• Trade and Value-Added EstimationDomestic Value Added from Technology Exports Method

  Estimates actual domestic value added contribution by using technology export values, 2022 OECD TIVA domestic VA shares, and export price indices.

  The advantage is that export data are more timely and directly linked to overseas capex. The downside is that domestic VA shares are lagging, supply chains evolve quickly, and export price index reliability is limited.

• Cross-validationNet Real Technology Exports Method

  Estimates net value creation in exports by deducting real technology imports used for domestic demand from real technology exports.

  This approach is close to the domestic value-added method but requires deflating both exports and imports, so cross-country differences are larger; therefore the report does not present it as the primary estimate.

• Forecast ModelRegression of Technology Exports on U.S. Hyperscaler Capex

  Uses each economy’s sensitivity of technology exports to U.S. hyperscaler capex to estimate 2026 contributions.

  The report stresses the sample is only about 13 quarters and technology changes are fast, so 2026 estimates should be treated as approximations.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Taiwan Economy and TWD

  Core beneficiary of AI hardware and advanced logic chip supply chains

  Strengths

  TSMC is a leader in advanced logic chips, and Taiwan also has AI server design and assembly capabilities. In 2025, technology exports contributed around 5.9 percentage points to real GDP growth.

  Weaknesses

  Growth is highly concentrated in the technology manufacturing segment, which may create K-shaped recovery dynamics with limited job and wage spillovers.

  Comparison

  Compared with other Asian economies, Taiwan is the most AI-leveraged economy in both the region and globally, according to the report.

  Risks

  Technology cycle downturn, supply-chain relocation, revision of price indices and GDP data, and geopolitical risk.

• South Korea Economy and KRW

  Supplier of high-bandwidth memory and high-capacity NAND, benefiting from AI server and data-center demand

  Strengths

  SK Hynix and Samsung have critical positions in high-bandwidth memory, NAND, and data-center storage; 2026 technology-export contribution is estimated at around 1.0 percentage point.

  Weaknesses

  The magnitude of benefit is lower than Taiwan, and growth divergence between manufacturing and non-manufacturing sectors is pronounced.

  Comparison

  South Korea benefits more than most Asian economies, but less than Taiwan.

  Risks

  Rising energy prices, storage-cycle volatility, and weak non-tech sector growth.

• Malaysia Economy and MYR

  Beneficiary through packaging and testing and data-center investment

  Strengths

  Malaysia has a meaningful share of global semiconductor packaging and testing, and data-center investment is relatively meaningful versus GDP; the report is constructive on MYR.

  Weaknesses

  It has a smaller direct role in core AI hardware segments than Taiwan and South Korea, and part of its tech exports may not be AI-related.

  Comparison

  Among Southeast Asian economies, Malaysia shows relatively clear benefits, with dual support from technology exports and data-center investment.

  Risks

  Energy shocks, external demand volatility, and limited margin/value capture in packaging and testing.

• Singapore Economy

  Beneficiary in semiconductor equipment, chip manufacturing, packaging and testing, and electronics exports

  Strengths

  Participates in wafer manufacturing, integrated circuit design, packaging and testing, and semiconductor equipment chain links; electronics exports have strengthened recently.

  Weaknesses

  Relative to its size, data-center investment and manufacturing tailwinds may be smaller than in Malaysia and Taiwan.

  Comparison

  Benefits more than low AI-exposure economies, but technology-export growth impact is below some Taiwan and Vietnam estimates.

  Risks

  Open economy means high sensitivity to global trade, energy prices, and the technology cycle.

• Vietnam Economy

  Beneficiary via electronics assembly and emerging packaging and testing

  Strengths

  Strong base in consumer electronics assembly, with expansion toward outsourced assembly and testing; the table shows a 2026 technology-contribution estimate of 3.4 percentage points.

  Weaknesses

  The report notes that most tech exports may not be AI-related, so domestic value capture could be limited.

  Comparison

  Export-based estimates show a larger contribution, while the industry-GDP method suggests a more muted impact.

  Risks

  High share of non-AI exports, dependence on imported intermediate goods, and errors in price and value-added estimation.

• China Economy and CNY

  Large-scale electronics manufacturer and participant in traditional semiconductor chains, but with lower direct exposure to leading-edge AI supply chain nodes

  Strengths

  Massive electronics manufacturing scale, supplying large volumes of low-end passive components, materials, and domestic AI-chain parts; the report is constructive on CNY.

  Weaknesses

  Most advanced-logic chip core nodes are largely outside China, and direct growth impact is more subdued due to U.S.-led export controls.

  Comparison

  Absolute technology manufacturing scale is very large, but AI-investment spillover growth elasticity is lower than Taiwan and South Korea.

  Risks

  Export controls, geopolitics, traditional chip-cycle dynamics, and external demand shifts.

Key data

• Global AI Capital ExpenditurePotentially above US$1.0 trillion in 2026The five major U.S. hyperscalers’ 2026 capex consensus is about US$750 billion; global AI capex is higher than that.
• AI Hardware SpendingOver US$600 billionThe report says 60%-70% of AI-related investments are used for servers, data storage, and networking equipment.
• Incremental Asian-Related Electronics Exports to the U.S.Over US$200 billion annualizedFrom 2022 to early 2026, Asian-related electronics exports to the U.S. increased substantially.
• Taiwan Technology Export Contribution to 2025 Real GDP Growth5.9 percentage pointsThe report says Taiwan is a leading manufacturer of high-end logic chips and AI servers, making it the most AI-sensitive economy.
• Taiwan Predicted Technology Export Contribution to 2026 Real GDP Growth4.5 percentage pointsAssumes AI investment continues to grow quickly, though at a slower pace.
• South Korea Technology Export Contribution0.8 percentage points in 2025, 1.0 percentage points in 2026South Korea is a key supplier of high-end storage chips, high-bandwidth memory, and NAND.
• Taiwan 2025 Current-Account SurplusClose to 20% of GDPThe report says the 2026 tracked value may exceed 2025.
• South Korea 2025 Current-Account Surplus6.6% of GDPThe report expects that even with higher oil prices, South Korea’s surplus may still reach a double-digit share of GDP in 2026.
• Data-Center Investment Contribution to Malaysia GrowthSlightly above 1 percentage point in 2025, potentially above 1 percentage point in 2026Thailand may also exceed 1 percentage point in 2026, supported by recent approvals of major projects.

Impact & implications

The macro impact of the AI investment boom is not evenly distributed across Asia. Taiwan and South Korea benefit most directly through advanced chips, server, and storage links, with an upward revision to growth forecasts and current-account outlooks; Malaysia, Singapore, Vietnam, and Thailand also receive supplementary support through packaging and testing, electronics manufacturing, and data-center investment. China, while constrained in direct exposure to frontier AI chain segments, still participates at scale through low-end components and traditional semiconductor chain segments, though direct growth impulse is more muted due to export controls affecting leading-edge nodes. From an investment perspective, the report is more constructive on North Asian technology-export economies and related currencies, while noting that strong headline growth does not necessarily translate into equal spillovers to jobs, wages, or inflation.

Risks

• AI-related estimates are highly approximate and sensitive to rapid technology changes, supply-chain relocation, lagged domestic value-added data, and uncertainty in price indices.
• Official GDP, industry value-added, and price statistics may be revised in the future.
• AI hardware demand may slow, leading to lower-than-expected technology-export and server-demand growth.
• Rising energy prices will pressure energy-importing economies and could offset part of the AI export tailwind.
• Growth could become K-shaped, with strong tech-sector performance but limited spillovers to jobs, wages, investment, and inflation.
• Service exports may be adversely affected by the AI investment boom; the report says this will be covered separately.
• Trade tariffs, export controls, and geopolitical tensions could reshape the semiconductor and server supply-chain footprint.

What to watch

• Whether 2026 capex for the five major U.S. hyperscalers stays close to the US$750 billion consensus.
• Whether global AI capital expenditure remains above US$1 trillion and whether hardware spending stays in the 60%-70% range.
• Growth rates of Taiwan exports of servers and high-end chips to the U.S., and South Korea exports of high-bandwidth memory and NAND.
• The pace at which data-center investment projects are implemented in Malaysia and Thailand, and their actual GDP impact.
• Current-account surpluses and TWD, KRW, MYR, and CNY performance in Asian technology-export economies.
• Whether growth divergence between technology and non-technology sectors, and between exports and domestic demand, is converging.
• Energy-price shocks weighing on India, Thailand, Indonesia, the Philippines, and other low-AI-exposure or high-fuel-import economies.
• Revisions and updates to key data series, including OECD TIVA, industry GDP, and export price indices.