Taiwan’s AI boom is spreading beyond TSMC to the wider economy

Taiwan’s growth is broadening as domestic demand picks up

Taiwan’s economy continues to grow strongly. In the second quarter of 2026, real GDP rose 12.93% year-on-year. The combined contribution of private consumption and fixed investment to growth also increased from 4.39 percentage points in the first quarter to 5.63 percentage points, showing that domestic demand is providing greater support. As AI-related companies increase investment and distribute profits, the economic benefits of this boom are gradually spreading to other industries.

Related article: Taiwan Outlook 4Q26 | From chips to memory, Taiwan can’t stop winning

Figure 1: Taiwan’s year-on-year real GDP growth

Table 1: Beyond exports, private consumption and fixed investment are driving Taiwan’s economy

Corporate capacity expansion is an important channel through which AI demand spreads to other industries. In the second quarter, domestic fixed asset purchases by Taiwan’s manufacturing sector, excluding land, reached NT$956.3 billion, up 53.7% year-on-year. Machinery and miscellaneous equipment accounted for 77.1%, while buildings and construction works accounted for 22.6%. The Ministry of Economic Affairs noted that demand for AI and data centres was driving capacity expansion across the semiconductor, memory and server supply chains. Building factories, purchasing equipment and improving supporting infrastructure benefit equipment suppliers, engineering contractors and business service providers.

Record tax receipts are giving Taiwan’s government room to share the gains of the AI boom

The AI boom is also reaching the wider economy through the public finances. According to the Ministry of Finance, net tax revenue for January to August 2026 reached NT$3.07 trillion, up 25.2% from a year earlier and 78.0% of the full-year budget with four months to go. The securities transaction tax was the largest source of the gain. It raised NT$452.7 billion, up 170.8%, as heavy trading in AI-related shares lifted market turnover, and it has already reached 181% of its full-year budget. Although it is only 15% of collections, it accounts for 46% of the year-on-year increase (Table 2). Corporate income tax is the largest line, at 26% of collections, and rose 24.9%. Because it is largely collected on the previous year’s profits, we believe 2026’s record earnings should feed through more fully into later years’ collections.

Table 2: Taiwan net tax revenue, January to August 2026

The government is preparing to share part of the gains. On 20 August, the Cabinet approved a NT$3.93 trillion central government budget for 2027, which balances revenue and spending without new borrowing and goes to the Legislative Yuan for review. Social-welfare spending rises 41.1% to NT$1.19 trillion, and funding for local governments rises 16.6% to NT$1.36 trillion, mostly through centrally allocated tax revenue of NT$972.5 billion. Local governments also share in the gains: their net tax revenue rose 63.5% in January to August, helped by a 2026 change that gives them a larger share of business tax.

The budget also includes a universal cash payment of NT$10,000 per citizen. A government official said the payment could be paid before the Lunar New Year in February if the budget passes by year-end. The government’s stated aim is to share the gains of growth. If approved, the budget would carry them to households through welfare spending and the cash payment, and to regions through local government funding.

Power and cooling demand shows how AI investment reaches suppliers beyond chips

The public finances are one route by which the AI boom spreads. The other runs through companies. Demand for AI infrastructure comes both from capacity expansion within Taiwan and from global procurement, and the suppliers closest to that spending are the ones that build and equip it. AI computing capacity cannot come online without the infrastructure around it.

NVIDIA’s GB300 NVL72 brings together 72 GPUs in a single rack and uses liquid cooling. As computing capacity becomes more concentrated, power supply and cooling requirements also increase: data centres must deliver sufficient, reliable power to servers and remove the heat generated by computing. Even when the chips have arrived, computing capacity cannot come online as scheduled if the supporting infrastructure is not ready. Customers increasing their chip purchases must also invest in power supply, cooling and related facilities.

As we noted in NVIDIA update, assessing the value of an AI system requires considering how much computing capacity it can deliver within a limited power supply, alongside the total cost of equipment, electricity, cooling and maintenance. This also explains why supporting equipment is becoming an important part of AI investment: improvements in chip performance still require corresponding power and cooling capacity to deliver their benefits in actual operations.

Delta Electronics shows how Taiwanese companies are meeting this wave of AI infrastructure demand. Its power products convert and supply electricity, while its cooling products remove the heat generated by operating equipment. The company also integrates these products into systems ready for deployment, broadening the range of products and services it supplies. As global investment in data centres increases, customer demand for power supply, cooling and system integration is rising alongside it. This brings Delta more business and demonstrates how Taiwanese companies beyond chip manufacturing can benefit from the AI boom.

This demand is already reflected in business results. In the first half of 2026, data centre-related business accounted for more than half of Delta’s group revenue. Its Infrastructure segment, which includes data centre infrastructure solutions, recorded year-on-year revenue growth of 63% and segment profit growth of 142%. Profit growing faster than revenue indicates that the segment has improved its profitability while taking on more business. Demand for power supply and cooling has translated into business results for suppliers of supporting equipment.

To meet rapidly growing demand for its products, Delta also needs to increase its own investment in advance. The company’s capital expenditure reached NT$27.5 billion in the first half, with capacity expansion spanning Taiwan and overseas markets. As factory construction progresses in the second half, construction payments and investment in equipment will also increase. Customer demand to expand AI computing capacity is therefore also driving Delta’s spending on production equipment and construction.

The systems business also requires more staff. Large equipment needs corresponding production space, while technical support and services after delivery require the company to build customer-facing teams. These investments will continue as the business expands, extending AI demand from equipment supply to engineering, technical and service work.

Other Taiwanese companies are participating in different parts of this wave of construction. Fortune Electric is expanding transformer capacity in response to global power grid upgrades and demand from AI data centres. CTCI’s project for TSMC’s reclaimed water plant at the Southern Taiwan Science Park covers investment, design, construction and long-term operation, supplying the water needed for wafer manufacturing. The former provides power equipment, while the latter provides engineering and water services. For investors in the Taiwan equity market, demand arising from the technology industry’s expansion is also creating opportunities for companies with these different areas of expertise.

Telecoms operators are turning AI infrastructure demand into ongoing service revenue

Businesses also need data centres, networks and day-to-day services to keep their computing capacity running. This creates ongoing demand for telecoms operators, which already run data centres, networks and cloud services for corporate customers.

Chunghwa Telecom is one example. Its business spans mobile communications, fixed-line services and broadband, while also providing businesses with data centres, network connectivity and cloud services. It can help customers build and connect their information systems, as well as provide the space and services needed to operate their equipment, enabling the telecoms sector to participate in this wave of AI investment.

For businesses looking to use AI, building their own data centres and purchasing equipment require upfront investment in both capital and technical staff. Chunghwa Telecom offers the option of renting data centre space and computing capacity. Its hicloud AI computing cloud, for example, offers hourly billing, allowing businesses to select GPU resources according to their computing needs without purchasing an entire set of equipment at once. Dedicated lines and cloud connectivity also allow customers to connect their existing systems to cloud resources. This enables Chunghwa Telecom to earn service revenue from customers’ ongoing use of data centres, networks and computing capacity, in addition to taking on construction projects.

In the first half of 2026, the value of information and communications technology (ICT) contracts secured by Chunghwa Telecom had already reached the previous full year’s level. Its Lunping AI data centre, which opened in August of the same year, further expanded its data centre services. For telecoms operators, the ability to capture corporate AI spending depends both on the number of construction projects they secure and on whether new facilities can attract customers to rent them on an ongoing basis. This is how data centre and network assets can generate further revenue.

Demand for cross-border connectivity is also already reflected in revenue. In August, increased cross-border data transmission by AI supply chain customers drove year-on-year growth of more than 60% in Chunghwa Telecom’s international circuit revenue. As businesses expand production and operations across different regions, they need to connect their locations. The overseas expansion of Taiwan’s supply chain therefore creates service opportunities for telecoms operators, extending the benefits to businesses that help companies transmit data.

Taiwan Mobile has also launched services combining GPU computing capacity, dedicated lines and cybersecurity protection. Both companies are using their existing networks, data centres and corporate customer bases to expand AI-related services, extending growth opportunities for Taiwan’s telecoms sector from personal communications to corporate demand.

Corporate expansion creates financing demand for banks

Businesses expanding their AI operations need funding as well as equipment and services. When suppliers require advance payment to secure materials, businesses must fund procurement before delivering their products and receiving payment from customers. Bank financing can bridge the funding gap between paying for purchases and collecting sales proceeds.

In May this year, Taipei Fubon Bank and First Commercial Bank jointly arranged a syndicated loan for Phison equivalent to NT$12 billion, with 11 banks participating. As AI applications drove up demand for storage and suppliers required advance payment, the two banks structured prepayment financing to support Phison’s material purchases over the next three years. As orders increase across the supply chain, financing needs extend from loans for factories and equipment to working capital for procurement. The joint participation of First Commercial Bank and Taipei Fubon Bank illustrates how different banks can secure this business through their corporate customer relationships.

How do rising pay and dividends support consumption and wealth management?

Corporate profits also flow to employees through pay and bonuses. Micron reportedly announced in September that it would pay a thank-you bonus of NT$1 million per person to employees in Taiwan who joined before 29 August 2025, while new employees joining in fiscal 2026 would receive a payment pro-rated according to their length of service. Delta also offers year-end, performance and profit-sharing bonuses, allowing employees to share in its business results.

Looking at earnings across the economy, average regular earnings for all employees in Taiwan’s industrial and services sectors rose 3.32% year-on-year in July. Including bonuses and overtime pay, average total earnings rose 8.25%. By sharing profits, businesses increase the funds available to households and support consumption of local goods and services. This is an important link between corporate performance and household income: in addition to fixed monthly pay, employees can share in their employers’ business results through variable compensation.

Alongside bonuses for employees, companies distribute profits to shareholders through dividends. TSMC’s cash dividends in 2026 total approximately NT$24 per share, up one-third from NT$18 in 2025. We believe growth in profits and dividend payouts could increase the funds available for shareholders to allocate. Beyond everyday consumption, these funds can be held as deposits or allocated to funds, insurance, and equity and bond products, creating demand for financial institutions’ services.

Such activity is also reflected in macroeconomic consumption data. Taiwan’s central bank noted that insurance and financial service fees, among other items, accounted for 37% of private consumption growth in the second quarter, up from 24% in the first quarter. As customers increase their wealth management and trading activity, the resulting service charges represent both financial services purchased by households and revenue for financial institutions.

Fubon’s wealth management results illustrate the revenue scale of this business. In the first half of 2026, Taipei Fubon Bank’s net fee income from wealth management reached NT$14.48 billion, up 44.2% year-on-year. Within this, net fee income from insurance and funds rose from NT$5.31 billion and NT$3.43 billion in the same period last year to NT$7.02 billion and NT$4.93 billion, respectively, while net fee income from equity and bond products increased from NT$0.87 billion to NT$2.03 billion. Equity and bond products grew faster, but the combined increase of approximately NT$3.21 billion from insurance and funds remained an important source of additional wealth management revenue.

As customers allocate more of their assets, banks can expand their fee-generating businesses beyond lending. The sales mix across products also affects wealth management revenue growth. The breakdown above therefore helps explain which services enable the financial sector to capture these fund flows.

The securities business is more directly affected by trading activity. Over the same period, Fubon Securities’ net fee income from brokerage and wealth management reached NT$13.04 billion, up 145.6% year-on-year, while its brokerage market share rose from 6.70% to 7.14% (excluding foreign brokers). Active trading in Taiwanese equities increased demand for trading services, while the rise in market share shows that Fubon also expanded its share of the market. From financing corporate procurement to helping households allocate assets, the financial sector can meet funding needs arising from the AI boom through a range of services.

Table 3: How three types of business benefit from AI demand

Investment implications: Broader spillover effects give Taiwan’s market more lasting growth momentum

Figure 2: Taiwan Capitalization Weighted Stock Index (TAIEX) and earnings per share

Figure 3: TAIEX forward valuation

We believe this wave of AI spillover effects is only just beginning. Several industries in Taiwan, from power equipment and engineering to telecommunications and finance, are already meeting the resulting demand. As companies continue to expand production capacity, new data centres enter service and subsequent procurement creates financing needs, this investment will continue to generate business for a range of companies. Profits also flow to households through pay and dividends, supporting consumption and asset allocation. This allows more industries to share in the AI boom and gives Taiwan’s market a broader base of earnings growth.

Although short-term swings in sentiment and valuations may continue to cause volatility in Taiwanese equities, we therefore remain positive on the market’s medium- to long-term outlook and maintain our 4-star ‘Very Attractive’ rating. Based on a fair P/E of 20 times, our end-2028 target is 74,640 points, implying potential upside of around 51.4% from the closing level on 8 October. Investing in Taiwan offers exposure both to semiconductor growth and to the opportunities created as demand extends to other industries.

Table 4: TAIEX earnings forecasts and target