TAIPEI (Taiwan News) — Taiwanese AI server makers are ramping up production to meet surging demand from the US, the country’s largest market for generative AI hardware, Nikkei Asia reported Thursday.
Top electronics manufacturers Foxconn, Quanta Computer, and Wistron are expanding output in Taiwan, the US, and Southeast Asia to supply tech giants like Nvidia and major cloud service providers.
Quanta CFO Yang Chun-lie (楊俊烈) said the company plans to at least double AI server production capacity by the end of 2026. “We will at least double our production capacity by the end of 2026,” he told investors during the company’s November earnings briefing.
Once primarily a notebook assembler for clients including Apple, Quanta has rapidly grown its AI server business, which now accounts for roughly half of its revenue. From January to September, the company posted NT$1.49 trillion (US$47.8 billion) in revenue, up 49.5% year-on-year, a record for the period.
Quanta’s main production facilities for AI servers are located in Taiwan, Thailand, and the US, with future expansions targeting North America and Southeast Asia.
According to Market Intelligence and Consulting Institute, Taiwanese companies now produce 90% of the world’s AI servers. Electronics manufactures that focused on PCs and smartphones have leveraged their expertise to compete for high-value AI contracts.
Foxconn, Quanta, and Wistron dominate this market, with key manufacturing processes for Nvidia’s next-generation AI servers next year concentrated among them. Servers destined for the US, which handle sensitive data, are mostly manufactured in Taiwan and Southeast Asia rather than China.
Taiwan’s AI server exports to the US from January to October nearly doubled year-on-year to NT$840.2 billion, trade statistics show. The growth is even higher when components and related products are included, contributing to Taiwan’s widening trade surplus with Washington.
Wistron recently opened a factory in Hsinchu and announced plans to invest NT$5.28 billion to expand the facility. The company is also set to begin mass production in Texas next year, serving clients beyond Nvidia.
Foxconn, which holds a 40% share of the global AI server market, is boosting production in North America and plans to begin server output in Japan at a former Sharp facility. The company aims to meet demand for sovereign AI, systems developed and operated entirely within the purchasing country.
Rising electricity demand poses a challenge for all three manufacturers. Taiwan’s last operating nuclear plant closed in May, while US data center expansion has raised concerns about potential power shortages amid the AI boom.
Despite occasional concerns about an investment bubble, market research firm TrendForce projects global capital spending by the top eight cloud providers will reach NT$18.74 trillion in 2026, up 40% from 2025, indicating robust near-term demand for AI infrastructure.
