TAIPEI, Jan 24, 2026, 03:17 (GMT+8)
- Pegatron plans to finish its first U.S. factory in Texas by the end of March, aiming to start trial production in late March or April.
- The plant plans to produce artificial intelligence servers powered by Nvidia chips, as Taiwan’s contract manufacturers expand U.S. production capacity.
- Pegatron is expecting triple-digit growth in AI-server revenue this year, expanding its reach beyond GPU servers to include inference and custom-chip systems.
Pegatron’s first factory in the U.S., located in Texas, is set to be finished by the end of March, the company’s CEO said Friday. Trial production — test runs before full-scale manufacturing — should kick off in late March or April. The facility will focus on AI server products, including systems powered by Nvidia chips. (Reuters)
The schedule hands Pegatron a rare asset among Taiwan contract manufacturers: a factory it plans to operate in the U.S., not merely a sales or service outpost. It’s also a wager that the fastest-growing segment of data center investment will be built closer to American clients.
Washington has been pushing Taiwan, which holds a sizable trade surplus with the U.S., to boost onshore investment. A recent U.S.-Taiwan trade agreement trimmed tariffs on many Taiwanese exports from 20% down to 15%, backed by a commitment of $250 billion in Taiwanese investments focused on semiconductors, energy, and AI, according to a Reuters report. (Reuters)
Pegatron President and CEO Kuang-Chih Cheng told reporters in Taipei that the Texas facility will be the company’s first factory “established and operated” in the U.S. “The U.S. plant is our first factory established and operated in the United States,” he confirmed. (The Economic Times)
Pegatron revealed it purchased an industrial building and land in Texas last October, following other Taiwanese tech firms like Foxconn, Inventec, and Wistron investing in the state. “We should be able to begin trial production in late March or April,” Cheng told reporters. (Investing)
The company supplies Apple, Microsoft, and Tesla, announcing its Texas plant will produce AI server products. These servers are loaded with top-tier chips designed to train AI models or handle large-scale AI operations.
At a company event Friday, Cheng revealed Pegatron expects its AI server revenue to jump by triple digits this year, following a similar boom last year. The company is moving beyond just graphics processing unit (GPU) servers — GPUs handle heavy math calculations — and branching into “inference” servers that run trained AI models. They’re also developing systems using application-specific integrated circuits, or ASICs, which are custom chips built for a single task. (Taipei Times)
Co-CEO Johnson Teng revealed that Pegatron has landed orders for ASIC-powered servers, with shipments set to begin this year. He added that several projects target transformer-based training and inference tasks, the backbone of many current AI models. Cheng noted that their efforts now go beyond just board-level manufacturing, moving into full system integration—including advanced cooling solutions.
Pegatron revealed a new AI server plant in Mexico started trial production late last year and is now set for mass production. In Taiwan, Cheng noted the company constructed two new factories in Taoyuan, scheduled to kick off production this year.
The U.S. timeline is tight, and the ramp-up isn’t assured. Construction hiccups, staffing issues, or shortages of crucial components—particularly high-end chips and the liquid-cooling equipment essential for AI servers—could delay trial runs well into the second quarter, with competitors vying for the same supplies.
Since Trump’s first term, Pegatron has been shifting its manufacturing out of China, branching into Southeast Asia and Mexico. The company runs a maintenance base in Indiana and keeps an office in California. It’s directing most North American EV production toward Mexico and nearby areas to sidestep higher tariffs as EV localisation accelerates.