News
Wiwynn projects 50% growth in AI server shipments driven by custom ASICs and plans increased CapEx
Wednesday, February 25, 2026 at 01:21 AM
Wiwynn expects AI server shipments to grow by 50% in 2024, specifically driven by custom ASIC projects for cloud service providers rather than standard Nvidia systems. The company plans a significant increase in capital expenditure this year, exceeding last year's NT$13 billion, to support sustained demand for AI infrastructure into 2026.
Context
Taiwanese server manufacturer Wiwynn has projected a 50% increase in both AI and general-purpose server shipments for 2026, signaling a robust recovery in traditional cloud spending alongside the ongoing AI boom. This growth is primarily fueled by custom ASIC (application-specific integrated circuit) projects for major hyperscalers, including Microsoft, Meta, Amazon, and Google. While Nvidia-based GPU systems remain a market staple, Wiwynn’s specialized focus on non-Nvidia proprietary chips has allowed it to capture a dominant share of the inference-driven server market, with AI-related products now expected to exceed 50% of total revenue.
To support this expansion, Chair Emily Hong announced that capital expenditure will rise significantly this year from the NT$13 billion spent in 2025. This investment targets new manufacturing capacity in Taiwan, Mexico, and the United States to mitigate geopolitical risks and meet the infrastructure needs of the "Big Four" cloud providers. Management anticipates that demand will accelerate in the second half of 2026, bolstered by the mass production of next-generation chips like Amazon's Trainium 2 and upcoming custom accelerators for Google and OpenAI.
Related Companies
Wiwynn
6669
Amazon
AMZN