Foxconn chairman says AI server share could reach 50% within a couple of years

Foxconn Chairman Young Liu sees room to capture half the global AI server market within a couple of years. In an interview with CNBC’s Christine Tan published October 30, he paired that growth outlook with a warning about overinvestment, even as he said the company’s production capacity was fully booked for at least 18 months.

A larger share—and a different revenue mix

Liu put Foxconn’s current share of the global AI server market at about 40%. He said that could rise to roughly 45% next year before reaching 50% within a couple of years. Those figures describe his expectations for Foxconn’s competitive position, rather than growth rates for the market as a whole.

The ambition would deepen a shift at a manufacturer best known for assembling Apple’s iPhones. Liu expects AI infrastructure to generate 60% to 65% of Foxconn’s revenue in three to five years. That forecast excludes edge AI devices, which run AI directly on local hardware, rather than treating every AI-related device as infrastructure revenue.

His confidence rests partly on demand he believes has yet to arrive. AI model developers and cloud service providers have supplied much of the demand so far, Liu said. Governments and enterprises have not yet fully embraced the technology. He described AI demand as “quite strong,” alongside the disclosure that production capacity was already booked.

U.S. production starts with higher-margin equipment

To serve that demand, Foxconn intends to broaden its U.S. manufacturing presence and move further up the supply chain. Liu said a wider manufacturing footprint would help the company respond more quickly to the U.S. market. The plan extends beyond simply increasing the volume of equipment it already makes.

Higher U.S. manufacturing costs will shape the initial product mix. Foxconn will first focus on equipment with stronger margins, including AI server racks and data center equipment, Liu said. He expects greater production scale could eventually reduce manufacturing costs, making scale both a growth objective and part of his answer to the cost problem.

That manufacturing push sits within Foxconn’s effort to cover more of the AI infrastructure business, from servers and racks to entire data centers and computing services. Liu linked the potential spending bill to how far the company pursues vertical integration—taking on more stages of that supply chain itself.

Overinvestment will kill the company

Young Liu, Foxconn chairman, speaking to CNBC

One customer’s retreat is not an industry slowdown

Liu left room for Foxconn to spend more than its planned $3 billion to $5 billion a year on AI. Whether it does so would depend on the extent of that integration. The spending range is therefore a plan, not a firm ceiling, even as he promised to keep investment under tight control.

Asked what would happen if major customers reduced AI spending, Liu drew a distinction between a setback at one buyer and a wider downturn. Lower demand from one customer could be offset by growth elsewhere, he said. That response relies on other customers continuing to increase their purchases.

An industry-wide slowdown would pose a much larger challenge, Liu acknowledged, both for Foxconn and the broader AI supply chain. His outlook thus separates the opportunity to win more business from the risk of demand weakening across customers. Confidence in the first did not lead him to dismiss the second.