Vanguard International Semiconductor (VIS) is evaluating a second wafer fabrication plant in Singapore, even though its first facility there has yet to produce a single chip. The company's inaugural Singapore plant opened with its entire initial capacity already spoken for—a level of demand the chipmaker ties to artificial intelligence and the less glamorous specialty chips that AI infrastructure quietly depends on.

The expansion plans run through VSMC, the joint venture VIS operates with NXP Semiconductors.

A fab sold out before it switched on

VSMC inaugurated its first 300mm fab in Singapore's Tampines district on September 28, 2026. Production is expected to begin in the first quarter of 2027. The 300mm wafer format—the largest in volume use across the industry—carries more chips per wafer than smaller formats, a cost advantage that matters for high-volume production of mature-node parts.

That gap matters. The facility's initial capacity has already been fully booked, meaning customers committed to output that does not yet exist.

VIS Chairman Leuh Fang said customer demand has stretched the first facility's capacity beyond its limits. That pressure is what prompted the company to start evaluating a second plant—and booked-out capacity is precisely the demand case such an expansion decision rests on.

The economics of the first site are substantial. The investment is estimated at approximately US$6.7 billion to US$7.8 billion. Under the current plan, the Tampines fab will ramp to 44,000 wafers per month by 2029 and is expected to create about 1,600 jobs.

Ownership, technology, and "physical AI"

The joint venture is split 60% VIS and 40% NXP. That structure gives VIS the controlling stake while guaranteeing NXP a seat at the capacity table. Arrangements like this also let a foundry and a major customer share the capital burden of a multibillion-dollar build—no small consideration at the US$6.7 billion to US$7.8 billion scale the first fab carries.

VSMC focuses on manufacturing processes between 40nm and 130nm. These are mature nodes—older, larger chip geometries rather than the bleeding-edge processes used to produce flagship AI accelerators. Chips built on such nodes typically serve automotive, industrial, and other everyday applications rather than data center accelerators. The AI linkage runs through those surrounding systems: even AI infrastructure built on cutting-edge processors still depends on mature-node components for power, control, and interface functions.

NXP Chief Executive Rafael Sotomayor has framed the facility around a specific idea. He said it will significantly contribute to enabling "physical AI" across sectors including automotive and industrial applications.

Physical AI refers to intelligence that lives in machines interacting with the real world—cars, factory equipment, and industrial systems—rather than in a chatbot window.

What it means for the chip supply chain

The most immediate signal concerns capacity tightness. With the first VSMC fab fully booked before production starts in Q1 2027, buyers of mature-node chips may face a more competitive environment for supply. It also shows how AI's pull on the supply chain now reaches beyond leading-edge accelerators into the mature-node components that surround them.

For VIS and NXP, the second plant remains at the evaluation stage. No timeline, investment figure, or capacity target has appeared in the details shared so far, so it should be treated as a plan rather than a commitment.

The first fab's numbers offer a sense of scale, however. A facility costing approximately US$6.7 billion to US$7.8 billion and ramping to 44,000 wafers per month by 2029 is a long-horizon bet that a second site would presumably have to match or justify.

There are also risks worth tracking. The Tampines fab still has to execute its ramp, hitting its Q1 2027 production start and scaling toward its 2029 target on schedule.

Singapore, meanwhile, stands to benefit from the jobs and investment tied to the first site. The city-state is already one of the world's established semiconductor manufacturing locations, so a second plant would deepen a footprint built on an existing supplier ecosystem—though the details of any such facility have yet to be laid out.

The key things to watch are straightforward: a formal decision on the second plant, any updated investment figures, and whether the Tampines fab begins production on its expected Q1 2027 schedule.