29th-Sep-2026 — The semiconductor industry’s AI investment cycle has largely been associated with advanced logic, HBM memory and increasingly sophisticated packaging. But another part of the semiconductor supply chain is becoming increasingly important: the analog, mixed-signal and power-management devices required to connect artificial intelligence with the physical world.
That is the market being addressed by VisionPower Semiconductor Manufacturing Company (VSMC), the joint venture between Taiwan’s Vanguard International Semiconductor Corporation (VIS) and NXP Semiconductors.
On September 28, 2026, VSMC officially opened its first 300mm wafer fabrication facility in Tampines, Singapore. The plant has already entered risk production and is scheduled to begin volume production in the first quarter of 2027.
Unlike the multibillion-dollar fabs targeting single-digit nanometer logic, VSMC will manufacture semiconductor technologies ranging from 130nm to 40nm. Those nodes remain highly relevant for many of the chips that surround today’s most advanced processors.
| Item | VSMC Singapore Fab |
|---|---|
| Location | Tampines, Singapore |
| Wafer size | 300mm / 12-inch |
| Joint venture partners | VIS and NXP Semiconductors |
| VIS ownership | 60% |
| NXP ownership | 40% |
| Process technologies | 130nm to 40nm |
| Target products | Analog, mixed-signal, power management and interposer applications |
| Risk production | Underway |
| Volume production | Q1 2027 |
| Planned capacity | Approx. 44,000 wafers/month by 2029 |
| Jobs expected | Approx. 1,600 |
| First sample lot yield | Above 99% |
VSMC says its first sample lot has already been processed with a yield above 99%, an important milestone ahead of the production ramp. The factory is expected to reach roughly 44,000 300mm wafers per month by 2029 and create approximately 1,600 jobs.
At first glance, a fab operating between 130nm and 40nm may appear only loosely connected to the current AI investment cycle. But AI systems require considerably more than CPUs, GPUs and accelerators. As AI moves from data centers into automobiles, industrial equipment, robots, smart infrastructure and increasingly autonomous systems, those systems will require large numbers of supporting semiconductors.
These include:
Many of these functions do not require 3nm or 2nm transistors. In many cases, established semiconductor nodes can offer a more attractive combination of cost, analog performance, voltage capability, reliability and process maturity. This is one reason mature and specialty processes remain strategically important even while the industry’s most visible investment is concentrated at the leading edge. VSMC specifically lists HPC, mobile, automotive, industrial and consumer markets among the target applications for its new manufacturing capacity.
The other important element of VSMC’s strategy is wafer size. Many mature and specialty semiconductor products have historically been manufactured on 150mm and 200mm wafers. VSMC is bringing these technologies onto a 300mm manufacturing platform.
A 300mm wafer provides significantly more silicon area than a 200mm wafer, allowing substantially more dies to be produced per wafer, assuming the same die size and comparable process conditions.
The larger format can therefore improve manufacturing economics when utilization is high enough to justify the significantly more expensive fab infrastructure and equipment.
For high-volume products such as automotive, industrial and power-management ICs, this can become an important competitive advantage. It also reflects a broader industry trend: mature technology does not necessarily mean old manufacturing infrastructure. Processes may remain at 40nm, 65nm, 90nm or 130nm while moving onto more automated, higher-productivity 300mm manufacturing platforms.
Much of today’s AI semiconductor discussion begins and ends with compute. But a physical AI system requires a much larger semiconductor ecosystem. Consider an autonomous robot. The AI processor may represent its computational brain, but that processor depends on multiple additional semiconductor functions:
Sensors → Analog Front End → Data Conversion → Processing → Connectivity → Power Management → Motor Control → Actuation
A vehicle follows a similar architecture. Advanced driver-assistance and autonomous-driving systems require powerful processors, but they also depend on power management, radar, sensing, battery control, networking, interface devices and numerous analog and mixed-signal functions. Consequently, the expansion of AI into real-world applications could generate semiconductor demand across a much broader range of process nodes than the AI accelerator market alone would suggest. This is the strategic context behind VSMC’s focus.
The semiconductor industry’s technical roadmap often gives the impression that older process nodes are gradually disappearing. In reality, semiconductor nodes rarely disappear simply because a smaller geometry becomes available. Instead, different nodes become specialized for different product requirements.
For example:
| Technology Requirement | Why Mature/Specialty Nodes Can Remain Attractive |
|---|---|
| High-voltage operation | Specialty device structures can be integrated |
| Analog precision | Often more important than transistor density |
| Power management | Requires specialized power devices and passives |
| Automotive reliability | Mature processes can offer extensive manufacturing history |
| Cost-sensitive products | Leading-edge lithography is unnecessary |
| Embedded nonvolatile memory | Requires specialized process integration |
| Long product lifetime | Industrial and automotive products can remain in production for many years |
This makes specialty foundry capacity fundamentally different from leading-edge logic capacity. Customers are not simply purchasing the smallest transistor available. They are selecting a combination of device performance, voltage capability, passive components, reliability, cost and manufacturing longevity.
VSMC was established in September 2024 as a joint venture between VIS and NXP. VIS owns 60%, while NXP owns 40%. NXP has also stated that it is entitled to 40% of the fab’s capacity. The combination is strategically interesting. VIS contributes foundry manufacturing expertise and specialty-process experience, while NXP is one of the world’s major suppliers of automotive, industrial, embedded and analog semiconductors.
For NXP, the fab provides additional control over long-term manufacturing capacity while supporting geographic diversification. NXP has described VSMC as part of its broader hybrid manufacturing strategy, combining internal manufacturing, joint-venture capacity and external foundries.
The new fab also reinforces Singapore’s already substantial semiconductor manufacturing ecosystem. Singapore hosts operations from numerous semiconductor companies across wafer manufacturing, equipment, materials, assembly and test. For semiconductor manufacturers, the country offers established infrastructure, engineering expertise, political stability and proximity to the broader Asian semiconductor supply chain. The VSMC project also demonstrates how quickly new manufacturing capacity can be brought online. The joint venture was established in 2024, groundbreaking followed in December that year, and less than two years later the fab had already moved into risk production. VSMC says the factory uses a highly automated smart-manufacturing model incorporating AI and digital management technologies to improve efficiency and customer service.
The opening of VSMC illustrates a broader change taking place across the semiconductor industry. For much of the last decade, the industry’s strategic discussion centered on leading-edge transistor scaling. That remains critical. But automotive electrification, industrial automation, energy infrastructure, robotics and physical AI are increasing the strategic importance of semiconductor technologies that may be several generations behind the leading edge in terms of transistor dimensions.
A 40nm or 90nm chip may never receive the attention of the latest AI accelerator. Yet a system cannot function without the power-management ICs, sensor interfaces, communication devices and analog components surrounding that processor. The semiconductor opportunity created by AI may therefore ultimately prove much wider than advanced logic alone.
For fabless companies and IDMs, the most interesting question will be how much of VSMC’s capacity ultimately becomes accessible to external foundry customers and which specialty process platforms become available. The new capacity could become relevant to companies designing:
VSMC has said the facility will operate across 130nm to 40nm and serve both VIS and NXP requirements as the operation scales. Its current production capacity has reportedly already been booked by global customers, according to reporting from Taiwan’s Central News Agency. That early demand itself is a useful signal. Mature and specialty semiconductor manufacturing remains far from obsolete.
VSMC’s new Singapore fab is important precisely because it is not attempting to compete at the industry’s smallest process nodes. Instead, VIS and NXP are investing in the analog, mixed-signal, power and specialty semiconductor technologies that remain essential to automobiles, industrial systems, mobile electronics and the emerging physical-AI ecosystem.
With volume production targeted for Q1 2027 and capacity expected to reach approximately 44,000 300mm wafers per month by 2029, the new facility will become a significant addition to Asia’s specialty semiconductor manufacturing base. The AI semiconductor boom began with compute.
Its next phase may increasingly depend on the less visible chips that allow that computing power to sense, communicate, control power and interact with the physical world. And those chips will continue to create demand for 40nm, 65nm, 90nm and 130nm manufacturing for many years to come.