MPW is not a single product. Each foundry runs its own shuttle program at its own set of nodes, with its own schedule, pricing model and access conditions.
Teams that treat all MPW as equivalent often discover late that node availability, IP ecosystem and shuttle frequency vary significantly across foundries. A clear comparison early avoids wasted evaluation cycles.
TSMC MPW (VAT / Online Shuttle)
TSMC operates the VATS (Virtual Auto Tape-out Shuttle) program, also referred to as the TSMC Online Shuttle. It is the most widely referenced MPW service in the industry.
Node coverage spans from mature processes like 0.18um, 0.13um and 90nm through advanced FinFET nodes including 28nm, 16nm/12nm, 7nm, 5nm and 3nm family nodes.
Shuttle cadence varies by node. Mature nodes may run quarterly or more frequently; advanced-node shuttles often run on a fixed but less frequent calendar, sometimes 2–4 times per year depending on demand.
Access is generally open to registered account holders, though advanced nodes may require additional qualification and higher minimum participation fees.
- Broadest advanced-node MPW coverage among pure-play foundries
- Strong IP ecosystem via TSMC DesignWare partners and third-party vendors
- Higher per-shuttle cost at leading edges, but lower execution risk
- Documentation and support infrastructure are well established
SMIC MPW Shuttle
SMIC provides MPW shuttle services across its qualified process nodes, with particular strength in mature and specialty nodes relevant to China-based and export-controlled programs.
Publicly documented MPW coverage includes 0.35um, 0.18um, 0.13um, 90nm, 55nm, 40nm and 28nm nodes, with some availability at 14nm depending on qualification status and region.
SMIC shuttles typically run on a regular cadence, often aligned with Chinese mainland business cycles. Lead times from design-in to wafer delivery are competitive for programs focused on the Asia supply chain.
For programs subject to export controls or seeking a non-US-foundry path, SMIC represents a key alternative source of silicon capacity and process technology.
- Strong mature-node MPW availability (0.35um to 28nm)
- Competitive cost structure, especially in local currency terms
- Important option for programs requiring a non-US foundry route
- IP ecosystem is growing but narrower than TSMC at advanced nodes
UMC MPW Program
UMC offers an MPW shuttle program covering its specialty and mainstream nodes, with emphasis on mixed-signal, RF and specialty process differentiation.
Documented MPW nodes include 0.18um, 0.13um, 90nm, 55nm, 40nm, 28nm and 14nm, with particular strength in embedded HV, BCD, eFlash and RF-SOI technologies.
UMC's positioning is strongest for products where process specialization matters more than raw geometry scaling — power management, display driver, RF connectivity and IoT edge devices.
Shuttle frequency is generally regular, with multiple runs per year on popular nodes. Access is open to qualified customers through standard registration.
- Specialty-process focus (HV, BCD, eFlash, RF-SOI)
- Good fit for power management, display driver and IoT products
- Competitive pricing on specialty nodes vs. pure-play leaders
- Broader APAC presence with Taiwan and China operations
GlobalFoundries MPW
GlobalFoundries provides MPW shuttle access across its differentiated process portfolio, with emphasis on 22FDX, 12LP and 45/40RFSOEE platforms.
GF has shifted strategy away from leading-edge scaling toward specialty and embedded solutions. Its MPW program reflects this: strong coverage at 22FDX (FD-SOI) for low-power IoT and RF, 12LP for automotive and industrial, and 45RFSOEE for high-performance RF.
For programs targeting automotive (AFD), IoT edge or RF front-end applications, GF's MPW offering is a meaningful complement to TSMC-centric sourcing strategies.
Shuttle schedules are published regularly. GF's access model is straightforward for qualified accounts, with clear pricing and technical support channels.
- 22FDX FD-SOI platform is a unique MPW differentiator
- Strong automotive (AFD) and RF SOI offerings
- No direct competition at sub-7nm — focus is on differentiated nodes
- Good option for multi-source or risk-distribution strategies
How to Compare Foundries for Your MPW
The first filter should always be process-technology fit. If the product requires a specific node or a specific type of process (BCD, RF-SOI, FD-SOI), that narrows the foundry list quickly.
The second filter is IP availability. Even if a foundry offers the right node, the required IP blocks (SerDes, PLL, ADC, memory compilers) must be available and qualified at that node.
Third is logistics and schedule alignment. Shuttle frequency, tape-out deadlines, wafer delivery location and communication language all affect real program timing.
Fourth is long-term sourcing strategy. Teams planning to move from MPW to production should confirm that the same foundry can support volume manufacturing, or plan a process port early if a second-source path is needed.
Common Foundry Selection Mistakes
Choosing a foundry based only on advertised node geometry without checking actual MPW shuttle availability leads to delays when no upcoming shuttle slot matches the program schedule.
Assuming IP availability is identical across foundries at the same nominal node is incorrect. Each foundry has a different IP partner network and qualification list.
Ignoring export-control implications can create serious program risk. Some foundries and some nodes are subject to restrictions that vary by applicant nationality, end-use and destination.
Evaluating MPW cost without considering the full production-path transition cost underestimates total program expense. The cheapest MPW shuttle may lead to the most expensive production transfer.
