Semiconductor foundry technology influences the power consumption, performance, size and integration options available to designers of industrial electronics. For automation engineers and product teams, the relevant question is not simply which process node is smallest. The choice also depends on application requirements such as low-power operation, wireless connectivity, sensing, automotive qualification and the integration of multiple chips.
TSMC’s 2017 technology portfolio illustrates how a large foundry organized process and packaging options around distinct application markets, including the Internet of Things (IoT), automotive electronics and high-performance computing. The details are specific to that period, but the portfolio offers a useful view of how logic, specialty processes and packaging can serve different design priorities.
Semiconductor foundry technology serves varied design needs
A pure-play foundry manufactures semiconductor products designed by customers rather than selling chips under its own brand. This model allows customers to select manufacturing technologies and services without the foundry directly competing with them through its own semiconductor products. In TSMC’s 2017 profile, its broad customer base spanned computer, communications, consumer, industrial and standard semiconductor markets.
That range matters because different applications call for different combinations of process characteristics. Advanced logic processes can target performance and chip density, while other platforms emphasize low power, radio-frequency (RF) functions, embedded memory, sensors or power management. A technology portfolio organized around these needs gives designers options beyond a single progression toward smaller geometries.
Low-power processes support connected devices
Industrial IoT devices may combine processing, sensing and wireless communication while operating within tight energy constraints. TSMC’s 2017 portfolio included ultra-low-power (ULP) process options at 55nm, 40nm, 28nm and 22nm, as well as ultra-low-leakage development at 22nm. The company positioned these technologies for applications such as IoT devices, wearables, wireless microcontrollers and sensor hubs.
These process families show why node selection should be tied to product requirements rather than treated as a simple contest over feature size. For example, the report described 55ULP as integrating RF and embedded flash memory, which could simplify system-on-chip designs. It also identified 40nm ULP for wireless connectivity, application processors and sensor hubs. For engineering teams, integration choices can affect how functions are allocated across a product’s electronics. The appropriate decision still depends on the specific design, its power targets and the capabilities available in the selected process ecosystem.
The portfolio also included 22nm ULP and ultra-low-leakage technologies aimed at applications such as IoT and wearable devices. Those technologies were at different development stages in 2017, so they should be understood as historical roadmap information, not as a statement of current availability.
Specialty technologies connect chips to industrial functions
Many automation-related products rely on more than digital logic. RF processes support wireless communications; embedded flash can serve microcontrollers; and sensor technologies convert physical conditions into signals that electronic systems can process. TSMC’s 2017 specialty offerings included RF, embedded memory, resistive random-access memory (RRAM), bipolar-CMOS-DMOS (BCD) and single-photon avalanche diode (SPAD) technologies.
The report described SPAD as a platform for LiDAR applications and linked three-dimensional imaging and sensing with machine vision. This is relevant to industrial and automotive system designers evaluating sensing components, although the platform itself does not determine a complete system’s performance. Similarly, TSMC’s pilot line for piezoelectric technology was intended to support development of devices such as microphones, ultrasonic sensors and actuators, including applications in medical and health products.
Automotive electronics bring another set of requirements. TSMC reported that 16FF+ technology began production for automotive applications in 2017, while specified foundation IP passed AEC-Q100 Grade 1 qualification and was certified to ISO 26262 ASIL-B. These details illustrate the importance of checking qualification and functional-safety evidence for the relevant components and design—not assuming that a process label alone establishes system suitability.
Packaging is part of the system architecture
Advanced packaging can combine chips and memory in ways that address system-level requirements. In 2017, TSMC reported volume production of its second-generation Integrated Fan-Out Package on Package (InFO-PoP) for mobile products, integrating a system-on-chip with DRAM. It also began production of Chip on Wafer on Substrate (CoWoS) for high-performance computing, combining a system-on-chip with stacks of high-bandwidth memory on an interposer.
For design teams, these examples highlight that packaging should be considered alongside the process used to manufacture each chip. Integration may involve multiple dies, memory and interconnect structures, so choices can affect the product’s footprint and power-and-performance objectives. The source also described continued production of copper-bump and wafer-level chip-scale packaging technologies, reinforcing that packaging options vary with application and silicon generation.
Evaluate platforms against application requirements
TSMC framed its offerings around four application platforms: mobile, high-performance computing, automotive electronics and IoT. That approach reflects a practical design principle for industrial automation: start with the product’s functional requirements, then assess the combination of process technology, specialty functions, IP and packaging needed to meet them.
- For connected sensors and controllers: assess low-power logic, wireless connectivity and memory requirements together.
- For machine-vision or ranging systems: consider the sensing technology and its role within the wider electronics design.
- For automotive-related equipment: verify applicable component qualifications and functional-safety requirements.
- For compute-intensive designs: evaluate whether advanced logic and multi-chip packaging align with system needs.
Foundry ecosystems also matter. TSMC’s Open Innovation Platform and Grand Alliance brought together customers, electronic design automation providers, IP partners and suppliers. Such collaboration supports design enablement across process, implementation and backend services, and can be a relevant consideration when planning a product’s path from design to manufacturing.
Conclusion: match process and packaging to the product
TSMC’s 2017 portfolio demonstrates the breadth of choices behind semiconductor foundry technology: advanced logic, low-power processes, RF and memory integration, sensing platforms, automotive-related qualifications and advanced packaging. For industrial automation teams, the practical takeaway is to define application needs first and evaluate the complete technology platform against them. Power, connectivity, sensing, safety and integration requirements all influence the design; no single process node answers every question.