1 The Nanoelectronics Industry

The electronic components industry, generically described as “nanoelectronics,” is an industry with specificities that set it apart from almost all other industries. Its perimeter is expanding continuously; it started by relying on chemists and physicists handling semiconductor crystals; then added electrical engineers to build circuits and functional blocks; now it also employs considerable numbers of software and system engineers. Its customers achieve increased economic efficiency by allowing functionality to be integrated in components; this way, they allow their vendors to expand their competence and move up the value chain.

The nanoelectronics positioning in the global economy is often depicted as the reversed pyramid shown in Figure 1. At the tip of the pyramid, there is the nanoelectronics industry producing components – popularly known as “computer chips.” At the next level, “original equipment manufacturers” (OEMs) use the components to build electronic products with a market value roughly five times higher than that of the components. The electronic equipment industry enables information and communications services with a market value about five times higher than that of the equipment they use. This way, it can be estimated that nanoelectronics enable economic activities with a total value around 25 times higher than its own market value: in 2014, they approached $9000 billions, or 11% of the approximately $80,000 billions gross domestic product of the world. Their weight continues increasing year after year.

The electronic components are used in almost any artifact produced by the industry: they can be found everywhere, from the lock on a hotel door to the space shuttle. They are manufactured under extreme cleanliness conditions on slices of monocrystalline silicon called “wafers” in dedicated facilities called “wafer fabs.” A wafer fab operates highly sophisticated equipment using specialty materials to build hundreds or thousands of structures on each wafer. A structure can contain billions of devices, essentially transistors, but also resistors, capacitors, inductors, and so on; it is so complex that it can only be conceived using “electronic design automation” (EDA) tools, in fact computer programs that assemble predefined functionalities from a library containing blocks capable to perform arithmetic and logic calculations, memory blocks to store software and data, connectivity blocks, and so on. Before delivering them to the users, the structures are diced from the wafer, put in packages foreseen with electrical contacts, tested, and marked; these operations are performed in specialized “assembly lines.”

The nanoelectronics industry consists essentially of all the entities that contribute toward delivering electronic components to the OEMs: they are primarily “integrated devices manufacturers” (IDM) and their suppliers, although the IDM denomination is not exactly correct. First, not all component providers build “integrated” devices; in fact, the “discrete” components (such as individual transistors, diodes, etc.) continue being an important part of the total production, with specific components showing significant growth, such as light-emitting diodes (LEDs) used as lamps, power devices, or micro-electromechanical systems (MEMS). Second, not all component providers are also “manufacturers”; an increasing part is represented by an “emerging” value chain consisting of “fabless” companies using contract manufacturing executed by third parties called “foundries.” This trend started in 1987 with the establishment of the Taiwan Semiconductor Manufacturing Company (TSMC), the first “pure play” foundry, but became highly significant in the last 5 years since two fabless companies rank among the top 10 sales leaders. Third, a number of specialties (like equipment, materials, design automation or assembly and test) split off from the IDMs forming branches of a dedicated supply chain that must be also given proper consideration. Figure 2 illustrates the segmentation of the industry in different specialties and business models.

This overview of the nanoelectronics industry takes into account all types of discrete and/or integrated electronic components suppliers, together with their dedicated supply chains.

2 The Nanoelectronics Ecosystem

The nanoelectronics industry has one of the highest innovation rates in the economy, often ranking number 1 in terms of R&D expenditures as a percentage of sales. The industry capitalizes upon ingenuity from everywhere in the world, and from any sources, including commercial companies of all sizes, academic and institutional research, and individual investigators. It succeeded sustaining over more than half a century an unparalleled flux of innovation.

The extreme precision and cleanliness necessary to achieve reasonable manufacturing yields at nanometric scale results in unusually high fixed costs of the research and manufacturing infrastructure. It is actually quite impossible to confirm the value of an innovation at low technology readiness levels (TRLs)²: positive laboratory results are no more than a hope; successful implementations in realistic environments are no more than a possibility; any novel idea must be taken all the way to an operational environment before concluding on its viability. Since the operational environments are extremely costly, typically in the multibillion dollar range, the industry uses “lab–fabs,” that is, facilities used both for research and for manufacturing of commercial products that can absorb the majority of the fixed costs. This approach is practically adopted across the board.

Around each company operating lab–fabs, there is a considerable number of small- and medium-sized companies, of research institutes, and university laboratories collaborating to maintain a technology pipeline filled with new ideas that are continuously scrutinized and moved toward higher TRLs to narrow the selection to the ones that can be included in future recipes. The metaphor of the industry is an ecosystem, relying on the large sequoia trees to withstand fires and tempests in the forest, on medium-sized trees and small bushes to provide a habitat bringing creative ideas to life, and on grass root innovation from university and institutional research to maintain a soil reach in nutrients.

The industry makes effective use of project-oriented collaborative research; it is natural to find it well represented in programs carried out by alliances or consortia that naturally cross boundaries between geographic areas and between disciplines.

Also, its systemic and strategic significance attracts the attention of public entities; some of them get involved in setting directions and priorities, some other simply provide financial incentives to facilitate the progress or promote a particular location.

3 Miniaturization

The primary engine of progress in the industry is the “miniaturization.” Unparalleled advances in equipment, materials, and manufacturing techniques enable a continuous reduction in size of the elementary function, the transistor. The peculiarity of the semiconductor technology consists in the fact that this improves simultaneously not only all performances parameter but also the unit costs. This trend was recognized already in 1965 (see footnote 1), being known as the “Moore's law”; it initially stated that the number of components per integrated function will double every year. Today, it is usually formulated in terms of the number of components per unit area doubling every (so many) month. In fact, the number of months is of secondary importance as long as this quasi-exponential progression continues, as it did since half a century, in spite of periodical warnings about insurmountable barriers – always overcome by the ingenuity of the researchers in the field. This is described as the “More Moore” progression.

Nanoelectronics follows since 1994 the “International Technology Roadmap for Semiconductors”³ (ITRS) generated by hundreds of specialists from all around the world. It identifies the challenges to overcome and the timing of the industrial deployment of the successive technology generation called “nodes.” Each node is characterized by a “feature size” expressed in nanometers, a rather generic identifier for a whole new set of technology capabilities that obviously depend on many more parameters than just one geometric dimension. Each feature size is smaller by the square root of 2 than the previous one, so that every new node appears to cut in half the silicon real estate needed for a function, in reference to the Moore's law. Companies try to beat the ITRS schedule and be first to market with the next node; in fact, the differences in time are small, and industry moves more or less in lockstep. This quasi-synchronization induced by ITRS guarantees the demand for the equipment and materials suppliers that could therefore invest in R&D at least 5 years before a new node was expected, enabling in due time the subsequent development of new manufacturing processes. Nowadays the industry is considerably widening its markets, serving numerous applications with technology needs that do not always evolve in synchronicity. It becomes increasingly difficult to define a unique, all-encompassing roadmap. ITRS is currently in a restructuring process. It remains to be seen to which extent its success in providing guidance for the industry will continue.

Making the devices smaller require high capital investments in advanced wafer fabs in order to keep the manufacturing yield close to 100%; today, a viable fab costs in excess of $10 billions. Surprisingly, the more expensive the fab, the lower the unit costs of the products it builds, thanks to an overproportional increase in productivity and the beneficial effects of the economy of scale. The decision whether to operate or not own fabs is essential for each company: If the business volume is not commensurate with the capacity of a commercially viable fab, it is preferable to rely on contract manufacturing that can aggregate the demand from several users to reach the needed economy of scale. In this case, the business model may be “fab-lite” when outsourcing most standard but maintaining some proprietary manufacturing generating market differentiation, or entirely “fabless” when relying on system and circuit design to compete. This drives down the number of the companies that participate in the miniaturization race.

As the number of devices per unit area increases, complex functions that were realized before by OEMs can now be integrated on a chip by the components suppliers. Advanced components enable electronic equipment with increased capabilities, better performance, lower power consumption, and smaller form factors. Applications can move from being stationary to becoming mobile, then portable, and eventually even wearable by a person – or go even further enabling autonomous functionality incorporated in communicating objects building the “Internet of Things.”

New applications can be addressed at every stage on the road, fueling a continuous increase in demand that is yet far from saturation. Modern applications as high-performance computing, data centers, Internet routers, cloud computing, or big data primarily rely on the newest technology nodes. There is no doubt that nanoelectronics will continue on the miniaturization path that will fuel growth in the foreseeable future.

4 Functional Diversification

Although a new technology node is ready every 2 years or so, each node will be used in manufacturing for 10 or 20 years after introduction. As a technology generation matures, the cost diminishes and it becomes affordable to add new features in the manufacturing recipe to address specific application requirements. They usually include specific device architectures for nonvolatile memories, power, radio frequency, sensing, actuating using either electronic effects or micromachined structures. These enrichments prolong the life expectancy of a technology generation; increase the volume of the commercial production it enables; and improve the overall return on investments. Since they create value through diversification rather than through miniaturization, they are referred to as the “More than Moore” progression.

The “More Moore” and “More than Moore” directions have been for some time depicted as orthogonal. In fact, diversification builds upon processing capabilities introduced in the miniaturization progress.

In market surveys, diversification products are partially reported together with the integrated circuits (ICs) and partially separated under the title optoelectronics – sensors/actuators – discretes (O–S–D). However, the distinction is not always sharp, for example, camera chips are classified together with the LED lamps among the optoelectronics, although they may be closer to the ICs and surely benefit from miniaturization. ITRS 2013 recognizes that there are more innovation streams in the industry, but represents them running on three parallel paths, highlighting the synergy between the “More Moore” mainstream evolution, the “More than Moore” enrichment of existing technologies on one side, and the “Beyond CMOS” exploration of new avenues on the other side.

The diversification has an essential role in enabling nanoelectronics to penetrate additional application areas. Over the last 5 years, the O–S–D products grew only marginally faster that the ICs, benefitting in the first place the progress in optoelectronics, and to some extent in sensors/actuators, while discretes grew as fast as the ICs (Figure 3). Nonetheless, the O–S–D TAM represented a business opportunity of about $65 billions in 2014. This is large enough to entice even companies ranking in the top 25 sales leaders to participate, or even to specialize in this segment.

5 Embedding Software

At the beginning of the digital revolution, hardware and software used to be often interrelated and therefore codeveloped; for example, it was desirable to design computer instructions that could be executed during a single turn of the hard disk. Today, complex computing structures are manufactured as an integrated circuit, and it is mandatory to colocate on the same chip the software defining its functionality and thereby build a system on chip (SoC).

For clarification, not all software encountered in the industry matters here; design software tools, either generated in-house or purchased from outside vendors, software systems for manufacturing control, scheduling, logistics, HR, and so on are not of interest for this overview. Likewise, operating systems, Internet-based businesses, or the plethora of applications (“apps”) are usually considered as belonging to a separate industry.

Embedded software is a constitutive element of the products delivered to the customers of the industry and a major contributor to the value created in nanoelectronics. There is a commercial market for “embedded systems,” consisting typically of subassemblies of hardware and software providing well-defined functionality that can be assembled by the OEMs in their end products. It is currently estimated at about $150 billions per year, the value being attributed to both hardware (88%) and software (12%). These numbers are quoted here only as an example. In fact, most embedded systems are captive, being generated inside the nanoelectronic companies and/or by their customers. The value of the embedded systems in the captive production surely exceeds by far the commercial market, being estimated in the range of billion dollars per year; the share between hardware and software may differ considerably from the quoted values.

Absent reliable quantitative data, it shall be noted here that software became an essential competence of the nanoelectronics industry, an essential enabler for the usability of the nanoelectronic products, and for sure one of the elements with an increasing significance and weight in the future.

6 Restructuring the Value Chain

The nanoelectronics value chain has continuously evolved since its beginning in 1956 with the Shockley Semiconductor Laboratory (a division of Beckman Instruments, Inc.), quickly followed next year by the split off of Fairchild Semiconductor (as a division of Fairchild Camera and Instrument Corporation), and then by further 65 start-ups launched in the following 20 years. The technology also diffused through numerous licenses, both for captive production and commercial activities.

7 Opportunities and Perspectives

7.1 Emerging Market Opportunities

Often, the applications that created big surges in demand fueling nanoelectronics growth have been either underestimated or not foreseen at all. The last example is the explosion of smartphones, tablets, and other portable devices that blurred the boundaries between the computing, communication, and consumer market segments. It is therefore risky to state what the next big opportunity will be. Nonetheless, even if the details of the future products are yet to be defined, there are areas in which the growth is likely to accelerate.

A quick overview of the electronic systems market and the component consumption per market segment shown in Figure 4 indicates that in most markets the component penetration is in the range of 25%, except for the segment Industrial/Medical/Other for which it is less than 18% (government applications also show low penetration, but they are a segment too small to matter in this context). The last years have experienced an acceleration of the component consumption in automotive, and this trend is likely to continue under the impact of new technologies enabling various types of electric vehicles, highly automated or even autonomous driving, and on-board infotainment. The “Industrial/Medical/Other” sector however seems to present the biggest opportunity: It can increase its consumption of components by 50% only to be at a par with the other segments. This could well happen within the “Industry 4.0” concept put forward by a European initiative, paralleled by the "Industrial Internet" concept put forward in the Unites States of America. It is based on the observation that the industry historically moved from mechanization to electrification and to information technology, and now has reasons to expect that the next significant boost in productivity and capabilities will occur by merging Internet technologies in the industrial processes. There is almost a unanimous expectation that industry will strongly move in this direction, even if particular implementation examples are still in the process of taking shape.

Likewise, computers were initially intended for about 100 governments, then they became business machines addressing about 50 k corporation, and then they eventually became personal and could interest a billion people. The next step in growing consumption is foreseeable: It will consist in embedding computing capabilities in objects. The “Internet of Things” (IoT) will further increase the number of “users” by one or two orders of magnitude, boosting demand. The concrete implementation cases are still in the process of being defined, but there is quasi-unanimity that the IoT will occur, taking the industry to the next level.

Of course, the unforeseen products and services idea should not be forgotten. The nanoelectronic industry creates opportunities for anybody, located anywhere in the world, to change the world with the force of a good idea.

Independent consultant, former executive director of ECSEL and ENIAC Joint Undertakings Munich, Germany

Dr. Andreas Wild