Published: July 11, 2025
In the previous volume, I outlined the horizontal specialization in the semiconductor industry and showed the growing presence of fabless semiconductor companies. In this volume, I would like to look at the historical background of the emergence of fabless semiconductor companies.
Background of the Emergence of Fabless Semiconductor Companies—Changes in the Semiconductor Business
Contributions Outside the Semiconductor Industry
Emergence of Fabless Semiconductor Companies
Volume 14: History of Horizontal Specialization in the Semiconductor Industry—Emergence of Fabless Semiconductor Companies
Here I would like to look at the background of the emergence of fabless semiconductor companies.
As explained in the first volume, the number of elements implemented on a chip has increased exponentially following Moore’s Law. This increase in quantity was not only a quantitative increase, but also a major qualitative change in the nature of the semiconductor business. I believe this was a major factor in the emergence of fabless semiconductor companies. Let us look at this in more detail below.
Before the invention of the integrated circuit, semiconductor companies manufactured single elements called transistors, which are just general-purpose components. Like capacitors and resistors, they are a kind of electronic components.
At that time, the role of semiconductor companies (or semiconductor divisions of general electronics manufacturers) was to manufacture components called transistors, while the role of their users was to design circuits and systems. In the first place, what was on a semiconductor chip was not circuits, but a single element called a transistor. Therefore, there was no room for fabless semiconductor companies, which specialize in designing circuits and systems that integrate several circuits, as discussed in the previous volume. Semiconductors were about physics, not circuits and systems.
Even today, with well-developed horizontal specialization, integrated device manufacturers (IDMs) are still the main manufacturers of single discrete devices such as transistors, called discrete semiconductors.
It is worth noting that with today’s progress in the horizontal specialization, as there are companies that only undertake manufacturing, there are also discrete semiconductor fabless companies that specialize in transistor design. However, at a time when it was common for semiconductor companies to have their own semiconductor fabs and the concept of horizontal specialization did not exist, although I cannot say for sure, I believe that there would have been almost no fabless business model that specialized in design.
An integrated circuit was invented in the late 1950s, but large circuits were not immediately implemented. In the beginning, the number of integrated elements was only a few dozen, or at most a few hundred. Although there may be some specialized ones, the circuits implemented were basic, general-purpose circuits. It remained a general-purpose component.
For example, a typical product is called a standard logic IC, which contains several basic logic gates, as shown in the figure below, and is still sold by several companies. The following picture is an example.
Users were responsible for designing the entire system, and they realized a dedicated system by arranging general-purpose integrated circuits, such as the standard logic ICs described above, on a printed circuit board. Semiconductor companies are basically manufacturing companies that make general-purpose components, and a fabless company specializing in design would have made little or no sense. Standard circuits are the same no matter who designs them, and there is no way to differentiate them by design.
However, after the invention of integrated circuits, the number of elements on a chip continued to increase, and Moore’s Law was proposed in 1965 as discussed in Volume 1. About 10 years after the invention of integrated circuits, around 1970, it became possible to have more than 1,000 elements on a chip, and a single-chip LSI for calculator was introduced in 1971, as discussed in Volume 4. It can be said that a silicon chip underwent a “qualitative change” from a mere “component” to a “system.”
The design methods used at the time seemed to be mostly manual, relying on the specialized skills of engineers in each semiconductor company, and it was almost impossible for an outside engineer to design. The manual process also made it difficult to respond to all requests for the development of chips for calculators with custom designs. In the case of calculators, customization was achieved through software, and the hardware (circuitry) was generalized to solve this problem. The generalized hardware is a microprocessor.
Although semiconductor devices were reaching a level where they could implement a system, design methods had not yet caught up with them, and it took a little longer for fabless semiconductor companies specializing in design to emerge.
I joined the company in 1981, so I don’t know the design methods of that era, but I remember the scene in my rookie years: Designers were drawing transistors and wiring in pencil and coloring them in with colored pencils on a large sheet of graph paper made of a special kind of paper called Mylar sheet.
As predicted by Moore’s Law, the level of integration increased exponentially each year. The number of elements that could be integrated on a silicon chip increased dramatically, roughly from about 1,000 around 1970 to about 100,000 by the end of the 1970s and more than 1 million by the late 1980s. A particularly large scale of LSI was called VLSI (for Very-Large-Scale Integration) at the time, although the term is not often used today.
With this level of integration, a system rather than just circuits could be integrated on a chip, making it possible to realize the electronic circuits of various equipment with one or a few chips of integrated circuits. However, because of the manual design process described above, it was practically impossible for semiconductor companies to design all the dedicated VLSIs that users require.
To solve this problem, the design method was innovated and computer-aided design automation rapidly advanced. Design automation tools were started as in-house software specific to each semiconductor company but have been replaced by software from specialized third-party vendors. This is also a kind of horizontal specialization. Although large computers had previously been used, advances in semiconductor integration made computers smaller, and engineering workstations (EWS), compact computers designed for engineering tasks, became widely adopted.
In addition, an efficient method was developed to make a dedicated VLSI for each user’s equipment. Not described in detail here, they are semi-custom ICs called gate arrays or standard cells (or cell-based ICs) and user-programmable logic products called programmable logic devices (PLDs). Field programmable gate arrays (FPGAs), mentioned in Volume 5, are a type of PLD.
These changes have created an environment where users outside of semiconductor companies (customers, students, researchers, etc.) can design their own VLSI.
Another important change resulting from the progress of miniaturization is the sharp rise in factory construction costs that accompanied the progress of miniaturization, as briefly mentioned in Volume 5. This made it extremely difficult for startup companies to build factories and enter new markets, and at a certain point, startup companies had no choice but to become fabless to enter new markets.
The changes in design methods described above have not been achieved by the efforts of the semiconductor industry alone.
The 1980 book Introduction to VLSI Systems by Professor Mead of the California Institute of Technology and Conway of Xerox Corporation (preprint versions were published in 1978 and 1979) is said to have been a major influence. Their revolution in design methods is known as the Mead and Conway revolution. This book brought the details of VLSI design to people outside the semiconductor industry for the first time. It seems that classes were held at universities where students actually designed chips according to the book and evaluated the complete chips. Mead and Conway received Electronics magazine’s Achievement Award in 1981.
Professor Mead (Professor Emeritus since 1999), introduced in the first volume on Moore’s Law as the person who coined the term Moore’s Law, was awarded the 37th Kyoto Prize in 2022 for his contributions to the semiconductor industry. The reason for the award is “Leading Contributions to the Establishment of the Guiding Principles for VLSI Systems Design,” and Professor Mead’s contributions are considered to be very significant.
Note: Professor Mead’s official title is Gordon and Betty Moore Professor of Engineering and Applied Science, Emeritus, California Institute of Technology.
Note: Kyoto Prize: The Kyoto Prize is an international award that originated in Japan and was established in 1984 by the Inamori Foundation. The Inamori Foundation was founded by Kazuo Inamori, the founder of Kyocera Corporation.
I watched the video of a memorial lecture by Carver Mead, Professor Emeritus and laureate of the 2022 Kyoto Prize, entitled ‘Carver Mead “A Personal Journey Through the Information Revolution”’ on YouTube, where he talked about his consulting work for Intel founders Gordon Moore and Robert Noyce. It’s no wonder he coined the term Moore’s Law, as mentioned in the first volume, and I think his consulting work gave him an insight into the realities of the semiconductor industry that led to the Mead & Conway revolution and the publication of Introduction to VLSI Systems.
Let’s take a closer look at when and how fabless semiconductor companies were founded.
What is the world’s first fabless semiconductor company? I have searched the Internet to find out, but there seems to be no definitive answer. The website of LSI Computer Systems, Inc., a U.S. company founded in 1969, describes itself as “The first fabless Semiconductor Company” (as of April 2025). It appears certain that fabless semiconductor companies existed at least as early as 1969, but this seems to be a special case. A search on the Internet shows no evidence of fabless semiconductor companies being established for some time after that, and the wave of full-scale fabless emergence came in the 1980s. The 1980s was the era of VLSI, which came after the publication of the aforementioned Introduction to VLSI Systems in 1980.
Major fabless semiconductor companies founded in the early 1980s, and their founding dates are listed below. Altera was founded in 1983, Chips and Technologies (hereafter C&T), Xilinx and Cirrus Logic in 1984. All are US companies founded in Silicon Valley. It is reasonable to consider these years as the beginning of the subsequent rise of fabless companies. Of these fabless companies, Altera and Xilinx are PLD companies such as FPGAs, etc. C&T and Cirrus Logic were successfully making chips for PCs at the time.
Silicon Valley is a cluster of high-tech companies located along the southern shore of San Francisco Bay, including the cities of San Jose, Santa Clara, Sunnyvale, Cupertino, Palo Alto, and Mountain View, among others.
In the case of C&T and Cirrus Logic, the launch of the IBM PC in 1981 and the start of the PC industry were also factors in the emergence of the new companies mentioned above; they expanded the range of applications for which semiconductors could be used and increased the demand for semiconductors.
As mentioned in Volume 5, Xilinx and Altera were the two strongest FPGA companies, but Altera was acquired by Intel in 2015, and Xilinx was acquired by AMD in 2022. C&T was also acquired by Intel in 1997.
Note: Intel acquired Altera in 2015, as mentioned above. In 2024, it announced the launch of Altera as a standalone subsidiary, and in April 2025, it announced the sale of a majority stake in Altera. It is worth noting that Intel will continue to hold the remaining shares. The transaction is expected to close in the second half of 2025.
Since TSMC, the world’s first foundry, was established in 1987, foundries specializing in contract manufacturing did not yet exist at that time, and manufacturing was outsourced to IDMs. Our company (the semiconductor division of Ricoh Company, Ltd., which was the predecessor of the former Ricoh Electronic Devices) also did business with several fabless companies. Several other Japanese companies also contracted with fabless companies to manufacture their products. Of the fabless companies mentioned above, at least C&T, Xilinx, and Cirrus Logic used Japanese companies. I think the scale of the business at that time was such that it could be handled by using idle manufacturing capacity at IDMs. Unfortunately, no dedicated foundries for contract manufacturing were established in either Japan or the US during this period.
Incidentally, among the fabless companies in the top 10 by semiconductor revenue discussed in the previous issue, Qualcomm was founded in 1985, Broadcom in 1991, NVIDIA in 1993, and MediaTek in 1997. Note that Qualcomm was not a fabless semiconductor company when it was founded in 1985.
Finally, the background of the emergence of fabless semiconductor companies specializing in design can be summarized in the following bullet points.
I think we can summarize the above as 1 and 2 expanding the possibilities for startups to be born, 3 to 6 increasing the feasibility of fabless companies specializing in design, and 7 making fabless a necessity for startups.
In this volume, I have looked at fabless semiconductor companies that are increasing their presence in the market. In the next volume, I will take a closer look at the foundries that played a leading role in the “6. Existence of a company to undertake manufacturing” described above.
Click below to read this series.
Semiconductor Miniaturization:
Volume 1: Semiconductor Miniaturization: What is Moore’s Law?
Volume 2: Semiconductor Miniaturization and Manufacturing Process
Volume 3: Semiconductor Miniaturization and International Technology Roadmap
Volume 4: Semiconductor Miniaturization and Semiconductor Business
Volume 5: Semiconductor Miniaturization and Semiconductor Business (Part 2)
Volume 6: Semiconductor Miniaturization and Semiconductor Devices
Volume 7: Semiconductor Miniaturization: What is MOSFET Scaling?
Volume 8: Semiconductor Miniaturization: Limitations of MOSFET Scaling
Volume 9: Semiconductor Miniaturization and Analog Circuits
Shift to Larger Diameter Silicon Wafers:
Volume 10: Shift to Larger Diameter Silicon Wafers: How a Common Material, Silicon, Became a Main Player
Volume 11: Shift to Larger Diameter Silicon Wafers (Part 2): How Silicon Wafers Are Made
Volume 12: Shift to Larger Diameter Silicon Wafers (Part 3): Reasons and History
Horizontal Specialization in the Semiconductor Industry
Volume 13: Horizontal Specialization in the Semiconductor Industry and the Rise of Fabless Companies
Volume 14: History of Horizontal Specialization in the Semiconductor Industry–Emergence of Fabless Semiconductor Companies