Semiconductor Industry Consolidation Shows Maturation Along Auto’s Roadmap
A wave of business consolidations is currently moving through the semiconductor manufacturing industry, and analysts of all stripes agree that this is another sign of the industry’s natural maturation. Some people fret that maturation may correspond with less innovation, and that semiconductor chips could become commodities like steel. Instead, we should expect the dynamics of maturing to follow the roadmap set by the biggest high-tech manufacturing industry of the last century: automobiles.
After 50 years of Moore’s Law led predictable scaling down to minimal atomic dimension for active transistors in integrated circuits (IC), the roadmap for the semiconductor IC manufacturing industry is no longer simply that of technology advances. While technology will no doubt continue to advance, it will be at a slower pace, and more likely tuned to a specific application. Application-specific IC chips today could include not just ASICs, but also specialty mobile microprocessors such as Apple’s AX series, as well as memory modules for video servers. All of these chips are rated by some metric other than raw transistor-count or circuit speed.
The raw technology of an IC can be equated to internal-combustion-engine displacement and RPM. While the basic V8 motor retained similar appearance and displacement, through clever engineering, the horsepower, reliability and mileage of motors continued to increase over 50 years. We can compare the dynamics of these two high-tech industries by looking at how they have matured over time.
Ages of Industries
In the “learning youth” of the auto industry, the raw power of the engine was the main value-add to horseless carriages, and vehicles were not reliable, comfortable or affordable. Henry Ford’s design of the Model-T in 1907, and creation of the moving assembly line in 1912, signaled the beginning of the next era.
In the “exuberant adolescence” of the auto industry, hundreds of companies developed new technologies so that improved functionalities could be integrated into cars. Between 1910 and 1950, most of the technologies we associate with a modern car were first sold: automatic transmissions, hydraulic brakes, overhead-valve engines, front-wheel drive, superchargers, unibody construction, radios and even air-conditioning.
In the “confident maturity” era, the user experience of powerful technologies was enhanced with comfort and reliability. Innovations in cabin acoustics and ergonomics improved the user experience. Refined products competed in the global marketplace, leading to consolidation to just a few large companies serving each distinct market.
In the “rigid establishment” era of the auto industry, it’s a given that the raw-engine horsepower is there, AND it’s assumed that it is highly functional and reliable, AND the vehicle has to be cost-effective and comfortable, AND we now expect a sophisticated ecosystem to provide service and support for our unique situation. The table below shows these ages of the auto industry in comparison to the semi industry.
In 1949, Bell Labs of AT&T created the world’s first solid-state transistor, and a decade later, both Robert Noyce in California and Jack Kilby in Texas discovered how to create semiconductor ICs. So this was the youth era.
In 1965, Gordon Moore published “Moore’s Law” as a business development tool, and when Bob Dennard published “Dennard’s Law” on scaling in 1974, the way was shown for miniaturization and cost-reduction of technology to allow for integration into new products and markets. So this was the adolescent era.
In 1985, Apple released the Macintosh computer to bring the first intuitive user experience to personal computer users. Also, IC-controlled sensors and actuators begin to be used in autos to improve economy and safety. So this was the beginning of the mature era.
Just after New Year’s 2001, Apple released iTunes (for Macintosh), laying the foundation for the applications ecosystem of third-party developers and for iPod, iPhone, iPad, AppleWatch and AppleTV to interoperate with ever-improving Mac computers. Also, semiconductors in autos add even greater value in performance and infotainment, while improved battery-control subsystems enable reliable electronic vehicles. So we are now well past maturity into the establishment era.
We can guess the future of the IC manufacturing industry by looking at the history of the auto industry. Theoretically, the “rule of three” in mature markets will apply, although with IC functionalities transforming established market dynamics, there will likely be more room for business model innovation relative to the auto industry.
Maturity in an industry generally correlates with predictable growth and profits. The global auto industry was certainly mature in 1950 when 10 million vehicles were produced, and the figure below shows that sustained growth has led to shipping nearly 90 million vehicles in 2014.
Imec, a world-renowned nanotechnology research center headquartered in Leuven, Belgium, has evolved into the world’s default headquarters for all advanced IC R&D, and the annual imec Technology Forum showcases the consortium’s activities for its partners, journalists and analysts. In the last decade, this forum has evolved to focus more and more on applications and cross-functional technology integration across multicultural collaborative partnerships. Technology still provides the foundation for it all, although technology alone is no longer sufficient to solve the next set of the world’s problems.
AMD provided a great example of the new way forward for the IC industry with its massive R&D project to create the 3-D Fiji graphics processing unit (GPU). It is a credit to Lisa Su’s vision as former CTO and now CEO that the company has made the strategic investment to bring a far more comprehensive systems-level solution to the market, instead of just a faster multicore chip. Intel has shown willingness to devote 1,000 engineers to try to win the specialty communications chip slot in the upcoming Apple iPhone 7.
The semiconductor manufacturing industry is actually well past “maturity” at this point, creating ICs that integrate amazing technologies into affordable and reliable products that function within complex application-specific ecosystems. For an established semiconductor fab industry company—whether an integrated device manufacturer (IDM), fabless company, foundry, original equipment manufacturer (OEM), specialty material supplier, electronic design automation (EDA), or outsourced semiconductor assembly and test (OSAT) provider—the barriers to entry for new competitors are nearly insurmountable such that future business is effectively “theirs to lose.” There should be less volatility in the industry, as more application-specific technology integration precludes second-source supply chain strategies, allowing the whole ecosystem to avoid over-building capacity.
Just as most drivers today do not care about engine-displacement and RPMs, but instead care about horsepower and fuel economy, so too do most smartphone users not care about transistor-count and clock-speed, but instead care about video-rendering and battery life. It’s a given that the raw technology is inside there somewhere, AND it’s assumed that it all integrates seamlessly and reliably when in use, AND it’s got to be highly functional yet cost-effective, AND we now expect a sophisticated ecosystem to provide service and support for our unique situations.
All of this ensures fantastic user experiences, which drive continued product sales and services to create new markets and expand old markets, such that revenues for the global IC fab industry should continue to grow faster than the average global GDP CAGR of ~4 percent. Since the industry is >US$300 billion in global annual revenue, a single percent increase is still a few billion dollars. The next 50 years of global semiconductor manufacturing should see sustained growth and profits with application-specific ICs providing distributed intelligence to our world.