From the mainframe to the mini to the PC to the smartphone, our computing devices just keep getting smaller, faster, and cheaper. Each of these shifts depended on advances in semiconductor technology and the ongoing impact of Moore’s Law, as individual transistors gave way to the integrated circuit, and then the sophisticated system on a chip. And at every transitional stage, new companies emerged, became dominant, and then sat back as others developed the technology that drove the next iteration, outcompeting their predecessors for market share and profits.
Now comes the Internet of Things (IoT), once again transforming the semiconductor industry. This time, however, it’s different. Rather than being dependent on a single kind of device, the market for IoT-oriented chips is highly diverse. Semiconductors are needed for a wide array of aspects of cloud integration and connectivity, including computing, sensors, communications, and interactivity. Estimates of the impact of the IoT range wildly, but by every account it will be enormous; the installed base of connected devices is forecasted to double or even quintuple through 2020.
With the IoT, semiconductor innovation is no longer driven primarily by Moore’s Law, which predicted the rapid escalation in the number of transistors on each chip and the concomitant increase in performance speed and decline in cost. The value of the chips required to power the IoT is not measured in sheer speed; other factors, such as power consumption, miniaturization, software, configurability, and durability, are more important. This is because in the IoT, the chips are being put to a multitude of very targeted uses in a highly diverse group of industries, including automobiles, manufacturing, utilities, and others. That large variety of unique applications has in turn led to a proliferation of companies in the semiconductor industry — some large, some small, virtually all of them “fabless” — that have entered the fray to design chips for ever more specific purposes. And this has entirely transformed the industry’s competitive dynamics.
“Strategy That Works - The key to success is figuring out how to monetize one’s capabilities.”
The semiconductor industry is no longer dominated by a handful of the largest players. Companies must now follow a different path, competing not so much on their performance edge or lower price, but on making the best chips for specific purposes, what data capture and communications services their chips enable, and their ability to collaborate with other players across the IoT value chain.
This, in turn, is putting an increasingly large premium on how semiconductor companies decide to approach their markets — what their “way to play” is — and on the distinctive capabilities they deploy to carry out that approach. Semiconductor companies must be able to distinguish themselves in new ways, for example, the ability to partner with downstream players, to work flexibly with customers, and to base business models on the value not just of the chips themselves but of the software and IoT services built into them. And some companies are already working to become what we call “supercompetitors” — companies that use their differentiating capabilities system to shape their future and realign their industry around themselves.
The IoT presents a particularly daunting challenge for traditional semiconductor companies. In the first place, many continue to focus on their largest business segments — desktops, communications, and the like — a distraction from the opportunities offered by the IoT. Second, new market segments for IoT chips demand a much higher level of design heterogeneity and device integration, affecting not just the chips themselves but also what software is needed to run the chips, how the chips are integrated into their respective devices, and the way components are packaged within the devices, which would impact, for example, the casing surrounding the chips (see Exhibit 1).
What’s more, companies must also take into account the many industry verticals for which chips must be designed and produced; each vertical has different go-to-market approaches and customer requirements. Finally, companies are facing a lack of interoperability standards and customer concerns over IoT security.
These challenges are why most semiconductor manufacturers remain cautious in their response to the IoT. How they approach this new development varies: Some are building IoT-dedicated business units and alliances, others are adapting products and rebalancing product portfolios toward faster-growing segments, and still others are building scale through mergers and acquisitions. Some are pursuing all three approaches. However, certain players are taking bold steps to define and pursue a logical forward-looking strategy. These are the companies that have devised a consistent, coherent way to play, through a firm understanding of how they intend to create value — for their customers and for themselves — and that are busy putting together the distinctive capabilities and the product and service portfolio they need.
Given the opportunities and constraints of the IoT, the largest integrated and fabless semiconductor industry players are choosing one of three primary ways to play that will be viable at scale, each of which requires different capabilities and targets different customer segments (see Exhibit 2).
The design, marketing, and sales of semiconductor devices — what the companies above are primarily engaged in — have been largely decoupled from the actual fabrication of the chips. Fabrication has become increasingly geared toward the massively scaled manufacture of specific designs, and this will remain an essential capability for the foundries. However, new IoT-focused ways to play will emerge here as well. Commodity volume fabrication players, focusing on producing low-density chips in high volumes, will likely begin to have an impact, as will boutique foundries that focus on high-end applications at lower volumes. Finally, some fabrication players may even choose to move up the semiconductor value chain to provide packaging and intellectual property as well, focusing on a higher level of service provision.
The transformation in the semiconductor industry toward distinct ways to play will not be easy. Many companies are still participating in too many industry areas — they are pretty good at many things but truly great at nothing. This will become a problem, especially as they are forced to compete against an ever more diverse set of competitors. Companies aiming to become ecosystem enablers, for example, will find themselves competing with software, cloud, and open source players as well as companies that focus on licensing their IP, all of which will seek to build ecosystems enabling cross-industry solutions (see “The Strategist’s Guide to the Internet of Things”). Those wishing to become industry application leaders will need to compete with subsystem suppliers and collaborative design partnerships among the OEMs. And lean, rapid portfolio shapers will need to pay close attention to their customers’ needs as they reshuffle their portfolios to position themselves in profitable niches.
All these challenges are likely to result in continued M&A activity — following up on the $130 billion of global deals made in the semiconductor sector in 2015, a record-setting year, and on nearly as high a value of deals in 2016 — as companies merge with or buy others in order to fill gaps in both their capabilities and their product portfolios.
For example, in its efforts to become a broad ecosystem enabler, Intel has been especially active on the acquisition front. The company bought Altera, a designer of programmable chips for data centers, industry, and the IoT; artificial intelligence player Nervana Systems; and Itseez, which makes chips that enable computer vision. Each of these companies provides a critical piece of the IoT ecosystem puzzle.
That’s also why, before its acquisition by SoftBank, ARM purchased Sansa Security and Offspark, both makers of security software for the IoT, integrating their security technology into ARM’s mbed operating system, and extending the relevance of ARM’s designs to security-critical IoT applications for the automotive and other industries.
Given the far greater number of types of chips needed to build out the IoT, and the increase in the number of ways to play in the IoT’s competitive sphere, there will likely be even more winners among the companies providing these chips. In many ways, smaller companies will have a much greater advantage than they did when Intel was king, since barriers to entry in fabless design are much lower, and they can concentrate more narrowly on specific verticals and chips for applications and niche markets.
The current challenge for the largest semiconductor companies is their pursuit of several markets at once. These companies will find it increasingly difficult to compete for customers against businesses that can afford to be more decisive and focused. Many semiconductor companies, for example, currently maintain large units dedicated to cross-industry security, even as they try to pursue customers in specific industry verticals. Such companies’ investment and operating decisions could be difficult to reconcile, and their various business units may be unable to support one another in their efforts to build the different capabilities needed to succeed in carrying out these different ways to play.
Given the nascent state of the IoT, have any true supercompetitors emerged in the semiconductor industry? Not yet. The chip sector dedicated to implementing the IoT is simply too diversified — in terms of products and services, of standards, and of industry verticals that need to be served — to have coalesced around a few top companies. It’s as if the semiconductor industry has turned back the clock to resemble the U.S. auto industry of the 20th century, when more than 1,500 carmakers opened up shop and then quickly disappeared, succumbing to pressure from the Fords and GMs, the supercompetitors of the time that had perfected vertical integration and the assembly line.
Similarly, in the chip landscape, new supercompetitors will eventually arise. They will be the companies or consortia that succeed in building out a dominant cross-industry IoT ecosystem, or that develop a winning combination of chips and services for a particular industry, or that offer a flexible line of products and services to rapidly meet customer needs.
Getting there will require ambitious companies to be decisive about their value proposition and the way to play needed to create real value in this new world. They must understand in detail the capabilities they need to carry out that strategy, perfecting existing capabilities and building, buying, or collaborating to gain new ones. They must learn to scale up those capabilities to the degree necessary to make a profit in their chosen markets, and they must understand the degree to which the capabilities they do build will be valued by customers in those markets.
In other words, as in every other industry, the key to success is figuring out how to monetize one’s capabilities. The IoT offers companies in the semiconductor industry a wide-open field for competitive advantage, or for irrelevance.