The semiconductor industry is at an inflection point that will determine its course for years to come, according to an informal survey of respected industry executives. The days of fabless semiconductor startups’ receiving $100 million or $200 million to develop new generations of complex system-on-chip products are over. Venture capital firms have not received the returns required to justify such investments, and the partners who championed them are out.
So, what’s next? Will there be more semiconductor startups? How will they be funded? And in a severely curtailed funding environment, where will the groundbreaking innovations come from?
To appreciate the industry’s dilemma, consider the divergence between the cost of IC development (Figure 1) and the trend line for early-stage investments in startup companies (Figure 2).
After interviews with more than 25 individual stakeholders in the industry—including venture capitalists, established semiconductor company CEOs and startup CEOs—we came to the conclusion that emerging trends could transform the semiconductor sector from a vibrant, innovative engine of the technology economy to a slow-moving purveyor of “me too” products.
From our conversations, we identified three significant trends. First, the consolidation of smaller industry players into larger companies is proceeding at an accelerated pace. In fact, most exits by semiconductor companies over the past decade have been through acquisition; IPOs have been almost nonexistent. Even midsized public companies have been swallowed by large consolidators over the past few years.
Second, the VC-backed companies that have succeeded in raising new money are those that have specialized in a particular niche. Specialty analog and mixed-signal IC companies are prime examples; the larger geometries and lower integration levels typical of such devices have kept their development costs reasonable.
Finally, startup activity around SoCs has nearly dried up because of the extreme expense involved in developing those highly complex devices.
What kind of industry does that leave us? Regrettably, one that provides much less room for startups and venture capital. Startups may still play at the fringes of the industry, serving small markets with limited growth potential; some with significant intellectual property might manage to make themselves into IP module providers in the mode of ARM and MIPS. Neither scenario provides the blockbuster financial outcomes that fuel a vibrant VC-funded industry.
According to at least one highly respected CEO, the chip industry might look increasingly like the auto industry: mature, slow moving and dominated by a handful of huge companies. Other industry insiders believe the pharmaceutical model could prevail; small development-stage companies would create promising intellectual property and then sell it to entrenched players with the resources to bring it to market. We even spoke to one industry exec who predicts the return of big company laboratories—think Bell Labs, Xerox Parc and Sarnoff. In these scenarios there is no place for VCs to put their cash, and no place for truly disruptive technologies.
Is it inevitable that our industry will continue to mature to the point of marginalization, or are there other possible outcomes?
Looking back at the most recent wave of semiconductor startup successes provides clues to alternative futures. Broadcom and Marvell are good examples of companies that were formed in the 1990s and achieved major success in the last decade. Two elements came together to enable their success in the market.
First, both companies were fully committed to the fabless semiconductor model. That strategy freed up an enormous amount of capital that otherwise would have been dedicated to building huge fabrication facilities. More-efficient use of capital was necessary to their success but was not sufficient, in itself, to enable it. It had to be accompanied by disruptive innovation.
The disruptive innovation that companies like Broadcom and Marvell brought to the market was the teaming of system developers (communications systems engineers and scientists) with IC circuit developers—the raw talent required for system-on-chip development. Nearly every semiconductor executive we talked with emphasized that they now have more software developers on their staffs than IC designers. That shift is a direct outcome of the innovation brought to the industry by the SoC developers.
The companies now coming to market with new ideas for SoCs are not engaging in disruptive innovation; rather, they are sustaining innovation. Yes, they bring new ideas and new system techniques to silicon, but the existing SoC industry players are very capable of doing the same.
When Broadcom and Marvell entered the scene, the entrenched semiconductor industry players didn’t employ engineers who could design innovative adaptive filters or sophisticated modems and Reed-Solomon codecs. That was the job of the system OEMs. Now, nearly every semiconductor firm has a cadre of system engineers and developers; and because of escalating development costs, the advantage is with the incumbents.
When observed through this lens, trying to bring new investment to SoC startups does not make sense. We are better off asking what the disruptive-innovation event in semiconductors might be. SoCs are the last event, not the next one.
The executives we met all agreed that there is still demand for new innovation in semiconductors. They cited the ever-increasing need for bandwidth in communications and consumer electronics; for lighter, more efficient power sources in mobile devices and automobiles; and for new and smarter ways of delivering energy to consumers and businesses. What’s missing, they believe, are truly groundbreaking ways to deliver those capabilities.
So what do they think might reenergize semiconductor startup activity?
One possibility cited by several in our survey is a shift to the “chipless” semiconductor company pioneered by industry leaders like eSilicon and possibly Global Unichip. That movement could significantly reduce the up-front capital required to bring new semiconductor products to market.
Some CEOs said the integration of sensors and actuators onto semiconductors is gaining momentum and could be the next significant event in higher levels of integration. The ability to add gyroscopes, accelerometers, camera focusing elements, microphones and magnetometers to chips, especially if it can be done in standard CMOS processes, could be groundbreaking.
Indeed, in the view of some, microelectromechanical systems will allow this next level of functional integration and as a result will cause the next big surge of disruptive technology in semiconductors.
No one can predict with confidence what the next wave will be or when it will occur. But based on past industry experience, if there is to be a next wave, there is someone working right now on the breakthroughs that will disrupt the industry.
Can the semiconductor industry recapture the excitement, dynamism and energy we had not so long ago? The answer will be determined by the innovation decisions entrepreneurs are making right now.
We are at a turning point. Around the corner is either the next surge of semiconductor innovation or a semiconductor industry that will consolidate and focus on cost cutting—most likely outside of the United States.
About the Authors
Bruce Kimble is a semiconductor industry specialist and Marty McMahon a cleantech industry specialist with McDermott & Bull, a multispecialty executive search firm with offices throughout California. Matt Rhodes is a longtime executive and investor in the semiconductor industry. This article grew out of the authors’ concern about the state of funding for new innovations and startups.
|Figure 1. Cost (in millions of dollars) of developing integrated circuits as a
function of geometry. |
Source: Silicon Knowledge, Matt Rhodes
|Figure 2. First-sequence VC investments in semiconductors as a percentage of chip industry revenues.|
Source: www.gsaglobal.org, www.pwcmoneytree.com, www.nvca.org