Is it time to re-tool the business model?
04/01/2004
Solid State Technology asked experts about the growing perception that R&D is becoming too commercialized while the industry struggles to make innovation profitable.
Managing for innovation
 |
V.V. Zhirnov, R.K. Cavin, Semiconductor Research Corp., Research Triangle Park, North Carolina
Basic research in semiconductor sciences has been significantly under-funded for more than a decade in the US, an especially risky trend for an industry that, from its beginning, has strongly depended on fundamental science. Basic research in solid-state sciences in the 1930s and 1940s, well-supported by governments that were spurred on by defense needs, created the foundation for the electronic century.
Much of modern information technology has been built on the use of this research investment. A need exists for both increased investments in future technologies and the stimulation of latent geniuses by properly formulating and featuring fundamental challenges. Due to chronic under-funding of basic research in semiconductor sciences, it is questionable whether the research pipeline can provide technologies to sustain the cadence of the International Technology Roadmap for Semiconductors (ITRS).
Because of market-timing pressures, the semiconductor industry has focused much of its R&D resources on immediate technology generation. A focus on commercial interests (or more broadly on applications) is not necessarily an obstacle for fundamental research, however, as practical needs often drive fundamental science. For example, thermodynamics emerged as a result of the careful study of heat engines.
The industry has also long recognized the importance of basic science. Ten of 98 Nobel Prizes in physics were issued for research performed in industry; eight were given for research performed in the US semiconductor industry.
The semiconductor industry formed Semiconductor Research Corp. (SRC) in 1982 to increase the efficiency of limited basic research funds by funding exploratory research in universities. Through SRC, the industry is investing between $60–$70 million in 2004. This is a relatively modest sum compared to many government programs, and because of the complexity of challenges identified by the ITRS, the level of funding needs to be substantially increased, probably through increased investments by governments.
While we argue that increased basic research investments are essential for long-term progress, it is reasonable to wonder about the relation between large scientific programs and breakthrough discoveries. Funding is only one factor in the success of fundamental research — Albert Einstein is an interesting case study. He acquired two profound results that later changed the world. The first was the special theory of relativity, including the famous relation E = mc2. The second was the discovery of the concept of the quanta of radiation, and thus the photoelectric effect law. This second achievement, an important milestone in quantum mechanics development, was the basis upon which Einstein received the Nobel Prize. When these two works were published in 1905, Einstein's job was as a technical assistant in the Swiss Patent Office; he had no research funding.
Modern commercial progress has not supplanted the need for such thinkers and other Einstein-caliber geniuses. It is rather the opposite, because as industry grows in an evolutionary way, building on what it has created, there is a need for innovators to look at processes and ideas from a different point of view.
Thus, we need to examine what, aside from his pure genius, prompted Einstein to attack, on his own, these two fundamental areas of research. In his time, several well-articulated, widely known scientific puzzles engaged his creative mind. Examples include:
- Why is the speed of light a fundamental constant?
- Why is it that photoelectrons can be released from a solid by light illumination if (and only if) the wavelength of the light is less than a certain value called the threshold?
- Why is the energy of the released photoelectrons independent of the intensity of the light?
- Why had many attempts by prominent scientists failed to explain blackbody radiation?
It is important to articulate a similar set of fundamental questions related to semiconductor sciences today to engage creative minds. For example:
- Is it possible to remove heat from solid surfaces at a rate >1000W/cm2?
- Is it possible to do useful computation at energy consumption less than kBTln2 per binary operation?
As Jack Kilby, the 2000 Nobel Prize laureate for inventing semiconductor ICs, put it, "The definition of the problem becomes a major part of the innovation." Perhaps this is a clue to how we might engage and support contemporary geniuses. It is important not to merely increase investments in semiconductor science research, but to also clearly postulate and feature scientific challenges that must be met if progress is to continue.
For more information, contact Victor Zhirnov, research scientist, Semiconductor Research Corp., Brighton Hall, Suite 120, 1101 Slater Rd., Durham, NC 27703; ph 919/941-9454, fax 919/941-9466, e-mail [email protected].
The recovery is here, so where's the profit?
 |
Charles DiLisio, D-Side Advisors, Saratoga, California
No one needs to tell those of us who toil in the semiconductor space that this world has changed. In the last three years, the semiconductor business model mutated yet again and we all must change with the market. Three key elements of the new market are changes in market and end-user behavior, lack of broad markets, and constantly rising costs. As we work through the excitement, we will see this model play out.
Much of the big change centers on the "consumer-like" behavior of the new semiconductor market. Consumer-like means short product cycles, market segment fragmentation, and what the trade calls "nonspousal approval" products. Typically, electronic gizmos have price points at $200–$300, low enough so they don't require discussion with the spouse.
The customer has changed. The end-user markets have splintered into smaller segments, and users will no longer pay more for more functions. The consumer attitude is, "Why should I pay for a cell phone with a camera?" This consumer doesn't care about the cost of delivering and integrating new camera functionality with that of the phone, because the phone is free.
Next, the broad billion-dollar markets, with their DRAM-like fab runs, are gone. Still, IDMs search to find new volume markets such as the PC or display to take advantage of declining IC costs by expanding demand. We know that Moore's Law works best in broad markets that respond to increasing price cuts — but what happens when those markets don't appear?
To find a solution, IDMs and fabless companies try to develop a "platform product," the foundation around which the company can build derivatives and develop extensions, as Intel has done with microprocessors. This is the old, overly-familiar crapshoot.
We are also familiar with the difficulty of amortizing rising design and process costs as designs go below 0.13µm. Mask costs are rising to the $900,000–$1 million range; 300mm isn't going to be cheaper nor are process requirements of low-k and copper, as the foundries are finding out. However, that's just entry stakes, as design simulation and test costs raise the bet on a new design in the $15–$25 million range, leaving the platform game for the most experienced gamblers.
The OEM customer also has changed from a fellow engineer to a contract manufacturer who will only pay for functions required, nothing more. Added functionality is also seen as a negative because of the potential to add cost from more engineering requirements.
With all these changes, firms in the semiconductor value chain need to rethink their antiquated business models to understand how they will make money and not lose margin in 2004's market. Often, a new profitability strategy is required.
Profit is in the details: paying attention to a profit rule, which deals with the interrelationship between market opportunity, competitive actions, and investment. The profit rule is a function of both tangible costs — which include the real costs for development and design, plus other costs of business— and the intangible business circumstances of market opportunity, competition, and investment.
 |
Market opportunity = projected total available market; competitors = number of existing and projected competitors. Investment = that required to create the chip; in this case, we can use design costs as a surrogate. The result is a probability for product profitability index. Profit should be the goal. Volumes and market share are nice to have, but profit insures the life of the business. We believe that this rule needs to be ≥10 to consider the project worthwhile. The trick is how to keep that value, rather than pass it down the chain.
Blindly following the IC manufacturer's quest for a platform product, volume markets became a siren song in the new millennium. Volume represented great revenue and profits; however, getting the volume required large capital investments, while continually looking for ways to squeeze equipment suppliers. Along the way, IDMs and foundries tried to maintain their margins, forcing their vendor companies to "take it or leave it."
Today, those living along the semiconductor value chain are experiencing the worst of Moore's Law — increasing costs and rapidly declining pricing that cannot be amortized over larger and larger volumes because those large markets are gone. Along the IC value chain, the power of IDMs, foundries, EMSs (electronic manufacturing services or contract manufacturers), and OEMs continues to drive down margin and lower ROI. The final blow: EMSs and OEMs don't care about functionality or performance, but are more concerned about reliability, interoperability, or ease of design-in at their cost.
Once again it's a brave new world where luck and good business models play as great a part in success as do ideas and great engineering.
For more information, contact Charles DiLisio, president of D-Side Advisors; ph 408/255-4620, e-mail [email protected].