Silicon Valley, the heart of innovation, has its roots in the growth story of the transistor. With a whole generation of inventors and business ranging from William Shockley, Andrew Grove to Steve Jobs, Silicon Valley gets its name from the exponentially rapid progress made by semiconductor industry in the 70s and the 90s. The industry began with feature sizes greater than 2 square micron, and the self-feeding Moore’s Law (expounded by Gordon Moore, one of the Intel founders) ensured that the best intellects in Silicon Valley figured out ways to cram more functionality into each square millimeter of a silicon die. Today’s minimum feature size of 25 nanometers is equivalent to a 10000 fold increase in feature packing density for a silicon circuit.

Over the course of this steady progress, it appeared that innovation engine could be profitable forever. Over the past 15-20 years, semiconductor startups have tried to tackle problems such as increasing computing performance, improving graphics and user interface performance, and enabling communication innovations such as OFDM (a computation intensive communication signaling process that is bandwidth efficient and tolerant to interference) that would have been scientific fiction two decades ago.

Alas, the silicon industry is hungry for economies of scale more than ever. NRE costs for new semiconductor related ideas that rely on the smallest feature size are skyrocketing and are only sustainable in applications that benefit from enormous market sizes, such as the cellphone or consumer electronics industry. This is not lost on the capital allocators of the valley, the venture capitalists. The stream for semiconductor funding is drying up. This can be seen in Fig. 1 where the early stage startup funding as well as the overall semiconductor funding have been combined. This data is courtesy of the quarterly venture capital survey carried out by PriceWaterhouse Coopers.

The graph shows that semiconductor startup funding as a percentage of total venture dollars is decreasing. What does this bode for the fabless semiconductor industry?

In the past several years, a majority of the semiconductor ventures have been focused on the consumer electronics (such as TV tuners, processors for portable devices, processors for home entertainment etc.) and wireless standards (such as Wimax, femtocells, NFC etc). There are several reasons for clustering of VC backed early-stage semiconductor companies into a few technological areas:

- Semiconductor ventures need significantly more exit valuations because of the high R&D and production costs associated with custom semiconductor solutions. Only huge markets such as consumer electronics can even come close to concentrating such value into a single firm (instead of an ecosystem of firms).

- Those ventures that are attempting to commercialize a new technological break need a long gestation period to account for the numerous trials and errors inherent in cutting edge semiconductor design and to develop the new markets necessary. Here’s an example of such a venture.

The net result is that either the venture money funds early stage startups chasing huge but traditionally very price-competitive markets, or the VC money funds later stage rounds of the riskier ventures. Both in some sense are disadvantaged in terms of eventual returns on investment.

Semiconductor teams that design products combining traditional digital CMOS logic with other MEMS based designs will lead the pack as far as innovation is concerned. MEMS provide a natural bridge between the physical world and the abstract world of digital computing. Additionally, companies that specialize into developing applications for semiconductor designs, like an Ideo for semiconductors, could also harness the potential for innovation while minimizing the impact of the economics of semiconductor manufacturing.