No technology remains forever. The venerable buggy whip gave way to other forms of vehicular acceleration. Similarly, the very fabric and namesake of Silicon Valley may be under assault.
One of the few technical tenets of the semiconductor industry that has crossed over into mainstream vernacular is Moore’s Law. Not really a law, it’s a profound observation that the density or complexity of chips doubles every 18 months. For decades the notion has guided the strategy and investment of an entire industry. A key attribute of this, to date, inevitable advancement is that the cost per gate of silicon is less with each generation. Many predict now this is no longer the case. As the bleeding edge processes are deployed, it appears that the cost per gate on silicon is actually increasing.
Is this the end of Moore’s Law? Not quite yet. The concept is likely to continue to hold water, uh sand, for at least another generation or two. That said, the relentless march towards lower cost and greater efficiency may have been trumped by the physical limitations of electrons flowing through silicon in the same predictable way the industry has come to expect for over 50 years; transistors are about to cost more for the first time since the first one was built in Bell Labs in the ’50s. Further, the actual cost to design each new chip is growing at near geometric rates. Fewer (albeit larger) are being attempted each year.
So, I guess that means Silicon Valley can close up shop. Well, no. The worldwide appetite for semiconductors increases, seemingly, daily. But the world won’t be so thrilled about paying more money for less area. What will fill the void?
If nothing else, the semiconductor industry is renowned for some of the world’s boldest innovations. It has solved countless problems dating back to its inception with uncanny predictability. One could conclude that it’s the very challenges inherent in the semiconductor development process that have resulted in the breakthroughs that, today, enable a microelectronic revolution that reaches every corner of the globe.
As an example, one solution among several is, so-called, 2.5D packaging. Essentially, this technology allows for the simultaneous use of old and new chips in one complex package. It’s gaining favor because it can facilitate reusability, sharing among developers, lower development costs that match application needs and improved time-to-market. It may or may not be cheaper depending upon the amount of technology it aggregates from the entire system: absorbing memory and IO, and reducing the system footprint. However, it does allow the developer to retreat from the bleeding edge of chip development; the same edge that is challenging the economics of Moore’s Law and otherwise causing the semiconductor industry to pause and reevaluate its R&D investment strategies. It may provide an alternative to developing smaller and smaller transistors at increasing cost.
In any case, innovation is the life force of the semiconductor industry. Moore’s Law is the proxy for the greatest run in the history of modern product advancement, but that run may be showing signs of financial diminishing returns. However, the creativity and tenacity of the engineering community may carry the day one more time as the once-pedestrian chip package, in the form of 2.5D, becomes the enabler of a much better understood and sexier slice of silicon.
Time will tell. But don’t be shocked one day to hear we’ve spent our lives working in Packaging Valley…