The End of Moore’s Law…No Really This Time
By: Scott Aughenbaugh
For decades we have benefited from Moore’s Law—the predicted doubling of computer power every two years per unit of cost. Now, the leader of DARPA’s Microsystems Technology Office and others are predicting the demise of Moore’s Law in a decade. That prediction is generous— Moore’s Law will cease much sooner.
One cause for the coming deceleration in expanding computing power is a problem producing cost-competitive chips in larger quantities. To overcome this problem, some have considered increasing the size of the silicon wafer from 300 mm to 450 mm, roughly the difference in diameter between a vinyl record and a basketball hoop. Although it may seem like a modest increase, the change coupled with improvements in advanced lithography could put Moore’s Law back on schedule, at least until we reach the physical limits of the technology later this decade. With estimates for building new fabrication facilities to create these larger silicon wafers well above $10 billion, and only a few currently breaking ground, it will be a decade before this new production technique takes off.
The impending expiration of Moore’s Law comes at a time when budgets are stretched thin and cost efficiency is as important as ever. Continued improvement in computational abilities would benefit the millions of users in the government and commercial sectors whose computers and operating systems are outdated (approximately 10 percent of U.S. federal government computers still run on Windows XP, for instance). More critically, realizing the potential for significant advances in virtualization, phones and wearable devices, weapons systems, smart city initiatives, health informatics, and other areas depend on our continued improvement in computational abilities. The end of Moore’s Law should thus signal to lawmakers and companies the need to invest more heavily in the research and development of 3D designs, photonics, materials beyond silicon, and eventually the quantum realm.