SAN FRANCISCO: Moore's Law, the engineering principle which guides the development of microprocessors, could soon prove ineffective, according to a group of Intel fellows.
Speaking at a panel discussion during the company's 2012 IDF conference in San Francisco, Intel senior fellow and director for process architecture and integration Mark Bohr said that the rule, which predicts that the number of transistors on a chip will effectively double every two years, could be exhausted by 2023.
"I do not accept that it has to happen, but it probably will," Bohr told convention-goers.
"I do not see the end of Moore's law for about 10 years."
The end of the law, whose creation was credited to long-time Intel boss Gordon has long been feared by chipmakers. As fabrication methods become more precise and the size of transistors grows even smaller, fundamental issues such as energy efficiency threaten future developments.
While Intel has been able to keep up the pace of Moore's law through the development of components such as the hi-K transistor, many feel that the ceiling could soon be reached for silicon computing.
"We are trying to invest in longer range research to look at what can be another option for the transistor," said Intel fellow for technology and manufacturing Ian Alexander Young.
"We are not saying it is imminent, but we are hedging our bets."
The Intel fellows are hardly in bad company with their assertion. Moore himself suggested in 2005 that the law had 10 to 20 years of validity left.