Andy Grove, chairman of Intel Corp. today proved himself to be a technologist rather than economist, declining to offer a prediction for economic recovery and promoting potential future innovations. Speaking at the International Electron Devices Meeting (IEDM) in San Francisco, Grove predicted continued uncertainty in the chip industry.
"I have no way of guessing where we are on the cycle," Grove said. "We are not at the beginning but at the beginning of the end? I would not be so optimistic."
Grove pointed to a 6 percent unemployment rate, possible war in Iraq and economic meltdown in South America as potentially destabilizing factors to the economic health of the industry at large.
"[This recession] is unique because it follows a period of incredible overbuilding," Grove said. "In the late 1990s we saw a building of communications capacity, of fabrication capacity, manufacturing capacity. The industry, in retrospect, was operating way ahead of the underlying demand."
However, Grove maintained that the current downturn has not been as cruel to Intel as the 1984-85 downturn, when Intel’s margins were reduced 40 percent. Other than stock prices, Intel’s financial operations have not suffered as badly as they did in 1984-85, he said.
Touting the party line that innovation is key to coming out of a downturn on the upside, Grove’s keynote speech today examined the technological progress made within the manufacturing industry and how this has led to increasingly innovative and powerful chips. However, as chip making begins to test the physical limits of Moore’s Law, Grove explained why he believes this boundary should be tested and conquered.
"The question we have to ask at this point is why bother? Certainly not because we need more capacity. The reason we have to bother is because silicon devices have been and continue to be the foundation of the digital spiral that characterizes this industry."
Grove said his concept of a digital spiral represents the future innovations that will form the cornerstone of the future chip industry. Foreseeable applications for next-generation chip technology include probablistic computing, wireless everything, natural I/O and a killer environment, he said.
Pervasive computing will become enabled through a wireless everything model, with every device that could benefit from having an RF transceiver being built with one. Probablistic computing takes advantage of the gains in raw computational power by creating more precisely refined models and systems, such as Web sites that correct the user’s entry errors, or refinements of current database practices. For example, shopping sites that currently use proprietary databases would use probablistic computing, taking advantage of computational power to provide precise matches, rather than search a database to provide approximate ones. One future technology that Grove sees semiconductors as enabling is blood type analysis, with proteins in human blood being analyzed to identify disease and also diagnose appropriate treatments for that disease. Some of the ideas Grove discussed are admittedly a long way off, requiring significant improvements in microchip technology’s ability to perform MIPs. But Grove is optimistic the technology is already in place to eventually lead to such future innovations.
However, alongside this optimism, Grove and his colleagues at IEDM have been open about some of the challenges facing the IC industry, especially as the limits of Moore’s Law are tested.
"Power is going to be a significant challenge," said Mark Bohr, Intel senior fellow. "Coming up with new materials like high-k gate dielectric will help provide another generation or two respite from that problem. Sub-threshold leakage, there really is no fundamental solution to that problem so I think we have to find a chip architecture and design solution."
Both Grove and Bohr said that future breakthroughs would require a more efficient use of current semiconductor technology and also the development of entirely new technologies.
"Generations at least for Intel are coming every two years or so," Bohr said. "We have things in the pipeline that will get us through the end of this decade but then it gets questionable."