IBM Is Investing $3 Billion to Advance Computing Technology
Hold onto your hats, because computing as we know it is soon to change. The once accurate Moore’s Law is slowing, with microchip scaling slowing from the oft-cited doubling of the transistor density every two years to the point where traditional silicon scaling simply won’t be viable in the near future. Although this may be perceived as a setback for some, IBM isn’t having any of that. Instead, it’s seeing it as an opportunity to dive headlong into the future.
IBM expects semiconductors to scale from today’s 22 nanometers down to 7 nanometers For comparison, a strand of human DNA is 2.5 nanometers in diameter
That’s in part why it recently announced a $3 billion investment into new non-silicon technologies such as neurosynaptics, carbon nanotubes and even quantum computing. Although this sounds like “Star Trek: Generation Whatever” advances in computing technologies, Supratik Guha, director of physical science with IBM Research, says some of these technologies are closer to reality than we might think—and will actually be for the better.
Q. What are some of the computational walls we’re running into today, including, perhaps most significantly, silicon-based scaling?
A. Silicon technology has worked wonderfully for us. We keep scaling, shrinking chip size, supporting more devices in the same area, and we’re reducing the power usage and getting more and more performance. This has been going on for a while, and today we’re at the 22-nanometer node. But it’s our belief that silicon scaling will come to an end in a few more generations—in maybe eight years—because the crossover point may come at the 7-nanometer nodes. So the question is: Is there something else, a different material that could replace silicon at that point or do we just say this far and no further?
Scientists the world over, including at IBM, are looking for alternatives to silicon. For our part, we’ve been looking at a couple of alternatives. One is called carbon electronics, specifically carbon nanotubes. The second one is looking at III-V semiconducting compounds. These are materials based on the family of gallium-arsenide compounds, such as those used for making semiconductor lasers—the kind used in CDs and telecommunications.
Carbon nanotubes have been extremely promising. As individual devices, they work very well. When we make very small transistors of carbon nanotubes, which is to say the size is about 10 nanometers or so, they outperform other materials. The challenge now is in taking this and developing it into a viable technology, which means you have to be able to place billions of these transistors on a wafer—all of them perfect, all of them placed in a systematic manner very close to one another. This is what one of our teams is trying to do. Now please keep in mind that replacing silicon is no simple task, so this is a research endeavor at this point and there are challenges that need to be surmounted. If successful, however, the payoff can be enormous.
Quite a few other things are going on. As you start looking at an entire computing system, in addition to the logic processors, there’s memory and there’s the communication links between all of these.
Like what you just read? To receive technical tips and articles directly in your inbox twice per month, sign up for the EXTRA e-newsletter here.
comments powered by