On Saturday evening, at midnight no less, Intel (Nasdaq: INTC ) and IBM (NYSE: IBM ) separately announced that they would be using a new type of material to manufacture different components on next-generation 45-nanometer computer chips.
This is a significant development, and how these materials are incorporated into their chips over the next couple of years could have huge implications for the relative success of not just Intel and IBM, but a number of chip manufacturers as well.
Houston, we have a small problem
Here's why. Chip manufacturers are racing to double the number of transistors they pack onto their chips roughly every 18 months. This feat, known as Moore's Law, is responsible for bringing us a bevy of high-performance, low-cost computerized devices. But the industry has begun to experience some problems, as it has continuously had to shrink various chip components. The industry is now constructing transistors just 45 nanometers in size. To understand how small this is, just consider that 2,000 such components could fit in the period at the end of this sentence.
Most serious among the problems these tiny components cause are excessive heat and increased power consumption. As electrons are buzzing around in ever-closer proximity to one another as a result of smaller transistors, they are causing the chips to overheat. Also, as the parts shrink, many electrons are not behaving properly and are leaking away. (This can cause the unfortunate problem of a transistor being on when it should be off or vice versa.)
The new materials seek to address both of these problems. The culprits are not the transistors themselves. Rather, the problem occurs with the gate -- the component that switches a transistor on and off -- and something called the gate dielectric, which helps control the flow of electrons.
For years, the gate dielectric has been manufactured with silicon dioxide, a material with good insulating properties. Unfortunately, as the transistors have become smaller, the gate dielectrics have also had to become smaller, and they are now getting so thin that they are rapidly losing their ability to control electrons.
Chips ahoy! A new material to the rescue
Both Intel and IBM are reporting that they will be using a new material, called high-K, based on the element hafnium. Because high-K can hold more electrical charge, it is expected that it will help reduce leakage considerably.
The two companies begin to diverge at the next level -- the gate itself. Traditionally, the gate has been manufactured with polysilicon. In order to make the gate compatible with the new, high-K gate dielectric, the manufacturers have had to devise an entirely new material. Neither company was willing to disclose much about how they make their new material. Suffice it to say, though, that this "secret sauce" is quite important to the success of the chip.
Why does all this matter? Because, to the extent that these new materials are successful in cutting down on power consumption and helping next-generation chips perform faster, they should also help win over big clients such as Dell and Hewlett-Packard, which are constantly looking for better-performing and -functioning chips.
In the past week, I've written about how Intel has recently won over both Sun Microsystems (Nasdaq: SUNW ) and Google (Nasdaq: GOOG ) with improvements to its chips. The creation of these new materials will simply take the battle to the next-generation chips.
So who will win?
Currently, Intel seems to have the advantage, because it is planning on using its new materials this year in its Penryn family of chips. A company spokesperson says the new materials will be in five early versions of the chip later this year, and they will serve a variety of server, desktop, and laptop systems.
How long this advantage will last is a matter of some debate. This is because IBM, which is partnering with Advanced Micro Devices (NYSE: AMD ) , Sony (NYSE: SNE ) , and Toshiba in the development of its new materials, claims that its breakthrough is more significant. Unlike Intel, IBM has figured out how to embed its new metal gates directly into silicon. (Intel's gates must still sit atop a silicon architecture structure.) And this is important because as transistors shrink to the 32-nanometer and then the 22-nanometer level, IBM and the others can expect to use this additional space to increase performance capacity.
From my perspective, Intel will be the early winner and will use this new development to build on its recent momentum and increase its market share throughout 2007. By early 2008, however, IBM, AMD, and maybe even Texas Instruments (NYSE: TXN ) (which has indicated that it will be out with new chips based on similar material in early 2008), should be back in the game. They might even be able to surpass Intel by constructing an even better chip that embeds the gates directly into silicon.
Of course, the real winners will be you and I, the consumers. We will continue to benefit from better, faster, and cheaper computers that allow us to play more vivid video games, easily edit video clips, and perform a number of other processor-intensive computer tasks.
Interested in other chip-related foolishness? Check out these recent articles:
Fool contributor Jack Uldrich was asleep at midnight on Saturday and didn't learn of any of these developments until the following morning. He owns stock in Intel and IBM. The Fool has a strict disclosure policy.