Personally, I'm not quite sure why any sane person would want to do this, but Intel researchers claim that integrated silicon photonics devices will shortly make possible the electronic transfer of the entire Library of Congress in less than two minutes. So says one Justin Rattner, Intel's chief technology officer, speaking at this year's Integrated Photonics Research Conference in Monterey, CA. Well, bully for you, Justin. BTW, I have another terabit task for you.
The Intel researchers have developed a 50 Gbit/s integrated laser link based on conventional silicon technology. While this is no minor feat, the Intel prototype is actually only four optical channels, each running at 12.5 Gbit/s and combined onto a single fiber to transmit at data rates up to 50 Gbit/s. The catch is that you have to "scale up" to achieve transmission speeds of 40 Gbit/s and "scale out" to as many as 25 channels; then terabit transmissions are indeed possible... but not yet achievable. Much effort will be needed to get to the terabit level.
To give you some idea of the complexity involved, I quote from a recent Intel white paper: "Four hybrid silicon lasers are created by fusing a layer of indium phosphide onto the silicon waveguides during manufacturing. The link uses integrated silicon modulators with much more performance headroom. Using constructive and deconstructive interference, the high speed modulators act as downstream 'shutters' that place data onto light from the hybrid lasers. A downstream, passive multiplexer combines the four light-streams onto a single fiber-optic cable or waveguide."
So all that research and development is going to require time and money. In a recent interview in The New York Times, Intel chief executive officer Paul Otellini noted that Intel will be seeking major partnerships to invest in the manufacturing of complex integrated silicon photonics chips. Of course, this is all part and parcel of Intel's non-PC strategy.
In my view, this is a survival strategy for Intel. The company has done well manufacturing designs for others–especially Apple and the Windows-based PC makers. But downward price pressure for microprocessors is quickly turning them into commodity chips–not quite memory chips, but close. Intel expects to be a player in the smart phone market, but it is already behind in this lucrative sector. Still, Intel is a big company ($35 billion in revenue, $4.4 million profits in 2009, and employs 79,800 worldwide).
Anyway, I'm a big fan of Intel. As I write this column, I have been avidly watching the 2010 US Open Tennis Championships. In a younger incarnation as the managing editor of Tennis magazine, I used to cover (on sweaty feet!) the tournament on the insanely overcrowded and overheated courts of the USTA's Billie Jean King Tennis Center in Flushing Meadow and before that at the impossibly crowded and badly maintained grass courts of the Forest Hills Tennis Club in Queens.
Now, although I am hundreds of miles away from New York City, I get my tennis fix, day and night, courtesy of my iPad tablet and the five simultaneous streams of play provided by IBM and a whole passel of Intel microprocessors. How sweet it is to sit on our screened porch with a frosty or two to hand and homemade munchies to suit. Life does not get much better than this! Intel Inside? You better believe it.