The Carlsbad, California-based company announced on Monday that it has created an optical modulator that can transfer data at 10Gbps with conventional silicon processes. The company hopes to have samples out by the second quarter of 2006 and move into commercial production by late 2006 or early 2007, said CEO Alex Dickinson.
The achievement pits the company in a race against Intel, Primarion and many others to change the way computers, and ultimately chips within a computer, speak to each other. Some companies are trying to achieve 10Gbps speeds on copper wire, according to, among others, Crosslink Capital's Dave Epstein.
Others, such as Luxtera, believe the future belongs to swapping out wire for optical fibre. The company also says it is moving faster than Intel, which has said it may not bring out a silicon modulator until toward the end of the decade.
Unfortunately, producing optical equipment is expensive: Manufacturers often refer to their trade as a black art. Dickinson and other chip executives, however, believe that a lot of the costs come from the fact that optical emerged from the telecommunications industry.
"They use exotic materials and exotic packaging. There are a lot of different pieces," he said.
Still, manufacturing optical components on silicon has proved problematic. One issue, for instance, is that the wavelengths of light are often longer than the channels inside a silicon chip.
Luxtera's secret weapon is the California Institute of Technology. The company's technology comes out of the university, which has also been running computerised simulations for years on how to solve the problem. The software and know-how from the university can't be easily replicated, said Dickinson.
"Light in small waveguides behaves in very strange ways," Dickinson said.
The prototype was made on behalf of Luxtera by FreeScale, on its 130nm silicon processes. Over the next year, Luxtera will test the chip and work on issues associated with mass manufacturing.
Like Dickinson, many of the company's executives and advisers are familiar faces. Gabriele Sartori, a former AMD executive who helped popularise HyperTransport, serves as vice-president of marketing. Forest Baskett, a partner at New Enterprise Associates and a Stanford professor, sits on the board, while Arno Penzias, who won a Nobel Prize for his work on the Big Bang and is an NEA partner, sits on the advisory board. The company's CTO, Cary Gunn, was named one of the Top 100 innovators by MIT Technology Review in 2003.
The company has received US$24m in venture backing from, among others, August Capital and New Enterprise Associates.
Holy Cow ! Luxtera has a light-switch !
So, they worked for three years trying to turn their university reseach into a meaningful product and what have they accomplished ?
A silicon device that can ON/OFF-modulate light of an unknown wavelength at 10Gbit/s. WOW.
Is that all they've got ?
Looks like the heydays of the internet are back. You only have to call yourself nano-something these days and sell your old cookies to everybody.
Has Luxtera addressed any of the real challenges in Silicon Optics - like integrated Laser sources, low-loss waveguides, taming background noise, manufacturable packaging, temperature stability, ageing etc etc.
BTW:
In their press Blitz they keep talking about a "product". Does anybody know what their "product" really is ? Anything that would justify this term ? Anything beyond a hand-picked lab-sample light switch soldered on a high-speed board ?