Kate Hafner provided New York Times readers with an interesting profile of tarnished pundit George Gilder in a recent Sunday Finance section. She spent a harsh moment on Gilder's all-but-religious belief in the power of broadband, but tempered her critique by suggesting the optical world, at least, might be turning back Gilder's way.
If it is, the evidence is tenuous. Publications focusing on photonic components have shrunk for three years straight, to the point where they are no larger than brochures. Few talk about optical mesh or reconfigurable add-drop for serious traffic. And network speeds may plateau at 10 Gbits/s for a while.
This last point was the subject of a CMP Netseminar that discussed the finer points of emerging 40-Gbit network components. It would be nice to dwell on the arcane issues of CMOS vs. III-V materials, or how much dispersion compensation is optimized for 40- vs. 10-Gbit networks. But a more basic and profound question is: Are there any drivers for bringing network speeds beyond 10 Gbits/s, in the public network or enterprise LANs and clusters?
The public network experienced a similar halt in growth at 622 Mbits/s, when the arrival of low-cost dense wave-division multiplexing kept carriers from having to buy the next grade of fiber to handle single-wavelength Sonet speeds. But this time, the fiber overbuild is more fundamental. If there are too many 2.5- and 10-Gbit networks out there, no one's going to put in new cable plant at any speed. And if 10-Gbit networks become the new gold standard, commodity component pricing will further delay experimentation with 40-Gbit backbones.
Given the precarious state of carriers, a few brave system companies are retargeting their interest toward specialized government networks, and advanced LANs and storage-area networks in the enterprise. Switch and add-drop startups like Calient and Photuris are making such networks their main market.
That might be OK for a nimble systems company that can play at 10 and 40 Gbits. It also works for a semiconductor giant like Infineon Technologies, our Netseminar sponsor, which offers chips in both markets. But what of the small IC startup focused solely on 40 Gbits/s? Gilder will not be able to show them any "angels in the photons" soon.
Loring Wirbel is Communications editorial director for EE Times and its network publications.