While systems vendors enthusiastically debated the merits of various modulation schemes in a number of presentations, carriers say that they don't need 40G now, and that it doesn't loom large in their immediate plans.

It's important to distinguish what's just interesting from what's really fundamental to our industry, says David Selby, senior director, product delivery, European markets for Level 3.

"40-gig is getting a lot of attention, but it is in demand?" he said. "As a reasonable supplier of wavelength services, we don't see a huge commercial market for 40G lambdas."

"We're the same," said Tim Hubbard, head of 21CN technology and platform introduction for BT. "Today we do not have a requirement for 40G."

Hubbard doesn't see a requirement for 40G in BT's access network at all. Strictly speaking, he says, there's no demand on the core either. There are one or two "hot spots" in the network, where BT might consider deploying 40G, but the cost will have to come down.

Cost seems to be at the heart of the problem. With 40 Gbit/s wavelengths still costing more than four 10 Gbit/s wavelengths, carriers say the economic equation often only makes sense in metro or long-haul networks where the fibre is close to exhaustion, and adding more capacity would require opening up a trench.

40 Gbit/s router interfaces, often quoted as the other key driver for 40G deployment, also appear to be too expensive for widespread acceptance.

"I know how many OC768/STM256 interfaces we've sold on the CRS-1 and its very small [compared to the 10G market]," said Russ Esmacher, senior product line manager, optical technology business unit, Cisco. "When the standards get out there and get ratified, then we see potential volume."

Components vendors blame the high cost of 40G components on the lack of a coordinated approach to building the subsystems.

"What's really the challenge at 40G is not so much the cost or modulation scheme, it's getting the whole supply chain working," said Sinclair Vass, director EMEA for JDSU. "40G is a cobbled together technology — let's standardize on an approach so we can make something with a yield we can work from."

Vass also voiced his hope that the industry should not repeat these mistakes with 100G.