DSL Prime: Why VDSL Growth is Slowing

Although there are customers for VDSL, and although VDSL does everything ADSL2+ does, VDSL growth is slowing.

by Dave Burstein of DSL Prime and Future of TV [November 6, 2006]

Vashist Out at Ikanos as VDSL slows
TI predicts slow ADSL to VDSL shift
Rajesh Vashist built Ikanos into a dominating leader in VDSL, but now is leaving as sales fell Q3 and look to plummet in Q4. His achievements should not be underestimated. Ikanos delivered DMT fast VDSL chips at least a year before anyone else. Without Ikanos, QAM chips from Metalink and Infineon/Savan would probably be winning the market today because DMT allies Broadcom, ST, and TI were unable to demonstrate production chips. He raised sales to $40 million a quarter, successfully IPO'd the firm, and leaves with strong products on the way. Vashist was a demanding CEO, with high turnover in key positions, but until this quarter the results were remarkable.

G. Venkatesh led the board in this surprising move. Venkatash built Maverick Networks and sold it to Broadcom. He is joined by Mike Gulett, who was COO of DSL chipmaker Virata, Danial Faizullabhoy of Walden Capital, and Michael Goguen of Sequoia. They are all senior technologists, giving Ikanos one of the strongest (and most independent) boards in the industry.

The company has $100 million in cash and plenty of time to turn things around. Interim CEO Dan Atler's highest priority is solving the production issues on the new 90 nanometer chips and improving the firmware and system design to get maximum performance.

Rajesh was the strongest proponent of quickly moving from ADSL to VDSL, but few carriers have made that choice. VDSL2 includes everything in ADSL2+, so at some point will dominate the market. It offers years of small improvements in technology and 3 to 50 times the performance on short loops. I've previously reported one CTO's disappointments with power, and density. In addition, while the DSL Forum is moving ahead with interoperability testing, progress is severely limited. "Wise old hands" are reluctant to deploy units without proof they will be compatible in the future. Pre-DOCSIS 3.0 100 Mbps cable modems aren't hitting the market outside Asia, and 3.0 itself looks more like 2008, giving telcos breathing room.

Ben Sheppard of TI expects the gap to continue. "The period of time when VDSL2 shipments cross over ADSL still could be ten years away," he tells Loring Wirbel in EE Times. TI estimates ADSL sales don't plateau until 2009. Sheppard adds in an e-mail, "In short, my statement on the potential for a 10 year time horizon for VDSL2 over ADSL is driven primarily by the changes that are needed in the infrastructure to enable VDSL2 pervasiveness. Today there is a 95 million gap between ADSL CPE shipments and VDSL shipments. It is just going to take time for the VDSL2 market to reach its potential. "

Cyrus Namazi of Conexant is supplying VDSL to AT&T and Chunghwa, but agrees ADSL will stay strong. "We do not see VDSL as a 'replacement' technology for ADSL. The market drivers for VDSL are different to that of ADSL. Where high speed data access was the driver for ADSL demand, it's the lure of real time applications such as video that are driving the demand for VDSL. In most countries in the world, with the exception of Japan, it's this addition of selling video services to their existing voice and high speed data business that drives service providers to deploy VDSL. Given the cost and complexity trade offs between ADSL and VDSL, we believe that VDSL deployments will continue to be complementing ADSL, as opposed to replacing it, for some time." The VDSL sales gap is unmistakable after two weak quarters. So my comment similar to "ADSL is so last century. Today, it's fiber home or fiber + VDSL at 50 Mbps or more," simply isn't supported in most countries in 2006.

Moore's Law will improve the density and price, which currently is typically twice as high for VDSL. Texas Instruments, Broadcom, and at least two others had originally planned to have full rate VDSL chips on the market by the end of 2006. TI has just released UR8, a gateway on a chip that emphasizes ADSL and low end VDSL; their 100 meg symmetric chip is delayed. I expected the new entrants, especially TI, to price aggressively to win market share. Fab capacity is widely available, generally bringing down pricing; everyone from Chartered in Singapore to TSMC in Taiwan is reporting mixed sales prospects. Early reports from 65 nm and even 45 nm tests reported by EE Times are highly encouraging, suggesting costs will continue dropping. Power is a more difficult issue, especially for remote DSLAMs. Even in basement installations, the need for external power can add several dollars per month to the cost.

VDSL sales will grow significantly in the first half of 2007, when Germany, AT&T, Swisscom, Belgacom, KPN, and TDC Denmark plan to rapidly expand their TV offerings. I don't have evidence to judge whether crossover is coming soon after or delayed as TI expects. The key question is whether carriers currently doing ADSL2+ switch to VDSL for operating savings, speed, and future protection, assuming the price comes down. My belief in customer demand for higher speeds would suggest most ADSL deployments would do better as VDSL, because a significant minority of the homes will get much higher speeds. Power and space are much less significant in modems, so if VDSL chip prices come down carriers looking to future proof may shift quickly. Most North American carriers are unwilling to offer 50 Mbps service to customers close by, fearing complaints from those who can't get that speed.

Copyright 2006 Dave Burstein.
The DSL Prime Newsletter is reprinted with permission.

"The power of the printing press belongs solely to those who own the presses"
—A.J. Leibling

The Internet is the cheapest printing press ever invented.