DSL Prime: China Grows

Although actual statistics may be over-optimistic, China's DSL deployments already offer greater speeds than are availale in the U.S., and total subscribers should surpass the U.S. in 2005.

by Dave Burstein DSL Prime [December 4, 2003]

China: 10 million going towards 75 million
World #1 in DSL shortly, in broadband probably two years
Three different informed sources convince me China will pass 10 million DSL subscribers shortly, if it has not already done so, vaulting past Japan for world leadership. That amazing jump this year is confirmed by Chang Xiaobing of China Telecom, Ying Wu of UTStarcom, and William Bao Bean quoting the semi-official figures from CCID. (Xiaobing indicates his pr department will help distribute more accurate data.)

The exchange-listed part of CT alone did almost a million last quarter, going to 3.517 million by Sept. 30, from 2.562. Ying Wu spoke in New York of a China Netcom plan for 25 million subscribers, and suggests estimating CT at twice Netcom's figures. That implies 75 million DSL subscribers in a few years, more than twice as many as any other country.

The industry knows this, and companies like Alcatel, Siemens, and Nokia are shifting their emphasis to the Chinese market. Chinese giants Huawei and ZTE now bid for contracts around the world, including Britain and France. They haven't won many DSL bids, but my e-mail today brings news Huawei is deploying 10 Gbps SDH in Rio de Janeiro.

Caution, please, before assuming winners and losers from this growth. Add together market share projections of the main companies, and they equal 200 percent of the market. Earlier optimism about Chinese growth led the entire industry to overproduce, resulting in several dismal quarters for chip vendors and inventory writeoffs.

"We're doing it for China," artist Yang Fudong tells the NYT. "The purpose of going abroad is very simple. You take a look at what the world is doing and then you come back and you do it yourself." It's now time for the rest of the world to look at China in turn, including their rapid conversion to IP DSLAMs.

Broadband: Who has it?
Michael Minges of the ITU has interesting yearend 2002 comparisons.
Broadband lines per hundred population

1 Korea (Rep.) 21.9

2 Hong Kong 14.6

3 Canada 11.1

4 Taiwan 9.4

5 Belgium 8.4

Minges adds other factors, like Internet bandwidth per capita, education, telephone penetration both wired and wireless, creating a new "Digital Access Index." Darryl Nelson at DSL Reports notes "Not one of the 5 strongest economies in the World—the US, UK, Japan, Germany and France—made it into the top 10. Apart from Canada, ranked 10th, and Holland (6th), the leading ICT nations are exclusively Asian and Scandinavian."

The remarkable growth in Japan, as well as strength in Western Europe, may be changing the rankings soon, but the U.S. is unlikely to rise.

A year ago, E. and T., two well known economists, chided me for complaining about U.S. performance. Cable and DSL began earlier in the U.S., but by 2001 U.S. quarterly numbers were disappointing and the trend was clear. The U.S. still has an absolute lead in broadband subscribers, and probably will not be passed by China until 2005.

Broadband subscribers by region: 3Q03
More than half the world's DSL is now in Asia. More than half of cable is in North America.

(From Teresa Mastrangelo of RHK. Early numbers, but pretty close.)

DSL: Economical beyond rich countries
Chinese capex $160/customer—last year
E., an academic, began by flattering me "Your newsletter is a gust of information and very useful." In return, he ask for some data: "Roughly speaking, how much does the upgrade of a creaky 3rd world phone system to DSL cost per connected user?"

$160 per subscriber in 2002, China Telecom claims in their 2002 Annual Report. Today, a similar cost is right accurate for many countries. Equipment prices have since come down, but bids should be somewhat higher if you order less than CT's 1.6 million lines.

One recent Middle Eastern bid for DSLAMs in quantity 100,000 was around $60, which means with modems, subscriber management, routers, and similar the hardware cost per user was about $125.

In general, people able to pay for DSL service can be asked to pay $50-100 upfront for the modem/hook-up, so the initial charge can defray some of the telco investment. The harder problem is backhaul, unless you are the national telco with fiber connections already in place.

I suspect he was surprised at my answer to what he called a provocative question: "Why is BB important to a poor country?" I support DSL everywhere, but realistically "Broadband is not that important. 'No one can worship God or love his neighbour on an empty stomach,' much less use the Internet."

But while I'd spend public money on health first, that doesn't mean there's no room for DSL. "Turns out, offering DSL is not that expensive. Every exchange in Cairo can be done for $2 million, and would almost surely be self-supporting. You could do 20 key COs in, say, Eritrea, for what one expatriate aid program officer costs for a year."

The trap in many places, especially if backhaul is expensive, is the price of DSL has been set too high. The effect is the market stays small, and the resulting low volume keeps prices high. The Chinese have proven that market grows rapidly if the price is kept down.

ADSL2+ working for Beijing Harbour
Who wants obsolete DSLAMs?
Harbour, China's third largest domestic DSLAM producer, is adopting new technology rapidly in the hope of increasing orders from CT and CN, to whom they have already shipped nearly a million lines. Their Hammer10000 and Hammer3300-48 successfully passed the "ADSL2+New Technology" test in Hangzhou Telecom, Xinjiang Telecom and Jiangsu Telecom. Harbour offers either ADSL or VDSL cards, supported by a high performance IP-based switching fabric, ADSL multicast, gigabit uplink ports, and cluster management.

ADSL2+ is not working well in production yet, so I'm following all results from the field closely. I believe nearly all providers should switch to ADSL2+ gear as soon as it's proven reliable, because many customers with currently marginal performance will get twice the throughput from the new chipsets. At $300/customer for acquisition, even a small reduction in churn and customer turnoffs is huge to the bottom line. ADSL2 has much greater reach and performance.

The changes in the standard make a difference, while nearly all the chipmakers have made small enhancements in the chip design that carriers tell me result in significant improvements.

Japanese carriers, in a market begging for speed, are already in production, and often reach speeds of 10 and 15 Mbps. They have to carefully match modem and DSLAM, for interoperability still needs improvements.

Alcatel in the U.S. is holding back until at least the first quarter, which is probably prudent for most carriers.

Copyright 2003 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.