The SiGe (pronounced 'siggy') marketplace has exploded since 1997,
with the need for high-speed devices for communications. It's not
because SiGe is a latecomer -- Jack Kilby's first semiconductor
designs for an oscillator (U.S. Patent 3,138,743) at Texas Instruments
in August, 1958 used germanium wafers to build what was termed the
first "solid circuit."
SiGe work has progressed alongside the more-dominant silicon substrate
work, though silicon wafers have enjoyed a great advantage over the
past 44 years. The first silicon substrate circuit was designed by
Robert Noyce at Fairchild Camera and Instrument (U.S. Patent
#2,981,877) shortly after Kirby's work. Noyce, whose patent was
actually filed almost 6 months after Kilby's, was later a co-founder
of Intel.
Beyond power consumption and speed issues for communications devices,
recent improvements in deposition techniques have made SiGe more
practical and competitive with standard silicon substrates.
THE RESEARCH ISSUES
------------------------------------
As noted in my comments, there is at least one comprehensive study
done of the market by Semico Research titled "Silicon Germanium
Blooms: The Quick and Quiet Solution." Continuing to track this
market will involve following:
* more than 15 companies (there are more than 200 semiconductor
manufacturers)
* at least 11 markets or applications
* 10 or more research companies' reports
I've organized this analysis in that order and will include links to
the richest information areas.
MARKET PARTICIPANTS
-------------------
I've identified 15 companies with products in the market, though other
articles note that there are twice as many companies in development
with SiGe products. Additional potential participants mentioned are
Hitachi, Toshiba, NEC, STMicrolectronics, Alcatel, Daimler-Benz, TI,
National Semiconductor and Nortel. But here's who has products in the
market today:
IBM's Microelectronics Division, in Burlington, VT, has been
identified as the leader in the early SiGe market. Trevor Yancey, an
analyst with IC Insights, has been quoted as saying that IBM's share
of the Internet backbone (SONET) market may be as high as 80%. IBM's
announced SiGe technology capable of running as high as 210GHz,
putting them well ahead in technology. An IBM announcement in July,
2002 said that the division had shipped 100 million SiGe circuits.
Applied Microcircuits Corp. (AMCC), San Diego, has partnered with IBM
to provide high speed optical circuits for OC-192.
Other participants with products announced or under development
include:
Agere (former Lucent Technologies Microelectronics Group, Allentown,
PA)
Atmel (San Jose, CA)
Chartered Semiconductor (Singapore)
Commuicant Semiconductor (Frankfort)
Cypress Semiconductor (San Jose, CA)
Infineon's Communications Business Group (Munich)
Intel Communications Group (Santa Clara, CA)
Intersil (Irvine, CA)
Jazz Semiconductor (Long Beach, CA)
Philips Semiconductor (Einhoven, The Netherlands)
Sige Semiconductor (Ottawa, Canada)
Temic Telefunken (Heilbronn, Germany)
TSMC (Hsinchu, Taiwan)
SiGe APPLICATIONS
------------------
The ideal speeds for applying SiGe are in applications requiring
speeds of more than 2GHz, though in the 2-4GHz range there are
competing technologies such as silicon-on-insulator (SOI); RF-CMOS;
BiCMOS and gallenium arsenide (GaAs) that compete for market share.
1. High-speed fiber optics (SONET). The fiber optic backbone for
communications has set standards to carry the network beyond the
current 155Mbps - 2.5Gbps range of the backbone. Strategies
Unlimited, a research firm in Mountain View, CA, is conservative on
its estimates of the total SiGe potential but believes that this
portion of the market will be 16% of the total.
OC-48 (4Gbps): implementations starting in 2002 and beyond.
OC-192 (10Gbps): this is one specific target market in the AMCC and
IBM joint development work. Infineon has also announced products for
this market.
OC-768 (40Gbps): implementation starting in 2003. Agere, Infineon and
AMCC/IBM are expected to compete in a market for 30,000 to 50,000 SiGe
devices worth $120 million to $200 million:
Semiconductor Business News (June, 2001), "IC Makers Race to Offer
40-Gbit OC-768 Chips, But Disagree on Process Technologies
http://www.siliconstrategies.com/story/OEG20010604S0009
2. Mobile phone RF transceivers. The base of users is huge, with 1
billion subscribers worldwide. But penetration levels are near 70% in
some markets -- and uncertainty over location-based services using
GPS and other technologies -- makes predictions difficult.
Allied Business Intelligence just released a report "RF Power Devices:
Transistors, ICs, Power Modules and New Materials" (October, 2002)
that predicts power modules will be dominated by GaAs, with average
selling prices going from $2.52 today to less than $1:
http://www.alliedworld.com/pdfs/psem02pr.pdf
The market for SiGE, Allied Business Intelligence contends, is in GPS
for phone handsets. There the market for ICs will rise from $150
million in 2002 to almost $3 billion in 2006. The problem with the
forecasts is regulatory issues -- though the FCC requirements were
that starting Oct. 1, 2002 25% of cell phones had to pinpoint a users
location, all of the Big Three manufacturers (Nokia, Ericsson and
Motorola) petitioned for a delay of E-911 services. An excellent
background piece onE-911, GPS and SiGe is in Semiconductor Magazine,
"One Thing GPS is struggling to Find: a Market" (May, 2001):
http://www.semi.org/web/wmagazine.nsf/4f55b97743c2d02e882565bf006c2459/290d9272dd781c3288256a37006a27e0!OpenDocument
3. Wireless LAN (802.11), 2.4GHz. This market has been actively
discussed for at least 10 years. With Intel and Microsoft looking for
better wireless technologies, Bluetooth technology revived interest in
1998. Deliveries have been late, but Xilinx goes out on a limb for a
forecast of units -- and OEM semiconductor pricing:
2002: $17, 121 million units
2003: $8
2004: $7
2005: $6, 1.1 billion units
The Xilinx forecast in in it's "Bluetooth Introduction" (undated)
contains excellent breakdowns by cellular/mobile/desktop/access point
markets and is available here:
http://www.xilinx.com/esp/bluetooth/tutorials/intro.htm
IDC notes the notorious delays in the Bluetooth, but predicts growth
for semiconductors from $76.6 million in 2001 to $2.6 billion by 2006.
As with the Xilinx report, cellular handsets are the largest market
-- though IDC predicts headsets for cellphone will be 2nd (followed by
desktop PCs, notebook PCs and accessories). The IDC report is
"Wordwide Bluetooth Semiconductor Market Forecast and Analysis,
2001-2006" (June 2002):
http://www.idc.com/getdoc.jhtml?containerId=27462
On the WiFi side, Aberdeen Group says that long-term growth will be
good but that the "market faces imminent consolidation" as more than
25 vendors compete for business that actually shrank in 2001. The
market for chips was down last year to $240 million, a decline of 5%,
according to "Only Focused WiFi-x IC suppliers will Prosper" (April,
2002):
http://www.aberdeen.com/2001/research/04020010.asp
Growth for these wireless LANs will be 30% per year for the next 4
years, according to Aberdeen, with a $890 million semiconductor market
in 2006. In a separate report, done at the same time, Aberdeen also
warns that alternate technologies may prevent WiFi from ever really
taking hold:
http://www.aberdeen.com/2001/research/04020012.asp
4. Cable/DSL markets. Sige Semiconductor makes projections that the
digital IC tuner market for ICs would rise at 175% per year for the
first half of this decade, to $211 million in 2005. They cite an
Isupply study, though I haven't been able to find the research
service. This market for digital tuners includes high-definition TV
sets; cable modems and tuners. The number of tuners is expected to
grow by 40% per year -- with ICs growing at a higher rate due to the
addition of functions.
Cable and DSL unit predictions are covered in a report on IDC studies
in Semiconductor Business News titled "Cable modem subscribers to
reach 57.5 million by 2005" (June, 2001):
http://www.siliconstrategies.com/story/OEG20010625S0097
5. Pre-amplifiers for drives. Uses include DVD drives, for which IBM
has developed amplifiers, as well as hard drives. Little market data
is apparent.
MARKET RESEARCH COMPANIES
-------------------------
Overall forecasts for 2005-2006 put SiGe sales in the $1.6 billion to
$2.7 billion range, depending on analysts assumptions. Here are some
of the most-visible predictions:
Semico Research: Forecasts of 49% growth per year, primarily in
cellular, optical networking and disk drives. "SiGe will only replace
a portion of the RF components for each of these individual markets,"
says Joanne Itow, of Semico, but unit growth will be 55% per year
through 2006 to $2.7 billion. Itow's comments are from Semiconductor
Business News, "Foundries to Bring SiGe into Mainstream," (Sept. 10,
2002):
http://www.siliconstrategies.com/story/OEG20020910S0023
Again, Semico's complete study "Silicon Germanium Blooms" (August,
2002) is summarized on the company's web page:
http://www.semico.com/studies/moreinfo.asp?id=158
Strategies Unlimited: Like Semico, they've done detailed splits of the
marketplace in multi-client research, even profiling the high-speed
SONET market for ICs. Overall they see the market going from $450
million in 2002 to $1.8 billion in 2005. They assume 679% to be
wireless; 16% for high-speed networking like SONET applications.
An excellent article from The Industrial Physicist, "Silicon-Germanium
Gives Semiconductors the Edge" (June/July 2002) highlights Strategies
Unlimited forecasts:
http://www.aip.org/tip/INPHFA/vol-8/iss-3/p22.pdf
Aberdeen Group: has done a number of interesting analyses,
particularly of sub-markets like WiFi.
Other research groups which have done SiGe analyses:
IC Knowledge
Insight 64
Allied Business international
Intex Management Services
VLSI Research
Dataquest Semiconductor Service
And two other sources to monitor are, though SiGe seems to be below
the radar for SIA:
Semiconductor Industry Association (http://www.semichips.org/home.cfm)
SEMI (www.semi.org)
GOOGLE SEARCH STRATEGY
----------------------
Usually at the early stages of a market, information is ample as
research companies seek to attract new clients. Perhaps because of
the explosive emergence of SiGe from the shadows of an industry
dominated by MOS technology it's not the case here. It also may be
due to the continual delay (read: inaccurate forecasts) for
technologies like 802.11 and Bluetooth.
At any rate, I've worked on this question quite a bit and wouldn't
claim to have exhausted all sources. In particular, there are
journals which may not be accessible on the Internet -- but a good
library will have access to them electronically.
A highly profitable search strategy is to go to a SiGe vendor site,
then use Google's Advanced search to look at the site itself for
references to SiGe. You can also use a 'command line' entry in Google
(though many researchers have a Google toolbar in their browser to do
same):
site:www.semi.org SiGe
One often finds presentations, articles, analysts reports on a company
website that aren't very findable from the home page.
A quick overview of leading electronics publications, gives an idea of
who's visible in the market, so using Electronic News, Silicon
Strategies (CMP's site for multiple publications), EE Times, and Chip
Scale Review is a great way to start.
The term "SiGe" is so unique that it works to quickly isolate
information that you're seeking. However, if you're searching the
World Wide Web you want to use the following to avoid hits with
involving a Japanese company which isn't in this business:
SiGe + semiconductors
OTHER ISSUES
------------
Any good research will raise as many questions as it answers, but
there are lots of interesting ones that arise from this overview:
* is GaAs going to dominate in 2-3 GHz designs, as Allied Business
Intelligence says it will for cellphone handset? What's the
implication for Bluetooth designs?
* how close is Bluetooth to actually shipping? what issues are there
in customer acceptance of Bluetooth?
* who's getting the design wins for OC-48 products?
* how can data be developed for DVD and hard drives? There's good
information about the size and characteristics of the drive markets --
now how can it be profiled for potential SiGe applications?
Best regards,
Omnivorous-GA |
,
3 Comments
)