There is little question at this point that the telecom infrastructure market—the
major engine behind the meteoric rise in the embedded-computer business in 1999
and 2000—will continue to be depressed throughout most of 2003. And, it
will probably be well into 2004 before any real hard increases will be realized.
That’s the bad news. The good news is that analysts believe the industry
has seen the bottom of the trough and, while it may bounce along the bottom
for some time yet, there is little indication it will retract more.
Despite taking a mighty blow, the telecommunications and networking infrastructure
business still remains a big spender. More than $50 billion will be spent in
2003 on new equipment and materials. Further, suppliers to the industry such
as Lucent, Nortel and Ericsson continue to be viable and currently report revenue
of some $30 billion among the three of them. Granted this is one-fifth of what
it was when they were in their greatest glory, but it’s still sizable revenue.
Others fared similarly—some better, some worse.
The mid- and high-end embedded-computer industry (as opposed to deeply embedded)
tracked the trend of the communications infrastructure. And while its peaks
and valleys may not have been as severe because of the mix of industries it
addresses, it was negatively impacted by the general economic malaise this country
has been experiencing since the beginning of 2001 and one that was exacerbated
by the events of 9-11.
Figure 1 shows the total revenue of the embedded-computer industry from 1998
with estimates for 2002, 2003 and 2004. This is a composite of all industry
segments that will be dissected industry by industry. What’s interesting
about the performance of the embedded-computer industry is that it feeds from
so many sectors that it tends to be at least somewhat resilient to depressions
in any one segment of the overall economy. Figure 2 shows the historical book-to-bill
indicating that the industry is definitely in recovery.
It’s pretty clear that the embedded-computer market segment addressing
the communications infrastructure market was hardest hit. Figure 3 shows the
embedded-computer revenues for the sector. And while the future looks relatively
bleak, there is some light at the end of the tunnel.
First off, while many new programs have been cancelled or put on hold, some
have simply scaled back and are expected to accelerate over the next several
months and years. In addition, a lot of equipment that was installed during
the rapid growth of 1999 and 2000, is either wearing out, or is no longer economically
viable. Much of this was PC-based motherboard equipment that required special
cooling equipment, was often made overseas with inferior components and was
not built to handle the rigors of industrial use. In some cases, cooling was
provided by through-the-wall household air conditioners and about the only monitoring
equipment was smoke detectors to detect a unit smoking and perhaps a video camera
to know which one.
What’s emerging is a significant replacement market. Many users fully expected
the “stop-gap” equipment to last only a year or so and to quickly
be replaced by the new paradigm in high-performance, switched-fabric-based gear.
But depressed revenues—partially because of competitive end-user pricing
and partially because of unrealistic expectations—have forced these companies
to keep equipment beyond its anticipated lifetime. Now, however the same shrinking
earnings are forcing these companies to replace this gear with more efficient
and reliable equipment with as little disruption and as low a cost as possible.
Much of this equipment is being replaced with cPCI PICMG 2.16 or similar hardware
including industrial strength motherboards with PMC modules. While this business
will never replace the robust market of 1998, 1999 and 2000, it’s expected
to keep many of the embedded-computer companies at least in operation until
the next wave of activity in the communications business occurs.
That may not be too far off. Suppliers to that industry such as Intel, Microsoft,
Cisco, Dell and Oracle have amassed a large war chest and are looking to create
new markets. Between the five of them, they have almost 90 billion dollars in
cash and liquid assets for use as seed money to develop these new markets.
One of the brighter areas in the embedded-computer
space is the computer market for medical imaging, diagnostics and therapy. This
is lead by the major imaging machines; the MRI machines, PET and CAT scanners
and E-Beam tomography equipment.
The trend is toward smaller, less invasive and more affordable machines. Reductions
in magnetic field, X-ray and positron emissions are being compensated with more
powerful computer resolving power.
But this is only one aspect of the embedded-computer’s contribution to
medicine. Tens of thousands of instruments in hospitals, doctors and dentists
offices ranging from blood-gas analyzers and body function monitors to radiation
and other therapy equipment all use embedded computers ranging from VME and
cPCI to PC/104 and stand-alone SBCs with and without mezzanine add-ons.
But this is just a beginning. New equipment is being developed that uses the
Internet to remotely monitor individuals and equipment. The industry is just
beginning to see some of the benefits of this new Internet-based technology.
Figure 4 shows how the embedded computer market is growing in medical applications.
Major manufacturers of large medical equipment forecast a growth of something
in the area of 8% annually. However, over the past several years, the embedded-computer
component in that market has outpaced that 8% and is estimated to be in the
10% to 12% range.
The military market for embedded computers has been one of the few to prosper
over the past few years. Growth has gone from 12% to 15% from 2000 to 2001 and
is expected to grow at a rate of 18% for 2002 and 21% in 2003. Figure 5 shows
the relative growth of embedded computers in the military market. RTC’s
sister publication, COTS Journal, has published a market update on the military
market in its January issue. That update will be available in the magazine or
at the magazine’s website (http://www.cotsjournalonline.com/marketupdate).
A new and emerging market for embedded computers is
expected in the homeland security and defense area. It’s still too early
to tell exactly what that will comprise, however, early signs indicate that
much new hardware and software will be developed for everything from monitoring
and control to inspection and biometrics.
While some of these developments will obviously fall into the military and COTS
market area, much of it will use more commercial hardware and software. Baggage
scanners, for example, are basically specially modified CAT scanners. Similarly,
other systems that will be used in the homeland security and defense market
will be systems based primarily on commercial products. Though there is no direct
way to determine how this will affect the embedded computer industry, many analysts
suggest that by 2004 the homeland security market will employ more than $4 billion
worth of embedded computers annually for security systems.
Industrial control and automation continue to be significant players in the
mid- and high-end embedded computer market. While the number of embedded computers
in this market sector continue to grow, the real dollars are increasing at a
lesser rate. Figure 6 illustrates the historic growth and projected future of
embedded computers in the industrial control arena. Many early applications
using VME, Multibus or other powerful multi-board computer systems have migrated
to PC-based architectures. Often small single board computers such as EBX form-factor
devices or even PC/104 SBCs are more than enough to power many machine and industrial
control applications. In other less demanding applications, industrial motherboards
serve with PCI slotcard I/O.
The Rest of the Players
While communications, medicine, homeland security and the military are significant
markets within the embedded-computer community, they are by no means the only
players. In fact, one of the characteristics that distinguishes the embedded-computer
industry is the diversity of applications. These range from sophisticated military
control and command systems to retail kiosks; from medical imaging to scientific
exploration; and from automated traffic control and easy-pass systems to all
variety of machine automation and industrial control.
Figure 7 breaks down four of these market areas. It quickly becomes evident
that these areas are all steadily growing at high, single-digit rates. While
this fails to provide the excitement of the 100% growth in the heyday of the
communications ramp up, it’s growth that has been steady for the past
Finally, the fallout from the communications infrastructure implosion has impacted
growth of technology in the embedded-computer arena. The largest impact has
been the change in what the market expected of the timing and technology for
high-speed serial, switched-fabrics. Early on, the lead approach was InfiniBand.
However the cost and time required to implement the technology has intimidated
even some of the major players. Lesser players have defaulted to more comfortable,
Sun, Dell and IBM seem to be holding the course on InfiniBand though IBM has
reportedly dropped its efforts at making IfB silicon. In the meantime, other
IfB former supporters, Intel, Microsoft and HP have reduced or dropped efforts
in that direction—at least until such time as a market for the technology
can be demonstrated.
Other approaches are also slow on the uptake. RapidIO had early support from
DSP chip makers Analog Devices and Texas Instruments, however that interest
appears to be waning. Mercury Computers, one of the RIO developers remains stalwart
to the cause—though little product is emerging. It’s now rumored that
the other developer, Motorola, along with Mercury will implement RIO in some
of its up-coming board-level products.
Even the odds-on favorite, PCI Express is slow to get off the ground. However,
with the advent of PCI Express Advanced Switching probably late next year, the
action is expected to heat up—perhaps in time for a rebound in the communications
business. In the meantime, StarFabric continues to pick up niche areas while
switched Ethernet—either in PICMG 2.16 or the forthcoming ATCA (advanced
telecommunications architecture)—continues to be the de facto standard
for switched serial interconnect. (Look for a lot more on the switched fabric
market in the upcoming Interconnect Strategies section of RTC and COTS Journal
in their February issues.)
Though not the sterling outlook that’s been projected in past years, the
future of the embedded-computer industry remains bright. While the communications
industry is re-inventing itself, other areas such as the military and medical
instrumentation remain bright. And, new applications surface regularly.