Hey Baldy and Clap,
You guys are great. But if the puns get any worse I will have you both banished to a gulag in Siberia!
Now, for some real news:
Making the Crusoe Argument -- Nov 10 2000
New York, Nov 10, 2000 (123Jump via COMTEX) -- Transmeta Corp. (TMTA) develops
and sells software-based microprocessors and develops additional hardware and
software technologies for Mobile Internet Computers - portable computing and
communication devices designed to provide an Internet experience comparable to
that traditionally supplied by a desktop personal computer, or PC. The company's
Crusoe family of microprocessors is targeted at the notebook and Internet
appliance segments of the Mobile Internet Computer market, and it provides
Crusoe microprocessors to suit both existing and emerging products within these
market segments. Rather than using the traditional method of designing
microprocessors primarily with hardware, Transmeta has developed a novel
approach which incorporates a substantial portion of microprocessor
functionality into its software, a process they've dubbed "Code Morphing." This
approach to designing microprocessors provides the first microprocessor solution
that simultaneously satisfies a wide range of user requirements for Mobile
Internet Computers, including compatibility with PC software, long battery life
and performance comparable to a desktop PC. The power-thrifty Crusoe is designed
to extend the battery life of portable computers and next-generation Internet
access devices. Transmeta has relied primarily on licensing agreements with IBM
(IBM) - which also manufactures the company's chips - and Toshiba (TOSBF) for
nearly all of its sales. Chip legend David Ditzel co-founded the company; Linux
operating system creator Linus Torvalds is on the development team. Entities
affiliated with Institutional Venture Partners own a small stake in Transmeta.
Transmeta filed regulatory documents on August 17 and went public on November 7
of this year. The offering was lead by Morgan Stanley Dean Witter, and is
co-managed by Deutsche Bank Alex. Brown, Salomon Smith Barney, Bank of America
Securities LLC, and SG Cowen. The size of the offering was 13 million shares
with pricing set at $21 per unit - well above its expected debut range of $16 to
$18. Strong investor demand saw the stock surge as high as $50.87 per share
before closing at $45.25 on opening day. The maximum offering value was $273
million, with an estimated potential market cap of $2.17 billion, based on 127.8
million shares outstanding after the IPO. There has been a world of speculation
about how good Transmeta's new Crusoe chip really is and will the company be
successful at cracking the market dominated by the Microsoft of chipmakers,
Intel (INTC). The company's stock surged more than 130% on its first day of
trading, leading many to speculate the company is overvalued. The company has
yet to make a profit, and has $120 million long-term debt. Transmeta has had
paltry revenues. For the six months ended June 30, 2000, it reported total
revenues of $348,000 with a pre-tax loss of $43.4 million. The company has
$112.4 million in cash. Considering these financial numbers, we are left
scratching our head while asking the question ... "why did Transmeta's IPO go so
well?" IPO's are all about gambles and Transmeta is entering a sufficiently
valuable market, meaning success could be richly rewarded. If we look at some
key issues we can discern that Transmeta stands a good chance of acquiring a
solid foothold in this market place. First of all a direct comparison between
Intel's and Transmeta's products (under current performance testing benchmarks)
is not possible, mainly due to differences in chip design. Intel's products are
based on the CISC (Complex Instruction Set Computing) concept which is founded
on the principle of incorporating large numbers of transistors onto a chip.
Intel has used steady advances in chip-building techniques to squeeze ever
greater numbers of transistors onto each generation of its chips, and therefore,
greater speed and performance. However, this design concept has a drawback ? the
greater the number of metal transistors, the greater the power consumption and,
consequently, the greater the build up of heat. Power consumption is not a real
concern for desktop computers which have an abundant power supply from a wall
socket, nor is heat build up a burden since a fan can easily be accommodated by
the computer casing. The same cannot be said of portable computers. With
notebooks, cooling is the more significant issue. Because of the condensed space
large fans are not possible, therefore more creative solutions are required. One
has to only recall that Intel's higher end mobile Pentium III processors were
delayed to market last year precisely over this issue ? the units were running
too hot. Then there is power consumption. A typical lithium ion battery in
notebooks today has an operating range of between two and three hours, depending
on usage. In contrast, Apple (AAPL) laptops (which use a design concept similar
to Transmeta's) typically have a 6 hour operating window. The main reason for
this is the RISC (Reduced Instruction Set Computing) concept, a chip building
technique that simplifies processor hardware and shifts the more complex action
to software. Transmeta used the RISC concept, or rather its successor VLIW (Very
Long Instruction Word), to develop its "Code Morphing" technology (the
chameleonlike ability to mimic the instructions for other microprocessors) used
in its Crusoe chips. This design concept allowed Transmeta to produce a chip
close to the processing speeds of Intel's products with one-tenth the power
consumption of Intel's best chips. Since Transmeta's product is the first real
innovation in the chip making industry in the last 14 years, a debate has broken
out over whether the Crusoe is truly innovative at all. Some industry analysts
point to the cool reception Transmeta has received from the big North American
computer companies, notably Compaq (CPQ), HP (HWP) and IBM. They point to the
fact that IBM's decision not to incorporate the Crusoe chip in its notebooks,
after being Transmeta's biggest proponent, suggests that there is a flaw in the
processor. To date there has been only one independent performance test of the
Crusoe chip, in which a 700 Mhz unit was pitted against a mobile Pentium III
processor running at the same clock speed. The tests confirmed that the Crusoe
chip was significantly slower, offering performance equivalence more akin to a
400-500 Mhz Pentium III. However, the group conducting the test agreed that the
Crusoe chip was faster every time they repeated the test. The primary reason for
this is the fact that instead of reading incoming command lines from the
hardware level, as does the Pentium III, the Crusoe chip's code morphing
technique reads the commands at the software level. According to Transmeta, the
advantage of software emulation is that it creates a smarter processor that
operates more efficiently on repeated commands that the chip has already
translated. This process can be best likened to memory cache which takes longer
when you first access it but becomes quicker every subsequent time. Dean
McCarron, an analyst with Mercury Research in Scottsdale, Ariz., argues that the
only way to solve this performance degradation is to increase the clock speed of
the processor, however, in doing so the power saving advantage the Crusoe enjoys
would be lost. McCarron points out that, in the past, companies such as Digital
(now owned by Compaq) tried the software emulation approach with its Alpha
processors but ran into the same performance degradation problems. Performance
degradation is being cited as the primary reason for Crusoe's cool reception
from North American computer producers. Most manufacturers have adopted a "wait
and see" approach. However, IBM's decision last month to drop Crusoe from its
product line up could have more to do with marketing than performance. In
November of 1999 and February 2000, Transmeta amended its technology-licencing
agreements with IBM and Toshiba respectively. Under these licensing deals,
reached in 1997 with IBM and 1998 with Toshiba, Transmeta gave both companies
the right to manufacture and market X86-compatible products based on its
technology. IBM and Toshiba still have the right to make and market Transmeta's
products, but neither company can use the technology to underpin an
X86-compatible processor. Most industry analysts agree that this licensing
buyback will benefit Transmeta in the long-run, but at the same time believe the
company has taken a huge risk in doing so. With most of the company's revenue in
1997, 1998 and 1999 coming from these licensing arrangements, Transmeta is now
dependent solely on its own product sales mainly through OEM's and distributors.
However, this seems to be changing. Unlike their North American counterparts,
Japanese computer manufacturers have been quite receptive to Crusoe. In fact
Sony (SNE), Fujitsu (FJTSY) and Hitachi (HIT) have all released products
equipped with the Crusoe processor, with NEC (NIPNY) expected to roll out a
notebook based on the chip sometime next year. The mere fact that four large
Japanese computer manufacturers have adopted Crusoe suggests that Transmeta's
product is more than up to its task. Instead of forecasting Transmeta's future
in a simple black and white, David vs. Goliath way, we must step back and put
things into perspective. Transmeta is not aiming to compete with Intel or AMD
(AMD) directly ... at least not for now. The company's main focus is on the
portable computing and emerging Internet appliance niche. Let's face it, how
many people need a 1 Ghz processor to surf the Net? For most consumer
applications a 500 Mhz processor is more than enough unless used for gaming, and
how many gamers buy laptops? ? Very few. Unfortunately, marketing has got the
better of consumers and analysts. With most advertisements pushing processor
speed as the main factor in computing, consumers have bought the hype that in
order to have the best product you must have the quickest processor. Sure it's
nice to have speed, but a notebook user is more concerned with battery life and
(unwittingly) heat. If these criteria were used for benchmark testing in the
notebook market, Transmeta would win hands down. The one thing all analysts can
agree on is the fact that Crusoe's power stinginess is accurate. The processor
consumes 1 watt of power at 700 Mhz, as opposed to Intel's 1.5 at the same clock
speed. Both Intel and AMD have introduced new power saving features - SmartStep
from Intel and PowerNow from AMD - as interim solutions until the launch of new
products next year. However, these features pale in comparison to the
sophistication of Crusoe. Using its Code Morphing capabilities the Crusoe chip
determines whether or not you need all the processing power available. For
example, if you have a 700 Mhz chip, but only need 500 Mhz to achieve a
particular task, the processor will automatically scale back its speed to meet
your requirements. Conventional CISC processors attempt to regulate power by
running the processor at full speed then halting it to a near stop, this process
is known as "Duty Cycle." In doing so, the processor may shut off at a critical
moment in an application operation. For the user this can be perceptible and
annoying ? especially during movies. In comparison, Crusoe regulates power by
rapidly adjusting its clock speed to whatever level the application requires,
without requiring the operating system to restart or re-boot the RAM. Since the
rapid switching occurs at such a fast pace, the change is not noticeable to the
user. This greatly improves battery power conservation, eliminates fans and
decreases the overall weight of the notebook. Another benefit of Code Morphing
is cost. By switching more of the complex processing to software you eliminate
the number of transistors on the chip and thus reduce the cost of manufacturing.
In fact, the cost of Transmeta's chips range from $119 to $329, as opposed to
Intel's mobile Pentium III products which typically range from $508 to $722.
Another key benefit of Code Morphing is that it greatly reduces the likelihood
of manufacturing errors and since most of the processor functionality is done in
software, Transmeta could provide performance updates for the Crusoe immediately
via its Web site - compared with long waits for next generation CISC processors.
That Transmeta has a good thing going is recognized, at least tacitly, by Intel.
According to Dean McCarron, and based on Intel's mobile chip launch this June, a
recent re-organization took place within Intel to look at new ways of building
mobile processors. According to McCarron, Intel has decided to stop taking
desktop processor components and re-packaging them as mobile processors, opting
instead to design new mobile chips from the ground up. If this is the case, then
recent comments espoused by Transmeta's co-founder David Ditzel that the company
enjoyed a five-year lead over Intel could have substance. Intel has consistently
introduced new CISC based products in two-year cycles over the past ten years.
If it were to start from scratch, by perhaps switching closer to RISC processing
or other technologies, it would indicate that development time for a new product
could ostensibly double the current two-year Intel cycle. In the meantime, the
company announced this week that it completed the development of its 0.13 micron
(130 nanometer) generation logic technology, allowing it to manufacture chips
with transistors that are approximately 1/1000th the width of a human hair. This
advanced process technology will begin volume production next year and deliver a
new generation of high performance microprocessors, which may contain more that
100 million transistors and run at multi-Ghz clock speeds. Intel claims that its
130nm process will operate at 1.3 volts or less, lowering the voltage over
current chips by 20%. There is no doubt that Transmeta has a "real" product. The
company currently has a technological advantage over the established industry
players, but the clock is ticking. Intel is a slow moving giant but it is
nevertheless a giant, it has stood up and taken notice of what Transmeta has
produced. Intel will eventually hit back with a competitive product, therefore,
Transmeta needs to aggressively push into the market it is trying to penetrate
while at the same time working hard to diminish the performance gap between
itself and Intel. The company possesses significant intellectual capital and is
spending a great deal on research and development, indicating that they are
capable of overcoming the issue of performance degradation. If Transmeta is
successful at persuading computer manufacturers to incorporate its products, the
main issue will become manufacturing capacity. The company does not have its own
manufacturing capabilities but relies on IBM facilities for all of its
production. This could change in the future as there have been reports that
Transmeta has approached Taiwan Semiconductor Manufacturing about possible chip
production. All in all, the chip industry is set to heat up in the next few
years. Will Transmeta succeed? That depends on the skills of its leadership, but
it certainly seems they have taken the lead in the first half of this match.
CONTACT: For more information, contact 123Jump.com, Inc.
212-968-8700
Send email to: info@123jump.com
Or, visit 123Jump at: 123jump.com
All Rights Reserved. (c) Copyright: 2000 123jump.com, Inc. |
