| | Named IPO of the Week by the IPO-Fund...
For many years road warrior executives and salespeople found the term "mobile
computing" to be an oxymoron. Anything that was "mobile" enough to take out in
the field was likely not powerful enough to run truly useful applications. On the
flip side, machines advertised as desktop replacements usually turned out to be
awkward and bulky, or lacked sufficient battery power to allow the user to work
untethered from an electrical outlet.
By the time the '90's rolled around many of these problems were solved with the
advent of modern notebook computers---that is with the exception of the battery
power problem. Scant headway was made attacking this problem because as
manufacturers added more processing power, better color screens, and assorted
on-board accessories like CD-ROM drives, any efficiency wrung out of better
battery technology (such as lithium ion) was quickly consumed by ever more
power thirsty devices. As the growth of desktop PCs began to falter and it
became clear that notebook PCs would be the new growth driver, microprocessor
makers like Intel and AMD stepped up efforts to come up with mobile versions of
their desktop processors that used power more efficiently.
Transmeta Corporation, our IPO of the Week, differentiates its chips by
dispensing with the old silicon based approach, which tries to cram more
complexity onto an ever smaller chip. As die size shrinks and chip-makers are
able to increase the number of transistors on a given chip, processing power
should increase. The problem that Intel and AMD's approach runs into is that
there are physical limits to how many transistors can be fit on existing chip
designs, including the dominant x86 architecture.
Beyond just physical limitations and chip complexity come more practical
concerns like power usage, heat dissipation, form factor and cost. The faster a
processor operates the more power it consumes. For a desktop PC, power
consumption is not a major issue. But the usefulness of a mobile computing
device is directly tied to its ability to operate independent of a wired power
source. Even given the advances in technology, most notebook PCs still lack the
ability to operate for any extended period of time without an electrical outlet hook
up. Not only do these devices use more power as processor speeds increase,
but they also generate excess heat. In order to keep to keep a notebook PC
running properly, cooling mechanisms have to be incorporated into the unit.
Adding fans, and the attendant motors to run the fans, increases design
complexity and cost while limiting manufacturers' flexibility to develop smaller
devices with useable feature sets.
Ironically, it was Gordon Moore, the co-founder of Intel, who first postulated the
problem that exponential chip size shrinkage would pose if processing speed
doubled every twelve to twenty-four months as he predicted. Reduced Instruction
Set Computing, or RISC, was one solution that companies such as Motorola,
IBM, ARM Holdings and MIPS have explored as a solution. RISC attempts to
take some of the strain off of the central processing unit (CPU) by having simple
instructions executed out of memory, effectively reducing demands placed on the
CPU. The problem with RISC computing is that processor speeds have increased
faster than memory speeds, creating a bottleneck for tasks executed outside of
the processor. RISC overcomes some of the problems posed by the architecture
employed in Intel-based chips but does so at the cost of processing speed.
Transmeta attempts to sidestep many of these issues by dispensing with a
hardware-centered solution and using software to perform many of the functions
normally performed by the physical processor. The heart of the Transmeta
solution lies in the emulation software that translates the software instruction
sets of x86 architecture-based chips into a simpler set of instructions for the
hardware component to execute. Because the chips have simpler instructions to
process power consumption is reduced. A typical Transmeta chip will use less
than 1 watt of power versus anywhere from six to ten watts of power usage for a
comparably power x86-based chip. With reduced power consumption comes
less of a need for heat dissipators. Manufacturers are not only free to make
smaller devices, but they can also make more powerful devices.
Because the physical design of the chip is less complex, Transmeta is also able
to make more powerful chips at a lower unit cost. That cost advantage is passed
off to consumers. Transmeta chips should eventually command prices of less
than $200. Intel, even manufacturing at volume, does not offer many chips
competitive in that price range. Most such chips offered by Intel are either not
performance leaders or are at the end of their production cycle and are effectively
part of an inventory liquidation program. A less obvious benefit is that
Transmeta's chips are software upgradable. This means that the cost of adding
functionality or optimizing the chip for new applications is cheaper.
Of course if it was this cut and dry everyone would be taking on the
microprocessor giants. Execution risk is the key. There are enough instances of
failed challengers to suggest that budging the incumbents won't be easy. Cyrix
tried and failed. The Power PC threat posed by the alliance of Apple, Motorola,
and IBM was similarly rebuked. And certainly Intel and AMD will not wait around
and cede the marketplace to Transmeta. With their formidable R&D budgets,
access to capital and deep bench of experienced engineers there were be more
challenges coming. Intel is intent on sticking to their mantra that 'only the
paranoid survive' and are hard at work improving its Speed Step technology.
Similarly, AMD is busy with its own efforts for a mobile Athlon chip. Competition
is coming from other directions as well. National Semiconductor claims that they
have designed a chip with lower power requirements than Transmeta's chips.
Apart from all the competition to sign up OEMs, there is just as much as
competition to win over the wallets of end users, including that most important of
all segments, the corporate market. Because the vast majority of laptop buyers
comes from corporate customers, satisfying their demands will be critical to
Transmeta's success initially. Corporate purchasers are a naturally cautious
bunch unwilling to take chances on products that are perceived to offer less
performance or are less reliable. Thanks in part to the wonderfully successful
Intel Inside ad campaign, AMD still struggles with the perception that it is an
off-brand. As the saying in corporate America goes: no one ever got fired for
buying IBM. Ironically, it was IBM that nixed the idea of including a Transmeta
processor in one of its notebooks just this past year. In this case no one got fired
for buying Intel.
Despite the possible pitfalls, it is clear that Transmeta has changed the terms of
the debate. If imitation is the sincerest form of flattery, then it is telling that a
large incumbent like Intel starts to talk like its smaller upstart brethren about
battery power and portability. That is enough to tell us that Transmeta is on to
something and a tacit acknowledgement of the merit of its challenge.
ipo-fund.com |
|
