Jeff,
upside.com
Cutting Edge
Cheap PCs, Future Markets
July 07, 1998
By Jim Evans
How does the current emphasis on the sub-$1,000 PC affect
Rambus, besides the potential for huge volumes of memory
chips going out the door with your technology? We view the
sub-$1,000 PC as beneficial to us, but it depends on what you believe
a sub-$1,000 PC will be. If you believe it will be good for e-mail and
little else, you could take a 386 processor and build a really cheap PC.
It wouldn't do anything exciting, but then it's not going to need
high-performance processors, high-performance memory or great
graphics. That wouldn't be good for us. We think what the market
wants is great performance, great graphics, great multimedia--and [it]
wants it cheap. When you have a game system or sub-$1,000 PC,
you can't have a lot of semiconductors, otherwise you couldn't sell it at
that price. So you have to get the most performance [from] a small
number of chips, and that also means not that many DRAMs. If you
want to have high performance, each DRAM has to be fast.
You've also ventured into the
communications market.
Non-PC applications are as big a
market for us as the PC
applications. Of course, in the PC
space there are a couple of players,
and it's one market. In the non-PC
space, there are a lot of segments
and a lot of players in those
segments. But the non-PC spaces
are about half our business.
What are the markets?
Consumer, communications and
non-x86 computers, as well as
miscellaneous things such as
copiers and printers. We look to have a good fit for Gigabit Ethernet,
asynchronous transfer mode (ATM), fiber channel, routers and
switches. There are three communications systems shipping with our
technology: a fiber channel product, a video-connect product and
ATM products. We see good prospects for large-volume growth in
communications.
Will communications be a growing percentage of your business?
Yeah, because communications is new. [We didn't ship any]
communications products six months ago. In the consumer area,
between Nintendo and Sony, they'll probably sell 3 million games a
month. That's not up in the PC volume, but it's good volume. We're
also seeing digital imaging (like that of video game systems) going into
TVs, digital [or high-definition] TVs. The first one using Rambus'
technology was shown earlier this year by Panasonic [Industrial Co. of
Secaucus, N.J.] and will be on sale by Christmas. Initially, it will be a
low-volume market, but if [HDTV] is successful, there will come a
time when 50 million of those things a year will be sold. All of them
could be using Rambus' technology. There will also be digital VCRs,
and Rambus has a good fit anyplace there's image manipulation. So the
biggest potential growth for us, after PCs, is the consumer area.
No second thoughts
Was there ever a time during
Rambus' early years when you
said, "Maybe I shouldn't have
left AMD"? There was never a
time when I said I shouldn't have
done it because even in the worst
of times, I was having fun. I wasn't
designed for working in big
companies. I've got a low tolerance
level for committees and
bureaucrats looking over my
shoulder, telling me what I should
be doing when it's supposed to be
my business to run.
In a startup, people are focused
because it's do-or-die. People join
startups because they care passionately about what they're trying to
do--and we all have big financial incentives. But I remember being in
Tokyo in 1991 or 1992, as our cash had gone below $1 million. I'm
going, "What am I doing over here?" That was probably the lowest
point I had. But there were a lot of times during our existence where if
things hadn't clicked, we could have been history. After all, we had to
[persuade] the whole industry to switch to our technology. Not all at
the same time, but we had to [persuade] a lot of them to switch at the
same time.
How did you come to join Rambus? There were two events that
clicked together. One of the venture capitalists I made contact with
[when I left AMD] was Bruce Dunlevie, who was on the board of
Rambus. About the same time, [Rambus Chairman] Bill Davidow
called me after John East, another former AMD executive [currently
CEO of Actel Corp.], gave him my name. In the same 24 hours I
touched base with two guys who were both on the board of Rambus.
At the end of the first meeting I said, "What these guys are doing is a
home-run idea. It's risky, but if we can make the technology work, the
need is obvious." The following week I accepted the job.
Was it tough to pitch your business model to Wall Street when
you went through your IPO road show? The road show was
complicated because there was no company like us in the
semiconductor area. The closest analogies we could point to were the
software companies, but that was quite a ways away, so initially there
was a lot of confusion about the model.
How did you overcome that? Semiconductor companies can have
big disadvantages from the financial community's point of view because
they need large amounts of capital. We were able to show the
advantages of our [licensing] business model. Like a software
company, we could generate cash early and not need this huge
[amount of] capital that a fab-oriented semiconductor company has to
have. Even a fabless company uses a lot of capital for inventory and
for access to wafers.
The recent startups and IPOs in the chip business seem to be
going more in the direction of what Rambus is doing. What has
changed? To some extent, because our IPO was successful, it could
be that a lot of people took their business plans and re-edited them to
make them look like [our] plan, as opposed to having thought about it
that way from the start.
Any examples? No. But success breeds imitators. There are trends
that will result in more companies like us. As chips become more
complex, it's getting harder for a single company to be able to develop
an entire chip. So startups are picking an area of expertise and doing
what we did: licensing the technology.
But the semiconductor IP model is not a guaranteed pathway to riches,
no more than being a fab semiconductor company 20 years ago was a
guarantee. The key to the semiconductor IP model is that you've got to
have a big market, and you've got to [persuade] chip and systems
companies to use the technology. The companies are only going to do
that if there's a big win for them: You've got to have a technology that
has a big value, so you're either reducing cost or creating performance
increases. And the royalties you charge have to be a small fraction of
that value, otherwise there's no incentive for the chip and systems
companies to adopt the technology.
Jim Evans is finance editor at UPSIDE.
----------------------------
|
