Posted at 4:46 p.m. PDT Sunday, July 4, 1999
AMD's Next Round With
Intel
Amid K6 sales gloom, the super-fast Athlon chip gives it
a fighting chance to make points in PC market
BY TOM QUINLAN
Mercury News Staff Writer
THE clouds that seem to hover perpetually over 1 AMD Place in
Sunnyvale have never seemed darker.
Last month Advanced Micro Devices Inc. announced that it
expected to report an operating loss of more than $200 million for
the quarter that ended in June -- largely because it was unable to
sell about 40 percent of the microprocessor chips it made during
that three-month span. A number of manufacturers had essentially
replaced the AMD K6 line with Intel Corp.'s Celeron processors.
And even as the company starts shipping
its super-fast Athlon processor, previously
known as the K7, personal computers that
use AMD's revolutionary design won't start
showing up until August at the earliest. It's
highly unlikely, AMD officials
acknowledge, that the company will become profitable this year.
Survival might be considered a stretch, let alone success. But as
dark as the clouds gathering around AMD are, much of the industry
is instead focusing on the silver lining.
''AMD's future is in its own hands now,'' noted Nathan
Brookwood, founder of the market research firm Insight 64. ''It's
been a long time since that was the case.''
Even former skeptics now believe that AMD has a good chance of
creating a significant -- and profitable -- niche for itself in the PC
market, something they weren't that confident about previously.
''I would say that AMD has a pretty good chance with Athlon,''
noted Michael Feibus of Mercury Research.
Two months ago Feibus questioned whether AMD would have the
ability to create and support the other products -- such as chip sets
and mother boards -- the chip would need to ultimately succeed.
Now, however, ''there are problems (in doing that) but they're the
sort of problems that occur with every new generation of chips.''
Optimism surrounding AMD's long-term future revolves around the
technical prowess of the Athlon itself and the fact that Intel -- rather
than being able to focus on exploiting the weaknesses of its rivals --
has its own problems to deal with.
The Athlon is a breakthrough product in a number of ways for AMD. Although it still uses the same
core set of instructions as Intel's processor line to maintain compatibility, the chip's system bus --
the electrical connection between memory and the processor -- at 200 MHz is twice as fast as that
used in Intel's current line of processors.
The floating point unit, which is heavily used in graphics and modeling applications, is also different,
and, according to critics, far faster than that used in Intel's Pentium III processor line. And with
speeds ranging from 500 to 600 MHz, Athlon's clock rate will at least match anything Intel can
offer.
Although AMD has not released any details about Athlon's performance yet, some sources who
have tested the chip described it as consistently outperforming Intel's line by anywhere from 5 to 50
percent, depending on the application being used. It's a perception that some computer company
executives also support.
''This chip is a screamer,'' noted an executive for one computer maker that has been demonstrating
an Athlon-based PC behind closed doors. ''The performance buyer is going to demand this chip.''
That's quite a change from AMD's traditional market niche. Despite AMD's best efforts, the K6
line has been overwhelmingly found in PCs priced at less then $1,000, where the competition was
fierce and the profits slim -- especially after Intel started introducing more powerful Celeron chips
with a starting price of less than $75.
Key customers
But AMD's most significant customers -- including Hewlett-Packard Co., International Business
Machines Corp., Compaq Computer Corp. and Gateway 2000 -- are all showing prototypes of
PCs that feature the Athlon and prices that approach the $2,500 mark, sources familiar with the
systems said.
If AMD can continue to introduce improvements to the Athlon line, it should be able to gain market
share and consistently make profits, complementary goals that have proved elusive for AMD.
''AMD has demonstrated that the mix of processors you sell is incredibly important. Intel's strength
is that they have a near-monopoly at the high end where profits are much higher,'' said Scott
Randall, of Soundview Technology Group. Although AMD overall holds about 16 percent of the
microprocessor market -- and has exceeded 50 percent in some market segments at various times
-- AMD's continuing losses have shown its market share gains to be an expensive victory,''
Brookwood said. ''It's been profitless prosperity.''
In large part, that reflects Intel's aggressive competition after entering the market far later than rivals
AMD and Cyrix.
Although Intel didn't introduce its first Celeron until 1998 -- almost a year after the first sub-$1,000
PCs from Compaq started shipping -- by January of this year Intel was able to promise Celeron
processors with clock speeds of 500 MHz and prices that started at less than $75.
And Intel basically threw out its once-sacred pricing model when it came to the low end. If
competitors were offering a PC manufacturer a special deal, Intel could offer to meet or beat it, a
dramatic departure from a one-size-fits-all price list that only responded to volume.
''We're going to do what it takes to be competitive in this market,'' noted Paul Otellini, executive
vice president and general manager of Intel Architecture Business Group, outlining Intel's Celeron
strategy earlier this year.
The results were dramatic. Intel gained market share, and prices for PCs continued to drop sharply.
The sub-$1,000 PC market morphed into the sub-$800 market and finally the under-$500 market.
Profits suffered an equally sharp drop.
Unable to match Intel's performance, or to achieve profitability as chip prices continued to drop,
National Semiconductor Corp. pulled the plug on its Cyrix microprocessor subsidiary, finally
agreeing last week to sell control of the division to Taiwan chip-set maker Via Technologies Inc.
When National dropped out, AMD founder and CEO Jerry Sanders noted that to compete with
Intel a company needed to commit hundreds of millions of dollars a quarter, and to decide whether
they were in for the long haul or out. ''And we're in,'' Sanders said at the time.
Complaints from Sanders
That didn't stop Sanders from complaining on numerous occasions that Intel, fat from profits
derived from the powerful processor that went into higher-priced desktop computers, workstations
and servers, was cutting prices so low on entry-level systems that it was impossible for anyone to
make a profit.
Certainly it became increasingly difficult for AMD. Not counting the most recent quarter's expected
$200 million loss, of AMD's last four quarterly reports two were profitable and two were not, with
the net result a $169.5 million loss.
For its part, Intel spokesmen insist that the Celeron line, however low it was priced, has
consistently contributed both to its cash flow and to its bottom line.
Intel's aggressiveness -- and the steady improvement of its Celeron line -- also forced AMD to try
to stretch the performance of the K6 far beyond what it was originally designed for. The result was
a breakdown in AMD's manufacturing capabilities that prevented AMD from making enough chips
to meet demand for more than six months.
By the time AMD solved the problem, the market had moved to Intel, leaving AMD with a surplus
of 2.3 million chips, and a reputation for being an unreliable supplier. A number of customers went
with Intel, and, AMD executives acknowledge, haven't come back yet.
Those are a lot of problems for one chip design -- and an unproven one -- to solve. But many
familiar with the Athlon give it a fighting chance.
''AMD has never consistently been able to match the performance of Intel's line, let alone beat it,''
Brookwood said.
The Athlon could be different, however.
''It's likely that AMD will have the best performing desktop chip on the market for the rest of this
year and until mid-2000 at least,'' Brookwood added.
To a certain extent that's because Intel has run into some manufacturing problems of its own.
Coppermine, the first chip Intel expects to produce using the same 0.18 micron process that AMD
is using for Athlon, has been delayed until late September at the earliest.
New chip-set design
Coppermine also uses a faster system bus, but much of the chip's performance will be tied into a
new chip set design that supports a new type of memory known as RDRAM, designed and
licensed by Rambus Inc.
''I absolutely believe that ultimately, Rambus' memory will offer better performance than standard
memory,'' Feibus said, comparing the new memory chip design to the static RAM memory chips
used to increase performance today. ''But it won't be faster initially. I think Intel just didn't realize
that there was so much life left in SRAM.''
More broadly, however, Intel may simply not be focusing enough attention on its competitors,
Feibus suggested.
In the past, ''you could always count on Intel having something up its sleeve that it could use to
pound the competition,'' he said. ''It doesn't look like they have anything for Athlon, though.''
But as Sanders has noted in the past, AMD is competing with an 800-pound gorilla, and one that
doesn't shy away from a battle.
''I think you can expect to see Intel be very aggressive in this marketplace, just as they were in the
low end with Celeron,'' said Mark Edelstone, a financial analyst for Morgan Stanley Dean Witter.
''AMD will be lucky to produce a million (Athlon) chips this year,'' Edelstone added, a prediction
not too much different from AMD's own forecasts, and a factor that could keep Intel's response
rather mild for the rest of this year.
But if AMD successfully ramps up production of Athlon in 2000, expect Intel to respond in much
the same way they did in the low end of the market. ''Intel has shown how aggressive they can be
and I expect them to be every bit as aggressive in this market,'' Edelstone said.
|
