SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?


To: Cirruslvr who wrote (46315)1/16/1999 4:46:00 PM
From: Cirruslvr  Read Replies (2) | Respond to of 1571591
 
K6-3 500 to come out with the other k6-3s!?
______________________________________________________________________
Manufacturing glitch drags AMD below Wall Street
estimates

By Mark Hachman
Electronic Buyers' News
(01/13/99, 09:42:33 PM EDT)

Advanced Micro Devices' inability to meet its earnings projections for its fourth
quarter further roiled analysts' stomachs, already sour after AMD's
microprocessor prices failed to meet expectations.

AMD, Sunnyvale, Calif., did turn a profit for the quarter, recording $22.3 million
of net income on a 29% increase in revenue, to $788.8 million. A year ago, AMD
lost $12.3 million while posting revenue of $613.1 million.

For 1998 in total, AMD again reported record revenue of $2.5 billion, up 8% from
1997. But AMD's $104.0 million net loss widened from the year before, when it
lost $21.1 million.

Analysts chose to focus on two negative points. First, AMD's earnings per share of
15 cents missed the consensus estimate of 19 cents a share; that loss also factored
about $12 million in unexpected pre-tax savings, making the case that AMD's
earnings could have been half what the company reported.

Second, AMD's fatal flaw again plagued the company. A manufacturing glitch,
which also contributed to AMD's net loss of $55.8 milion in the first quarter of
1998, caused some K6-2 processors designed for 400 MHz to only run accurately
at speeds of 350 MHz or below. That forced AMD to lower their prices. Although
the lithography error proved to be relatively easy to fix, the flawed chip
replacements will be available only in February.

The 850,000 to 1 million microprocessor units affected by the flaw equated to
about a $35 to $40 decrease in the average K6-2 price, said W.J. “Jerry” Sanders
III, AMD's chairman and chief executive. Essentially, AMD lost “at least” $20
million and as much as $40 million from that flaw alone, Sanders said.

The unexpected drop in prices was unsettling to some. “I wish I had brought some
airsick bags,” said A.A. “Tad” LaFountain III, analyst with Needham & Co., New
York.

“Terrible ASPs, just horrible,” complained another Wall Street analyst.

AMD's average selling price (ASP) of its microprocessors was about $88 or $89,
Sanders said; the company had previously set a $100 ASP target for the fourth
quarter and for 1999, spurred by sales of its new K6-2 chips.

Nevertheless, AMD shipped 5.5 million processors during the fourth quarter, of
which about 400,000 were at speeds of 400 MHz and above. Of that tally, about
1.4 million and 1.3 million chips were shipped to Europe and Asia, respectively.

AMD's share of the sub-$1,000 U.S retail PC market grew from 33% in November
to 36.5% in December, according to AMD's tallies -- even after Intel began cutting
the prices of its Celeron processors. “Once again, the Computational Products
Group did a great job,” Sanders said.

Like Intel Corp., Sanders said AMD's production would be ramped up to exceed
its first quarter 1999 production estimate, pegged to remain steady at the 5.5
million units AMD manufactured in the fourth quarter. AMD's goal is to increase
wafer production from 40,000 wafer starts per month now, to 60,000 starts at the
end of 1999. During the holiday selling season, AMD was unable to satisfy
demand for its microprocessors, Sanders said.

Sanders added that an “acceleration of acceptance” of the AMD brand should also
contribute to increased adoption of the forthcoming K6-3 processor, built from the
same K6-2 core as the flawed chips but with on-chip cache memory.

The K6-3 launch has not been delayed, although Sanders said he wants to be
assured that the K6-3 will ship at 500 MHz when initial K6-3 shipments begin in
the first quarter
. During that period, AMD should announce a “top-tier” customer
both in the desktop and mobile computing market
, Sanders said.


More important will be a push towards faster, more expensive processors.
Although flat ASPs are unlikely, AMD has hitched its wagon to its microprocessor
prices.

“If ASPs do not move up it will be difficult to exceed modest growth for the year,”
Sanders said. But he added that AMD could grow 10% if ASPs remain flat, which
he considered a doubtful proposition.

Right now, however, the K6-2 is being priced against Intel's low-cost Celeron
brand, not what one analyst wanted to hear.

“Big revenue boost, big unit upside, but my thesis for the company is that they're
not getting paid for performance,” said Scott Randall, analyst at SoundView
Financial Group, Stamford, Conn.

“AMD's stated reality is their willingness to price against Celeron,” Randall
added. “The K6-2 is a better product, with distinct differences, but it's all about
clock speed and memory [in PCs].”

Intel can also now afford to play low-ball in the low-cost PC market,
compensating with its high-margin Xeon processors, LaFountain said. “When the
K7 comes out, AMD can play the same game [against Xeon],” he said.
“Everyone's always talking about a winner and a loser...the winner is the
customer, and the loser is profitability.”

Although AMD's non-memory businesses reported flat growth, Sanders said it was
virtually certain that its communications, programmable logic, and memory groups
lost money, especially in flash memory.

“Our working hypothesis is that non-microprocessor sales will be flat in Q1, with
growth resuming in Q2,” thanks to new communications products and high-density
devices from its PLD subsidiary, Vantis Corp, said Ben Anixter, vice-president of
external affairs. Sanders further qualified the statement, saying the model excluded
flash memory, whose ASP is declining at about 35% to 40% annually.

Wall Street analysts also noted that AMD had unexpectedly accounted for $3
million in interest expenditures in the third quarter 1998. In addition, analysts said
they had been asked to project $20 million in research expenses tied to AMD's
partnership with Motorola. Only $11 million was actually spent.

Sanders specifically apologized to one analyst concerning the R&D expense,
stating that the uncertainty was due to the inability to accurately forecast
Motorola's own spending.

Analysts' models were also further confused by the dilutive power of an additional
20 million shares of stock AMD plans to issue at some unknown date.

“When you operate at break-even, problems become magnified,” LaFountain said.
“What was the effect of Intel's delay of its 450-MHz Xeon processors? You never
saw it.”
ebns.com
______________________________________________________________________

I wonder why the article said Sanders wants to be assured.



To: Cirruslvr who wrote (46315)1/16/1999 4:48:00 PM
From: Pravin Kamdar  Read Replies (1) | Respond to of 1571591
 
Must read:

techweb.com

Intel may be backing themselves into a corner. DDR SDRAM will be very popular. I suspect that Intel will use it with the Celeron line.

Pravin.



To: Cirruslvr who wrote (46315)1/17/1999 1:11:00 AM
From: Brian Hutcheson  Read Replies (1) | Respond to of 1571591
 
C L , re. were the process problems only with .35
On the .35 they could not ramp the volume while maintaining clock speed increase . They had problems obtaining 266mhz parts and any kind of volume even on lower clock 200mhz.
They started .25 at the beginning of '98 and produced over 10 million
K6 with speeds up to 400mhz .
So the .25 process has been much more successful .
They have obviously had some problems as the latest CC indicates but if the K6-3 and K7 are initially being produced on .25 or .22 they should have a good ramp as the .25 is a proven process for them .
The K7 on .22 IMO should be better than Intel's offerings
Brian