<font color=red>a must read, imho. Intel blames rambus for its problems. Journey of the Rambus
from a saviour to a scapegoat in just a short time..... <ggg> Intel admits that pIII is not competitive
design. Considering that it is intel's main product for at least another year ......
Intel is expecting p4 to have lower margins......
09:32am EST 30-Oct-00 J.P. Morgan (RAGSDALE, TERRY (1-212) 648-9047) INTC INTC.
INTEL: RECENT MANAGEMENT MEETING --- PART I OF II
October 30, 2000
J.P. MORGAN SECURITIES INC. - EQUITY RESEARCH
TERRY RAGSDALE (1-212) 648-9047 ragsdale_terry@jpmorgan.com
Daan D. Coster (1-212) 648-3806 coster_daan@jpmorgan.com
Intel (BUY)
PART I OF II
RECENT MANAGEMENT MEETING: SOME INTERESTING BITS AND PIECES
Earnings Per Share P/E
INTC 52-Wk ------------------ ------------ MkCap
10/27 Rge 12/99 12/00 12/01 4Q/00 4Q/99 12/00E 12/01E Yld ($MM)
---- ----- ----- ----- ----- ----- ----- ---- ---- ---- -----
$46.38 $76-34 $1.17A $1.69E $1.72E $0.42E $0.50A 36.4 35.6 0.3% $311,362
Note: 12/00 and 12/01 estimates include investment gains of $0.39 and
$0.27 per share, respectively.
We met last week with Paul Otellini, head of Intel's microprocessor
business, and walked away with some decent color on various aspects of
Intel's business. Nothing to trigger a trade, mind you, but some tid-
bits. Our position on Intel's stock remains the same: We were not smart
enough to hop off the bus in the $70s, and we have no inclination to do so
in the $40s. We think a lot of bad news (and conservative modeling
assumptions) is already priced into Intel. The stock finally has
outperformed recently, up over 30% since its October 11 low of $35,
compared with a slight decline for the Philadelphia Semiconductor Index.
No, there's not much of a case for pounding the table at the moment, but
investor psychology on the stock is lousy, and historically that has
proven to be a good time to buy it. At 36 times 2001 earnings (including
forecasted investment gains), it's hard to call the stock cheap, but the
market has tended not to care much about Intel's valuation: When things
are good, no price is too high, and when things are not good, no price is
low enough. Our guess is that the news flow will improve from here,
especially on the launch of the P4 microprocessor in about three weeks.
Also, Intel will be hosting its webcast analyst meeting later this week
(November 1), and probably the theme of the meeting will not be how awful
things are. We suspect that Intel will provide further information on the
product roadmap and plans for the P4 launch and ramp. We maintain our BUY
rating on the stock. With the stock currently trading at $46, we need to
do something about our $46 target price, and we will revisit that issue
following the webcast.
In the new world of Reg FD, one has to temper one's expectations about
gleaning stock-moving information from meetings with management, unless
it's in the context of a widely attended analyst meeting. Even then, the
odds are that the major stuff will be disclosed in a press release before
the event starts. Intel's position on Reg FD is that the company's
investor relations policy has been consistent with FD for a long time.
Earnings guidance, for example, is always included in the quarterly
earnings release (and in considerable detail, including most income
statement line items), and any update to guidance has always been done via
a press release. Intel generally has been happy to provide color on its
business to analysts and investors in private meetings with management,
but managers clearly are coached not to spill material beans.
We discussed the following in our meeting with Mr. Otellini:
ú Earnings/PC market guidance: Mr. Otellini refused to
characterize Intel's 4Q/00 guidance and view of the PC market as
"conservative," preferring to use the word "cautious." In other words,
Intel sees reasons not to be too aggressive, but don't assume that there's
upside just yet. He said that Intel has not seen the weakness in Europe
spread to other regions (whew). The key issue is inventories at the OEM
level. Intel is comfortable with inventories in the distribution channel
(both chip distributors and PC resellers) and on its own books, but the
visibility on PC OEMs' inventories of microprocessors is less clear.
Finally, Mr. Otellini said that Intel can ship more than its guidance if
the demand is there.
ú Competitive situation: Asked about Intel's intentions on
pricing, Mr. Otellini was not very specific, but he provided some color.
Intel's microprocessor business was flat in units terms in 3Q/00, and
AMD's business was up 700,000 units. So, yes, AMD gained share, but not
very much. Mr. Otellini said that Intel expects to gain share in 4Q/00
but that the company's share remains in a comfortable historical range,
implying historical pricing trends. He also noted that Intel's earnings
guidance for 4Q/00 implies that Intel is not expecting serious erosion in
average selling prices (ASPs). One question that we remain unclear on its
Intel's gross margin situation in 2001. The company has said that it
expects to continue reducing manufacturing cost per unit sequentially for
the next four quarters through 3Q/01, but we don't see how this works.
The 0.18 micron ramp will be 90% done by 4Q/00, as will the conversion to
flip chip packaging. P4s will replace PIIIs in 1H/01, and it seems
inevitable that those margins will decline (larger die size). In mid-
2001, Intel will start ramping up 0.13 micron manufacturing technology,
which we would think would be a drag on margins initially. Perhaps the
answer is yield improvements on 0.18 micron technology, although Intel has
gotten so good at ramping manufacturing that the initial yields probably
already were pretty good. This one is a mystery, but maybe we'll get more
specifics on the webcast.
ú P4 ramp: According to the trade press (but not publicly
confirmed by Intel), the Pentium 4 will launch on November 20. It was not
so long ago that Intel tended to agree with the assertion that the Pentium
4 ramp would be pretty modest initially. The chip size is rather large
(implying high manufacturing cost), and the infrastructure, especially
Rambus DRAM, has been expensive and in limited availability. Intel's
plans have changed. Mr. Otellini said that Intel's confidence in P4 and
its ability to ramp to volume production naturally has increased as the
launch date has gotten closer. P4 will not be narrowly targeted at
workstations and servers like the Pentium Pro back in 1995. Mr. Otellini
was rather frank about the die size issue: The PC market is not as strong
as Intel had expected, which frees up some capacity to ramp P4 faster. In
effect, Intel has committed to fixed manufacturing costs, which means that
using the capacity to make more P4s is "free" compared with not using the
capacity at all. On the infrastructure side, Intel is comfortable with
both availability and cost and sees no "volume limiters." The recent
trend of lower memory prices helps. Although Mr. Otellini admitted that
cost reductions across the P4 platform (e.g., shrinking the processor on
0.13 micron technology, introducing a chipset that supports SDRAM, etc.)
will be important to continue ramping to even higher volumes in 2H/01, he
sees no problem getting to substantial volume in 1H/01.
ú P4 pricing will not be much different from the pricing of PII and PIII at
initial launch. P4 wafers will yield better margins than Celeron and will
be a good trade-off for Intel. Mr. Otellini was not willing to discuss
system price points on P4 launch to avoid preannouncing customers'
products. Mr. Otellini would not be drawn into a discussion of future P4
clock speeds beyond the 1.4 GHz level at launch, but the company's product
roadmap has leaked to the trade press, and it includes 1.7 GHz in 1Q/01
and 2.0 GHz in 2Q/01. This is a pretty good story, as clock speeds sell
processors. Mr. Otellini said that P4 performance will be roughly
equivalent to PIII on a clock-for-clock basis, but there will be a clear
clock speed separation between the high end of PIII and the low end of P4
(unlike when PIII was introduced and overlapped with PII).
ú Rambus pricing issues: Mr. Otellini admitted that Rambus memory
prices remain at substantial premiums to mainstream SDRAM and said that
Intel will have to help with subsidies this quarter and next quarter; he
called this "pump priming" to get the Rambus market to critical mass. Mr.
Otellini was no more specific than that, but the trade press is.
According to Electronic Buyers' News (EBN), Intel is offering PC OEMs a
$70 rebate per P4, not huge in the context of prices that are expected at
$950-975 for a 1.4 GHz P4 and over $1,000 for a 1.5 GHz P4. Further,
according to EBN, Intel is selling P4 to the non-OEM channels (motherboard
manufacturers, distributors, and resellers) bundled with Rambus DRAM
modules, thereby addressing both availability and pricing. EBN does not
say what the subsidy is in this case but does note that 64 megabyte Rambus
modules trade in the spot market currently at $200-250, compared with $58
for the equivalent SDRAM (133 MHz).
ú Transmeta: Asked about the very well publicized, super-power
efficient Transmeta Crusoe microprocessor for laptops, Mr. Otellini said
that "Transmeta is about to speak for itself." He didn't mean Transmeta's
IPO. Rather, he was referring to the first tests and benchmarks reported
in the trade press, specifically PC World and CT Magazine. Our German is
worse than rusty, so we couldn't make heads or tails of CT Magazine's
report on the web. But we found reviews from IT Week, PC World, and PC
Magazine of the first laptop PC to use Transmeta, which is the Sony Vaio
PictureBook PCG-C1VE (based on a 600 MHz Crusoe), and the initial reviews
are pretty weak.
09:33am EST 30-Oct-00 J.P. Morgan (RAGSDALE, TERRY (1-212) 648-9047) INTC INTC.
INTEL: RECENT MANAGEMENT MEETING --- PART II OF II
October 30, 2000
J.P. MORGAN SECURITIES INC. - EQUITY RESEARCH
TERRY RAGSDALE (1-212) 648-9047 ragsdale_terry@jpmorgan.com
Daan D. Coster (1-212) 648-3806 coster_daan@jpmorgan.com
Intel (BUY)
PART II OF II
RECENT MANAGEMENT MEETING: SOME INTERESTING BITS AND PIECES
Earnings Per Share P/E
INTC 52-Wk ------------------ ------------ MkCap
10/27 Rge 12/99 12/00 12/01 4Q/00 4Q/99 12/00E 12/01E Yld ($MM)
---- ----- ----- ----- ----- ----- ----- ---- ---- ---- -----
$46.38 $76-34 $1.17A $1.69E $1.72E $0.42E $0.50A 36.4 35.6 0.3% $311,362
Note: 12/00 and 12/01 estimates include investment gains of $0.39 and
$0.27 per share, respectively.
IT Week (October 23) says that the Crusoe-based Sony PictureBook's
performance is around half of a 500 MHz Celeron-based system for about the
same price (with no discussion of battery life). PC Magazine (October 18)
says that the Sony "PictureBook is a bit anemic, even when set to maximum
performance. The unit's Business Winstone 99 score was just 9.7, which is
a fraction of other ultraportables." PC Magazine's comparisons were to
another Sony Vaio ultraportable with an Intel Pentium III 600 MHz
(Winstone score 22.9, or more than twice the Crusoe machine's score) and
to a Toshiba Portege ultraportable with the same 600 MHz Intel processor
(Winstone score 30.4, or more than 3x the Crusoe machine's score). PC
Magazine found that the battery life of the Crusoe-based Sony PictureBook
just barely met the minimum of Sony's rated life of 2.5 to 5.5 hours.
Finally, PC World (October 13) compared the new Crusoe-based Sony
PictureBook to another recent Sony PictureBook, one based on an Intel
Pentium II (not III) at 400 MHz. The older Intel-based Sony PictureBook
had half as much main memory (64 megabytes versus 128 megabytes) and a
fraction of the video memory (2.5 megabytes versus 8 megabytes). On the
PC Worldbench 2000 benchmark, the Crusoe-based Sony was 19% slower than
the older Intel-based Sony, in spite of the Intel-based machine's clock
speed (400 MHz versus 600 MHz), main memory, and graphics memory
handicaps. The Crusoe-based Sony did have a longer battery life, at 150
minutes versus 100 minutes.
In short, the tests so far of Transmeta's Crusoe microprocessor by
independent third parties are rather disappointing. There is some
controversy about benchmarks: Transmeta claims that conventional
benchmarks do not mimic real-world PC usage and therefore do not capture
Crusoe's true performance and battery life characteristics as a real end
user would experience them. Specifically, Crusoe has the overhead of
translating Intel code into Crusoe code before running it. But if code is
run repeatedly, it need only be translated once and will run faster
(without further translation overhead) thereafter. Industry authorities
are divided on this issue, but at least a couple seem to agree with
Transmeta. We must admit to being skeptical of any company that claims
its advantages are not captured by conventional benchmarks, but the jury
is out. Mr. Otellini also reminded us that Intel has responded (not his
word) to the Transmeta threat with a more power-optimized version of the
mobile Pentium III. Our guess is that that will be more than sufficient
to put Transmeta to bed. After all, no matter what Transmeta has, the
game has always been more about marketing than about technology/performance.
And reviews that say Transmeta's performance/battery life cannot be proven
with industry standard benchmarks are not going to sell Transmeta-based PCs.
ú Execution issues: Asked about Intel's well-publicized and
formerly rare execution issues of the past couple of years, Mr. Otellini
started by saying that Intel's microprocessor business is "not suffering
from a lack of management attention." Yes, Intel is branching out into
other businesses such as comm ICs, but there's no doubt where the bread
and butter is. He admitted that many of Intel's execution problems have
stemmed from the selection several years ago of Rambus DRAM as the next
DRAM standard to support Intel's microprocessors (delay of the 820
chipset, the MTH recall, and the Timna product cancellation). One could
read into this that Intel plans to extricate itself from Rambus, which is
supported by CEO Craig Barrett's recent public comments that " . we made a
big bet on Rambus, and it did not work out . In retrospect, it was a
mistake to be dependent on a third party for a technology that gates your
performance." Intel has said that it is working on a P4 chipset that will
support SDRAM and has made it clear that the company is carefully
considering DDR support as well. Mr. Otellini said that Intel's recall of
the 1.13 GHz Pentium III resulted from a hole in Intel's testing process,
a hole that has always been there but luckily has not been an issue until
the 1.13 GHz PIII (in other words, not a recent execution problem). He
also admitted that Intel may have stretched too far on PIII to meet the
competitive challenge from AMD and vowed that it won't happen again.
Finally, Mr. Otellini pointed to Intel's recent management shuffling, one
of the most substantial ever for Intel, aimed at bringing under one roof
full responsibility for all aspects of the microprocessor business (e.g.,
microprocessors, chipsets, motherboards, etc.) and under another roof all
manufacturing, implying a clearer chain of command and accountability
going forward. |
