Intel's Less-Discussed Problems May Be Even More Severe
By Jim Seymour
Special to TheStreet.com
3/1/01 10:02 AM ET
"You can never save your way out of a recession. You can only spend your way out."
Quick test: The man who said that was: (A) Your choice of any Democratic president since 1940; or (B) Intel (INTC:Nasdaq - news - boards) CEO Craig Barrett.
Related Stories
You will be excused if you chose A. Because though you were probably right, part of your choice was no doubt based on the utter improbability of B, of a modern-day American CEO saying anything like that.
And yet Barrett said something very much like that at the Intel Developers Forum in San Jose, Calif., this week. In fairness, he didn't say, "You can only spend your way out," but rather, "The only way to get out of a recession stronger than you went into it is to have great new products."
And thus, he said, Intel is going to spend $12 billion on R&D and new production facilities this year.
Intel has long used its Developers Forum to speak as much to the Street as to PC developers. And much of the rest of what went down this week in San Jose was similarly aimed at Wall Street, not Tech Street. But Intel, trying to fight its way back from product and production bobbles, mismanaged new-product rollouts, product delays, product cancellations -- and a stock price that has fallen by more than half since September -- is getting harder and harder to figure, notwithstanding such "guidance" from management.
Some of the challenges Intel faces are well-known:
More aggressive competition from a newly revitalized Advanced Micro Devices (AMD:NYSE - news - boards).
Smaller performance improvements than expected from faster (and expensive) new Intel chips.
An overall slowdown in PC sales.
Production problems and resulting delayed shipments.
Even with those, Intel has managed occasional little micro-rallies, of the sort we saw Monday and Tuesday. No doubt this is due in part to misguided press reports -- my favorite headline so far this week was "Intel Rises on New Chip," though the demonstration Intel gave at the forum of a PC running a prototype of its new McKinley chip was hardly enough to move reporters' pencils, let alone the market.
These little market nudges are hard to understand, given the totality of what's going on in Santa Clara and elsewhere in the Intel world. Let's look at some of these issues more deeply:
The take rate on the new Pentium 4 is a disappointment. And that's putting it mildly. Intel soldiers on, saying the Pentium 4 is meeting or beating internal estimates. Sure, Craig. But in the free world, most observers wonder why Pentium 4 hasn't been able to grab a bigger slice of high-end PC sales, and why it hasn't been able even to shoulder aside the aging Pentium III.
AMD's Athlon CPUs rang Pentium 4's bell in test after test, and buyers say they don't see anything like the improvement in performance they expected in the new, still-slow-selling P4 boxes. Indeed, at the same clock speeds, the Pentium 4 is often slower than the Pentium III. Go figure; buyers certainly are.
The 64-bit Itanium chip, which really does -- or at least, did -- have the potential to change the world of computing, is late, very late. Indeed, the attention shown this week to the demo of the McKinley prototype at the Developers Forum (after a no-show at the International Solid-State Circuits Conference in San Francisco two weeks ago) illustrates the Itanium problem.
McKinley is the 1.2 gigahertz follow-on to the initial 800 megahertz Itanium. Developers are so frustrated by Intel's delays with the Itanium, upon which work began seven long years ago, that they have already shifted their focus to its successor.
Compaq (CPQ:NYSE - news - boards), an important partner in rolling out Itanium servers -- essential to Intel's plan to take on high-end RISC (reduced-instruction-set computing) CPUs from Sun Microsystems (SUNW:Nasdaq - news - boards), IBM (IBM:NYSE - news - boards) and Hewlett-Packard (HWP:NYSE - news - boards) -- is frothing at the mouth at the delays. And yet, Intel isn't just unable to deliver production quantities of Itaniums to its original equipment manufacturers -- it hasn't even released the chip design to production yet.
The focus of the PC marketplace has shifted dramatically, to midspeed value-priced boxes, such as those powered by AMD's excellent and cheap Durons, and Intel's OK but not-so-cheap Celerons and Pentium IIIs. Yet Intel continues to focus on faster and more powerful chips, where the only improvements most of us can see are in things like processing streaming-video clips a little more smoothly. No chipmaker should ever back away from constantly improving performance, but sometimes you've gotta acknowledge that you're still zigging while the market has zagged.
Intel's cutting back in many areas in R&D, while proclaiming a spend-our-way-out-of-it policy. I don't get that. For example, Intel quietly ordered its contractor to just finish up the bones of its new midtown R&D center in Austin, Texas, then stop. Yet its work in Austin is key to future products. (And leaving an empty, wall-less multistory hulk in the middle of downtown Austin is unlikely to help the company's rep there.)
AMD's new Lightning Data Transport, or LDT, bus is catching on fast. Hot graphics-chipmaker nVidia (NVDA:Nasdaq - news - boards) is one of the latest and most important LDT licensees. This is a head-on run at Intel's new inter-chip communications bus design for the Pentium 4 ... and as my PC Magazine colleague John Dvorak has pointed out, nVidia's clout has made it perhaps "the most dangerous chip designer in Silicon Valley."
Intel is still sticking with Rambus memory ... it says. And repeated that this week at the Developer Forum. Yet Intel also says it knows expensive Rambus (RMBS:Nasdaq - news - boards) RDRAM memory is the high-priced spread, and thus cannot push Pentium 4 machines into the price-sensitive PC mainstream. So it's rolling out later this year a special chipset for Pentium 4 machines that will allow them to use slightly slower but much less expensive, standard synchronous-DRAM, or SDRAM. And then a second Pentium 4-supporting chipset early next year, which will allow use of the even faster (but still cheaper than Rambus) double-data-rate DDR-SDRAM memory.
Big question: By then, will it matter? That's an eternity for a chip design like Pentium 4 to flop around in the market without a clear shot at taking over.
I like Intel. I took a lot of (deserved) flak here a few months ago for overestimating its resiliency and ability to hold on to a stock price twice today's close. I was wrong.
I find it hard to see Intel turning around anytime this year. Too many obstacles. You can't spend your way out of every problem, Craig.
--------------------------------------------------------------------------------
Jim Seymour is president of Seymour Group, an information-strategies consulting firm working with corporate clients in the U.S., Europe and Asia, and a longtime columnist for PC Magazine. Under no circumstances does the information in this column represent a recommendation to buy or sell stocks. At time of publication, Seymour had no positions in the stocks mentioned in this column, although positions can change at any time. Seymour does not write about companies that are, or have been recently, consulting clients of Seymour Group. While Seymour cannot provide investment advice or recommendations, he invites you to send your feedback to Jim Seymour .
--------------------------------------------------------------------------------
Send letters to the editor to letters@realmoney.com.
Read our conflicts and disclosure policy.
Order reprints of RealMoney.com articles. Top |
