AMD doubters persist; Are they right this time?

AMD doubters persist; Are they right this time?

Summary: With AMD's launch of its latest greatest Opteron server chip--formerly codenamed Barcelona--one of the biggest questions is whether lightning can strike twice.The first lightning strike came when AMD (all resources) launched the Athlon line and became a legitimate desktop player.

TOPICS: Processors, Intel

With AMD's launch of its latest greatest Opteron server chip--formerly codenamed Barcelona--one of the biggest questions is whether lightning can strike twice.

The first lightning strike came when AMD (all resources) launched the Athlon line and became a legitimate desktop player. Using that momentum AMD leapfrogged its larger rival on the server side, grabbed market share, found an energy consumption pitch and became quite the contender. For much of this decade, AMD has enjoyed a renaissance.

At first, few folks gave AMD much of a chance at the time. But AMD surged, Intel took its eye of the ball and customers basked in the glow--not to mention prices--of two real options for processors.

Now the debate is whether AMD's latest quad-core architecture can bloody Intel's nose again. Many of us--David Berlind cops to it and I was off too--were wrong about AMD's prospects a few years ago. We learned that you can't call an Intel victory too early. Another key lesson: It's not a zero sum game. AMD doesn't have to beat Intel to have a thriving business. Just keeping Intel honest can keep many customers happy.

But the arguments in Intel's favor are still there. Intel has a broad technology portfolio, scale and the financial heft to squeeze AMD's balance sheet with pricing pressure. Intel is clearly Goliath to AMD's David.

What remains to be seen is whether David could beat Goliath in a seven-game series.

I'm not going to claim to have the answers, but others are going to be making their forecasts. It's clear that AMD still has its doubters. Time will tell if the doubters are correct.

First up, Merrill Lynch analyst Srini Pajjuri. In a research note on Friday, he downgraded Intel to a neutral rating based on the theory that all of the good news about Intel is already baked into the stock price.

His commentary, however, about AMD was interesting.

"Delayed launch notwithstanding, Barcelona should help AMD gain some lost ground in servers. However, we are not expecting a repeat of 2005. Intel has a competitive server part and the current version of Barcelona is unlikely to create a meaningful performance gap with Intel."

Simply put, a tie isn't good enough for AMD.

UBS analyst Uche Orji has a similar take:

"With AMD having launched its quad-core Barcelona server processor, we have reassessed our view of AMD’s market position and valuation. While we learned that Barcelona is likely to reach speeds in the range of 2.6 GHz in 1Q08, falling short of our expectations of 3 GHz, the benchmark data AMD presented at the launch event showed that Barcelona’s performance will likely be competitive with Intel’s server chips, at least on certain benchmarks. This is in line with our view that AMD will be able to stabilize unit market share and grow average selling prices (ASPs), which in turn should help lift AMD to profitability by 4Q08. But other concerns remain. Holding us back from a more constructive stance on the stock is our ongoing view that AMD needs to implement a more substantial restructuring to lower expenses and reduce the overall complexity of its business. The longer AMD defers its primary focus to be on profitability rather than market share, the more uncertain is AMD’s future. On top of this are recent credit market uncertainty and management departures."

Orji's last point on the credit markets shouldn't be taken lightly. It's tough to get credit right now and AMD has $3.9 billion in net debt. Orji estimates that it may be tough for AMD to get more capital without significantly lowering expenses.

But if AMD executes that financial outlook changes.

Overall, AMD can't afford many missteps even after the launch of Barcelona. The consensus view seems to be that AMD has to be perfect. My plan is to follow the numbers on AMD when it comes to calling a winner in this battle. My hunch is that AMD has taken Intel's best punch, but future quarters will tell the tale.

Topics: Processors, Intel

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Talking Intel/AMD victory...

    is so subjective. If you look at Intel's desire - to be the only chip maker - AMD is victorious.

    If you look at solely IPS's then it's a toss up between INT and FP.

    I sell nothing but AMD. So in my book they are a winner no matter what.

    Let's wait and see what the market does... and says.

    AMD isn't going anywhere like LOVErock would have you believe. Are AMD fanboys disappointed? A little. Are they devastated? No way.

    All your major vendors are still offering AMD boxes for a reason. It is a viable alternative.
  • AMD Has always been a contender

    I've used their 486's, K5, K6, K6-2, K6-3, Duron, Athlon, etc. All have been exemplary performers and extremely reliable.

    AMD has been a player ever since they fab'ed Intel's chips back in the day.
  • "A tie isn't good enough"?

    The comment "a tie isn't good enough for AMD" makes no sense. The economics of a duopoly are problematic, and the economic problems with duopoly often drive it towards monopoly. The question is, given similar performance and similar pricing, will that price provide AMD the margins to continue in the particular business? Unless someone can show me the economic analysis to prove this out, this is just sensationalism. After all, even at Opteron's peak vs. Intel's Netburst Xeons, AMD's server market share never exceeded 25%.

    AMD did a horrible job at marketing its dual-core Opterons after Intel released Woodcrest. While Intel touted SPECint and SPECjbb, AMD did not stress its lab-proven performance advantage when running consolidated workloads on VMware.

    Intel's "quad core" Clovertown certainly had a price/performance advantage running Microsoft software and open-source software, but it was a price/performance disadvantage running IBM, BEA, and Oracle software. These apps remained mostly deployed on dual-core machines, and AMD had some advantages, especially with I/O.

    I think Opteron was hurt by Clovertown as it impacted the four-socket market, which was Opteron's sweet spot. I don't think a lot of customers bought Clovertowns instead of 4S Opteron boxes (Clovertown machines have memory capacity, memory bandwidth, and I/O limitations, making them poor for 4S workloads like VMware and Oracle), but I think a lot of customers instead waited for the quad-core market to settle out, to where a 2S quad-core system could replace a 4S workload.

    That all said, AnandTech's tests show a 2.0 GHz Barcelona to be on par to slightly better than the 2.33 GHz Clovertown. The Barcelona's cores have been improved to give better performance on databases and Java. The FP units are significantly better (AnandTech shows a 142% improvement, clock for clock), allowing Barcelona to give exceptional performance without recompiling to specific Intel SSE features. This should significantly help existing software. It is also the reason AMD had to slip Barcelona's launch by a month. The first tens of thousands of Barcelonas have already been snapped up by the likes of TACC and other HPC customers.

    Barcelona shellacs Clovertown on memory bandwidth and latency. This will certainly sell well with HPC and database customers.

    AMD still has a significant power consumption advantage over Intel, and will continue to for some time. Both Clovertown and Tigerton use FB-DIMM memory, and Tigerton requires an additional 64MB snoop filter adding to the power requirements.

    Yes Intel has faster Clovertowns at 2.66 and 2.93/3.0 GHz, but AMD will have a slightly faster Barcelona to match the 2.66 GHz Intel parts, and the 3.0 GHz parts are very expensive, and probably will not be the model most customers choose. Yes, Intel has Penryn coming. Penryn's primary changes faster clock rates, improved SSE units and more cache. But AMD has Shanghai coming which will increase its cache and speeds. Finally, AMD could Clovertown Penryn with its Montreal "8-core" processor. The real challenges I think come in the 2009 timeframe, when Intel's Nehalem Xeon faces AMD's Sandtiger.

    Of course, all of this is irrelevant if AMD continues to fail to market its products in an effective manner.

    But back to my original comment. I fail to understand why so many in the tech industry believe it has to be a monopoly vendor market. I notice this tends to come from the desktop PC-centric crowd. They remember OS/2 vs. Windows, Novell Netware vs. NT, Red Hat vs. SuSE, Alpha vs. Intel, Dell vs. Gateway etc. They fail to recall Compaq vs. IBM, Seagate vs. Western Digital, HP-UX vs. Solaris, etc.

    Of course, in those markets where monopolies prevail, these same people who cheer them on complain incessantly about the lack of features, large number of bugs, and lack of competition in these markets.

    If HP, IBM, and Sun can each have about 30% of the UNIX market (and three different processor architectures), and if HP, IBM, and Dell can all have more than 20% of the much larger x86 server market, there is certainly room for more than one x86 processor.
    • It makes PERFECT sense

      [i]The comment "a tie isn't good enough for AMD" makes no sense.[/i]

      Considering that from their strategic and cash position, AMD needs [i]sure-fire winners[/i] that puts them in a favorable light when compared to Intel. Intel's too big and has pockets far too deep for AMD to just make product that's simply good enough to compete. Intel's dominating position over the past year or two can easily sway prospective clients away from AMD, who had little to nothing to counter the seemingly endless performance juggernaut that Intel played in response to the Athlon. AMD [i]must[/i] convince all the various computer and server makers that they are indeed back in the game. Indeed, it has to be more than just merely getting into the game again because--in the back of everyone's mind--they all know Intel is such a stout, gargantuan behemoth currently flush with solid product that choosing them over AMD is a VERY safe choice.

      Without AMD armed with that giant-slayer of a bat, whey would makers feel any confidence whatsoever that Intel won't simply walk over AMD again when they finally got their act together? They've seen what happens when the sleeping giant awoke the last time, and it hasn't been all that long ago that Intel will be likely to forget the brutal lesson that they learned.
  • RE: AMD doubters persist; Are they right this time?

    AMD's tru Quad-core that can be made a Tri core to sell at a price/performance point that is mid way the Dual/Quad core price point. Get it!! Having a TRU quad gives you options the other guys don't have, & can't talk about. Yet I'd jump all over a Tri for my customers that don't have the $$ to afford a Quad. Yet wanted a performance step up from Dual core. When they want the latest iteration of whatever AMD had to offer, plug it right in and buy back the Tri. Customer saves money. I have a piece the next bloke that wants a box but doesn't have the $$ for a quad. And the world goes around. I say yes, AMD made mistakes. BUT they have in the end their technology act together. Just watch em in the next 6 months. As Intel shrinks old tech. And AMD births NEW Tech. I'm lovin it!
  • Lighting Strikes Thrice

    Just one thing: you unreasonably conflated the first and second lightning strikes... The original K7 Athlon was the first strike-- pulled AMD even with and ahead of Pentium 4 for a while (though of course they stumbled eventually, this made Intel move a lot faster than they had planned to on P4, which cut into their margins, as they weren't able to milk a successful product for years as they had done before).

    Then the A64/Opteron K8 should count as the second lightning strike-- broke open the server market, effectively killed off Itanium, got AMD its relationships with Dell, Sun, and IBM, put a stake through the heart of Netburst, and took the industry towards multicore.

    It's telling that Intel had the third product line available to counter AMD (Pentium-M->Core Duo->Core 2 Duo->Core 2 Quad)-- this reflects their greater resources as a company. But if you look at things now, Intel is basically back to having one viable line of CPU designs, where without AMD they would have had at least three they were positioning differently and making a lot more money on.

    AMD has actually put them into a relatively tight spot. If they can make Barcelona and its various 2, 3 and 8 core derivatives into a viable offering against Intel's server and mobile products, they can make quite a bit of money and continue to put pressure on Intel. In spite of their inferior products, they've gained unit share recently, so the hardware makers aren't interested in going back to an Intel-only world any time soon. So I think lightning could well strike the third time, at least in the sense of AMD winding up in a stronger market position at the best days of the Barcelona generation than they ever were in the K8 generation.

    Every time AMD rises and Intel knocks them back, Intel doesn't knock AMD back quite as far as the last time. AMD comes back stronger after each round. If that trend continues, Intel will lose its ability to control AMD altogether.
    • AMEN !

      lots of truth in that