AMD CEO Quits.

[Fingolfin_Noldor]

About bloody time

July 18, 2008
Another Loss at A.M.D. Leads to Chief’s Ouster
By JOHN MARKOFF and LAURIE J. FLYNN

SAN FRANCISCO — After the seventh consecutive quarterly loss at Advanced Micro Devices, the company’s board ousted Hector Ruiz as chief executive and replaced him with Dirk Meyer, a star engineering executive at the company.

A talented chip designer who had led an earlier turnaround of A.M.D.’s technology, Mr. Meyer, 46, had already been designated by Mr. Ruiz as his successor, but the timing of the transition on Thursday came as a surprise to industry executives and analysts.

Mr. Meyer, president and chief operating officer, is widely respected and admired by other A.M.D. technical employees and also has the confidence of Wall Street analysts.

Mr. Ruiz will remain as chairman of the chip maker, based in Sunnyvale, Calif.

“A.M.D. has not been living up to expectations,” said Nathan Brookwood, an analyst at Insight64, a technology consulting firm. “When you’re C.E.O., if there is any kind of accountability, you are on the line.”

Mr. Ruiz was instrumental in helping A.M.D., the perennial No. 2 chip maker, win market share from the leader, Intel, beginning in 2003. The company’s advanced line of Opteron chips offered performance advantages over Intel’s Pentium products, and A.M.D. won business from Hewlett-Packard, Dell and other Intel loyalists.

Intel ran into engineering difficulties and canceled projects that were performing poorly before they got to market during the period.

However, A.M.D.’s gains proved short-lived. Mr. Ruiz then presided over a period in which key products fell behind schedule and A.M.D. lost market share. In 2007, a server chip code-named Barcelona, which had four separate processor cores and beat Intel chips in performance and price, was delayed because of a design flaw.

The company began to recover this year, but continued to lose money. On Thursday, A.M.D. reported a loss of $1.18 billion in the second quarter, which included a loss of $920 million related to a portion of the operations of the graphics chip maker ATI Technologies. A.M.D. bought ATI, which makes chips for hand-held devices and digital TVs, in 2006.

Mr. Ruiz, 62, was born in Piedras Negras, Mexico, a small town across from Eagle Pass, Tex. His parents, who had white-collar jobs, emphasized education and he crossed the border to attend an American high school. He went on to study engineering at the University of Texas and Rice University. He began his career as a researcher at Texas Instruments before starting a 22-year career at Motorola.

Mr. Ruiz joined A.M.D. in 2000 and replaced its charismatic founder, Jerry Saunders, in 2002.

Mr. Brookwood, the analyst, called a decision by Mr. Ruiz to sue Intel for antitrust violations a “gutsy” move. He said there were many investigations into Intel’s business practices around the world. “Certainly there is something there,” he said.

In an interview, Mr. Ruiz said he remained optimistic about A.M.D. despite the earnings decline over the last two years. He hinted that he would focus on the company’s antitrust suit against Intel. “Not only can we right the ship, but Dirk and I believe that we are in the strongest position the company has been in going forward,” Mr. Ruiz said.

A.M.D. has been able to strengthen its management team in recent months, Mr. Meyer said. “We clearly have to execute better and more consistently,” he said in the same interview, noting that Intel’s biggest vulnerability is that the world abhors a monopoly.

Mr. Meyer began his career at Intel, and he also designed advanced microprocessors at the Digital Equipment Corporation. He joined A.M.D. in 1995, and was responsible for the Athlon processor family there, a bold gamble that began an era in which the smaller company proved more innovative and quicker than Intel, the dominant force in the industry.

“This gives Wall Street a chance to take another look,” said Cody Acree, a securities analyst with Stifel Nicolaus. “So many on Wall Street had turned their backs on A.M.D. With a company struggling the way they are, you need an excuse to take another look.”

Mr. Acree said Mr. Meyer had considerable credibility with analysts and investors because, unlike Mr. Ruiz, he was comfortable talking about A.M.D.’s challenges. “He has a more pragmatic rapport with the street,” Mr. Acree said. Mr. Ruiz, on the other hand, preferred to focus on A.M.D.’s strength, appearing sometimes to gloss over its considerable challenges, Mr. Acree said.

Finally, that fool at the top quits. Bad Execution and what not for the last 2 years, I'm surprised he lasted this long.

[phongn]

First he left Motorola SPS in not-so-hot shape (spun off into Freescale), now AMD, eh? He really spent far too much for ATI, and then they couldn't get K9 working ...

[White Haven]

While I'm sure he's stunningly unimpressive, I'm not sure the CEO is to blame for the fact th at AMD's CPUs have been getting stomped for the past two years or so. He may not have been helping things, and may have been causing other problems that the company can't endure at a time like that, but the core problem would still exist.

[phongn]

While I'm sure he's stunningly unimpressive, I'm not sure the CEO is to blame for the fact that AMD's CPUs have been getting stomped for the past two years or so. He may not have been helping things, and may have been causing other problems that the company can't endure at a time like that, but the core problem would still exist.

AMD probably could've recovered from the K9 failure normally and that's not really his fault. But the ATI purchase? The price of acquisition was far too high and they've written off billions from that, money they could ill-afford.

[Fingolfin_Noldor]

I think they should have done a Joint Venture instead of a single monolithic company that would have cost a lot more. What they really need now is to expand their chip design divisions to catch up. Ah well, AMD is a mess now.

[phongn]

I think they should have done a Joint Venture instead of a single monolithic company that would have cost a lot more. What they really need now is to expand their chip design divisions to catch up. Ah well, AMD is a mess now.

Expand their chip design divisions? With what money? AMD's laying off lots of people and while ATI is doing well, it isn't anywhere near recouping the the cost of purchase.

[Fingolfin_Noldor]

I think they should have done a Joint Venture instead of a single monolithic company that would have cost a lot more. What they really need now is to expand their chip design divisions to catch up. Ah well, AMD is a mess now.

Expand their chip design divisions? With what money? AMD's laying off lots of people and while ATI is doing well, it isn't anywhere near recouping the the cost of purchase.

Well, that is a problem they need to solve somehow. I guess what Dirk Meyers needs to do now is refocus AMD properly. Sigh...

[Arthur_Tuxedo]

I don't know if you guys have noticed, but right now AMD is beating Intel and NVidia in bang for buck across almost every market segment of CPU's and GPU's. Now matter how temporary that situation is, only a fool would count AMD out for good.

[Uraniun235]

I don't know if you guys have noticed, but right now AMD is beating Intel and NVidia in bang for buck across almost every market segment of CPU's and GPU's. Now matter how temporary that situation is, only a fool would count AMD out for good.

The GPU side I'll accept, but are there any reviews which indicate that AMD currently has superior performance/price numbers for most of their CPUs?

[White Haven]

AMD's only staying in shouting distance by shorting their brackets, so it's the AMD mid against the Intel low, and the AMD high against the Intel mid. So in the mid and low end spaces, they stay competitive but with lower profit margins, and in the high end space they simply lose.

[Arthur_Tuxedo]

That's true, but the fact remains that right now a 2.6 GHz Phenom is cheaper and uses less power than a 2.4 GHz Q6600, and a 2.5 GHz Athlon X2 BE has the same advantages over the E2220. Overclockers will still want the Intel chips, but for everyone else, AMD is currently the way to go. I can't imagine who would buy $500+ high-end CPU's, although there must be some small market for them or they'd stop making them.

On the GPU side, AMD's 4800 series simply trounces nVidia's offerings for price / performance, and the shorter profit margin argument is even more applicable to nVidia right now than it is to AMD vs Intel.

[White Haven]

So AMD has some better chips than an old quad core and a Celeron equivalent. That's...not a ringing endorsement. As for their video card division, no argument there.

[phongn]

I don't know if you guys have noticed, but right now AMD is beating Intel and NVidia in bang for buck across almost every market segment of CPU's and GPU's. Now matter how temporary that situation is, only a fool would count AMD out for good.

AMD can't afford to undercut Intel continuously. Even accounting for the ATI writedown, they've still lost over $300M in Q1! Bang for buck is all well and good, but if they're not making money they're in deep trouble.

That's true, but the fact remains that right now a 2.6 GHz Phenom is cheaper and uses less power than a 2.4 GHz Q6600, and a 2.5 GHz Athlon X2 BE has the same advantages over the E2220. Overclockers will still want the Intel chips, but for everyone else, AMD is currently the way to go. I can't imagine who would buy $500+ high-end CPU's, although there must be some small market for them or they'd stop making them.

Uh, no. You're wrong, totally wrong. AMD currently can try to win the low end (where they've traditionally done well) and they have a stranglehold in four-socket configurations thanks to their integrated memory controller. But everywhere else they've taken a severe beating. Single and dual-socket configurations remain strong for Intel and they all but own the fast-growing mobile market.

And as for mainstream CPUs? Nothing AMD has is competitive with Penryn, and a 3GHz E8400 is only $183 retail. Plus, Intel, in general, has superior chipsets compared to AMD's offerings.

On the GPU side, AMD's 4800 series simply trounces nVidia's offerings for price / performance, and the shorter profit margin argument is even more applicable to nVidia right now than it is to AMD vs Intel.

That may be so, but nVidia is making money right now. Q1 profit was $90M.

[Arthur_Tuxedo]

I don't know if you guys have noticed, but right now AMD is beating Intel and NVidia in bang for buck across almost every market segment of CPU's and GPU's. Now matter how temporary that situation is, only a fool would count AMD out for good.

AMD can't afford to undercut Intel continuously. Even accounting for the ATI writedown, they've still lost over $300M in Q1! Bang for buck is all well and good, but if they're not making money they're in deep trouble.

Didn't say they could afford to continuously undercut them. Just that they're competitive right now even though no one seems to have noticed, and only a fool would count them out for the long haul.

That's true, but the fact remains that right now a 2.6 GHz Phenom is cheaper and uses less power than a 2.4 GHz Q6600, and a 2.5 GHz Athlon X2 BE has the same advantages over the E2220. Overclockers will still want the Intel chips, but for everyone else, AMD is currently the way to go. I can't imagine who would buy $500+ high-end CPU's, although there must be some small market for them or they'd stop making them.

Uh, no. You're wrong, totally wrong. AMD currently can try to win the low end (where they've traditionally done well) and they have a stranglehold in four-socket configurations thanks to their integrated memory controller. But everywhere else they've taken a severe beating. Single and dual-socket configurations remain strong for Intel and they all but own the fast-growing mobile market.

And as for mainstream CPUs? Nothing AMD has is competitive with Penryn, and a 3GHz E8400 is only $183 retail. Plus, Intel, in general, has superior chipsets compared to AMD's offerings.

For mainstream dual-core Intel is king, but for mainstream quad-core AMD has the better offering for the money. I admit to overlooking the mainstream dual-core segment, but you can hardly go so far as to say they've only got the low-end and four-socket when the higher-clocked Phenoms so clearly trounce the Q6600 and Q9300 in price / performance.

On the GPU side, AMD's 4800 series simply trounces nVidia's offerings for price / performance, and the shorter profit margin argument is even more applicable to nVidia right now than it is to AMD vs Intel.

That may be so, but nVidia is making money right now. Q1 profit was $90M.

In Q1 nVidia was spanking ATI's chips across the spectrum. The situation going forward is quite different.

[phongn]

Didn't say they could afford to continuously undercut them. Just that they're competitive right now even though no one seems to have noticed, and only a fool would count them out for the long haul.

AMD's CPU group is only competitive because they're pricing things so low, which apparently I was unclear about. The iron laws of supply and demand are killing them.

For mainstream dual-core Intel is king, but for mainstream quad-core AMD has the better offering for the money. I admit to overlooking the mainstream dual-core segment, but you can hardly go so far as to say they've only got the low-end and four-socket when the higher-clocked Phenoms so clearly trounce the Q6600 and Q9300 in price / performance.

Trounce? A Phenom 9950 (US $235) has performance parity with cheaper Q6600 (US $210).

In Q1 nVidia was spanking ATI's chips across the spectrum. The situation going forward is quite different.

Not to the extent of AMD vs. Intel.

[The Kernel]

Overclocker's Ed Stroglio had a nice analysis of this event and AMD's overall problems since their inception.

Overclockers

The plug got pulled on Hector Ruiz.

Why? It's not too far-fetched to say he "died" not only for his own sins, but Jerry Sanders' sins, too. You can't fairly judge Ruiz unless you understand what he was handed and what he had to do with it.

What am I talking about? AMD has one huge flaw, always has: lack of money compared to Intel, and that's primarily because AMD is a victim of (secular) original sin.

Why is that? Just like Adam, Jerry Sanders made a big mistake at the creation when he decided to get into the CPU business. Rather than compete with Intel toe-to-toe from the dawn of the PC era, he was happy to reverse-engineer and make easy money cloning Intel chips.

It certainly saved AMD a lot of early R&D money, probably made Jerry Sanders a lot more money, and if AMD had gone toe-to-toe from the start, they may not have made it out of the Eighties.

However, the cloning strategy also doomed AMD to be a trailing also-ran, and permanently forfeited CPU leadership to Intel.

By the time the AMD Cloning Era ended, AMD was a financial midget compared to Intel. In an industry where Money Mattered, it was hopelessly weaker than Intel, with no prospect of overcoming or even getting close to Intel's ability to invest in R&D and especially production capacity.

It isn't unfair to say that AMD's efforts since the end of the cloning era have been mostly dedicated to undoing that grave initial error.

Lack of money is as ingrained a part of AMD corporate culture as the color green. It is the fundamental bottleneck of the company, and plays a large if unacknowledged part of its decision-making process. Because they can't throw money at situations or problems like Intel, they end up having to do things on the cheap, cut corners, gamble, pick their spots.

Yes, it was like David and Goliath, except this David couldn't afford a lethal sling. No matter what AMD did, no matter how much better an AMD CPU was than an Intel chip, Goliath had a guaranteed 75%-80% share of the market. Unlike the Biblical David, AMD had no chance of taking out Goliath with one shot. The only way this David could beat this Goliath, or even level the playing field, was to gnaw him down to size, and that would have taken five or more years of AMD superiority.

Still, during the last half of the nineties and the first few years of the new century, David was capable enough to at least stay on the same battlefield as Goliath, and even get a few kicks and punches in, especially when Goliath got drunk on arrogance and lay around in a stupor after running a CPU design into the ground.

This was all good fun, but even a befuddled Goliath could still use his bulk and just roll over on David and half-smother him to death when he started getting dangerous, rules or no rules.

Even worse, this David and Goliath competition was also an arms race, and the cost of the latest weapons skyrocketed. Looking down the road, it didn't take a rocket scientist to see that eventually, no matter how nimble David was, Goliath would win so long as David stayed so small.

David had to bulk up, and that was the imperative when Jerry Sanders handed over the reins to Hector. AMD had determined that if they didn't get a lot bigger, Goliath would eventually get them. And they were not wrong to think that.

But . . . they had no money in an industry that required it. What could Hector do?

Jerry Sanders certainly had more than a little of the Wizard of Oz about him; the problem was everyone knew that and took him with some blocks of salt.

Hector at first seemed to be a different story; he seemed to be a very sober, business-like type, quite unlike the personally flamboyant Sanders.

But just like the best spies are the last people anyone around them suspect would be spies; Hector was a stealth Wizard of Oz. The Ruiz strategy was simple, "If you don't have money, act like you do. Fake it 'til you make it." For a while, it worked.

Sanders had left Ruiz with a risky but rewarding CPU design (Hammer), and a new circuit design approach (SOI) which proved very troublesome but gave AMD some cooling relief while Intel chips were having heat stroke.

The combination of AMD achievements and Intel ineptitude gave AMD the better chips for a while, but still, again, no matter how much better the AMD chips were, Intel would still have 75% of the market since AMD could only make 25% of the chips.

However, this situation (along with lots of very easily obtained money) let AMD borrow its way into acting big. Instead of one fab, we'll have two, two plus. Don't worry Dell, we can supply you. Intel, it's time to be sued.

The momentum strategy was working quite well, but then two bad things happened.

First, Hector and Company were so busy chasing big business (often to the cost of small businesses) that they forgot why they were getting the business to begin with. They either didn't or couldn't keep up the pace of innovations that had put them ahead. In contrast, after running the PIII and PIV designs to destruction largely because they didn't want to do the hard job of fixing ever-more-leaky transistors, Intel bit the bullet and did the expensive work they had avoided so long to make CPUs far more efficient and less leaky than the later Pentiums.

Even more importantly, AMD didn't get while the getting was good. They relied on borrowed money rather than raise equity when their stock price was high. AMD could have easily raised $8-10 billion at that time. Sure, it would have diluted the stock, ticked off the then-current shareholders and maybe most importantly, hurt the value of the execs' stock options), but it would have made AMD financially strong and easily capable of building fabs and fending off any Intel price wars with a smile.

No, instead of shoring up AMD's finances when they had the chance, Ruiz and Company did the exact opposite and greatly weakened them by buying ATI. Many will say this was Ruiz's fatal error, and indeed, it was a huge blunder to pay over $5 billion dollars for a company on the verge of a long losing streak that probably could have been bought for $2 billion or less a year later.

But the bigger catastrophe wasn't paying too much for ATI, bad enough as that was. That was paying for it by emptying the piggy bank and filling it with IOUs rather than paying for it out of a stock issue. This was the financial equivalent of AMD putting a sign on its butt and saying to Intel, "Kick me." And Intel did, with price wars and C2Ds.

That was the tragedy. Anything after that was farce. The technical failures surrounding K10 would have been a disaster under the best of hands, but they were compounded by the spin strategy from hell. While AMD has always been an overly secretive company prone to hiding its weaknesses through deception, all previous records were shattered in 2007. Oz couldn't turn himself off.

The irony of all this is that while we've given two very good reasons why Ruiz deserved to be fired, that probably wasn't the reason why he was fired. The man literally talked himself out of his job, but that kind of talk was the reason why he initially got the job.

But it should never be forgotten, despite all his errors, that he did what he did to undo somebody else's colossal mistake made decades earlier.

The Last Part of the Legacy

Hector isn't quite dead nor even gone yet. He's got one last card to play to atone for sins.

Forget leading whatever "Asset Smart" strategy there is; that's just financial finagling. Hector's last chance to retrieve his reputation is to beat Intel in the courts and among the regulators.

Dirk Meyer cannot make AMD healthy. The most he can do is to retrench, stabilize the ship, show enough improvement to somehow get the extra money for enough further fabbing to stay in the game, and hope to fight another day, another year.

There is no quick fix for AMD. The only "quick fix" is to sell the company to someone with serious money and a willingness to spend it, yet still be able to make x86 processors under the AMD-Intel cross-licensing agreement.

Short of a sugar daddy, though, all by itself, AMD at best will resume a hand-to-mouth existence for a few more years, and more likely will face a slow, lingering drift into niche irrelevancy.

Unless, of course, GOD intervenes and cuts Goliath down to size, GOD as in "Governments Overriding Destiny."

As we've said before, Intel may well be fined for past behavior, may well deserve to be fined, but fines aren't going to make AMD healthy, nor will the proceeds from lawsuits.

No, what will matter is the degree (if any) to which governments will decide to intervene in the CPU business, either directly or indirectly, which really means to what degree will governments decide to throttle Intel.

There are so many possible variations on this theme that there's no point even starting to list them, dozens of possibilities and variations of those possibilities arising from multiple places, all interacting with each other and almost all occurring behind closed doors.

What is clear, though, is that AMD is going to have to be rescued/bailed out, in some way, shape or form by outside parties, or it will just shrivel up.

In some ways, AMD has become a litigation machine that makes computer equipment on the side. It isn't a unique situation, you could have called MCI that before it won its lawsuit against AT&T.

That's Hector's last chance, and if he leads AMD to victory in that war, all his mistakes will be forgotten.

But if he doesn't, he'll probably be remembered as the man who killed AMD.

[The Kernel]

That's true, but the fact remains that right now a 2.6 GHz Phenom is cheaper and uses less power than a 2.4 GHz Q6600,

The hell it does.



That's a comparison between the lower clocked 9850 and the Q6600 from ArsTechnica. The Q6600 uses 100 LESS WATTS at load then the 9850.

As for the price, the Pheom 9950 is currently selling for $235 compared to below $200 for the Q6600.

[Arthur_Tuxedo]

That's true, but the fact remains that right now a 2.6 GHz Phenom is cheaper and uses less power than a 2.4 GHz Q6600,

The hell it does.



That's a comparison between the lower clocked 9850 and the Q6600 from ArsTechnica. The Q6600 uses 100 LESS WATTS at load then the 9850.

As for the price, the Pheom 9950 is currently selling for $235 compared to below $200 for the Q6600.

Conceeded. Checked Newegg again and spotted my error. I was remembering the price of the Q9300 rather than the 6600 and the 95W TDP figure of the lower clocked Phenoms.

Anyway, I have no trouble swallowing the idea that the Quad-Core Intels might be more efficient than the AMD's, but those figures are pretty hard to believe. Is everything else being held equal in those rigs, because I don't see how load could be a hundred watts higher when even the Prescotts would barely scratch a hundred watts period under load.

[The Kernel]

Conceeded. Checked Newegg again and spotted my error. I was remembering the price of the Q9300 rather than the 6600 and the 95W TDP figure of the lower clocked Phenoms.

Anyway, I have no trouble swallowing the idea that the Quad-Core Intels might be more efficient than the AMD's, but those figures are pretty hard to believe. Is everything else being held equal in those rigs, because I don't see how load could be a hundred watts higher when even the Prescotts would barely scratch a hundred watts period under load.

TDP ratings are meaningless across different vendors. AMD uses a very weird system for TDP ratings that doesn't correlate much to actual power draw. And Intel's are ridiculously conservative; they claim a 65W TDP max on processors 45nm dual core procs that barely draw 20W under load.

And yes, the methodology on these tests is sound. The problem AMD has is that their chips are too big, too hot and 12-18 months process wise behind Intel.

[Arthur_Tuxedo]

Interesting. I had wondered why there was no change in TDP when Intel jumped from 65 to 45 nm. Thanks for the info.