A Terrible Quarter, Part 2 . . .

As we pointed out yesterday, AMD had a terrible first quarter.

Why should anybody besides investors care?

Under normal circumstances, the average person reading this isn’t obviously affected. It doesn’t matter to you if Intel makes one as opposed to two billion dollars profit a quarter. If anything, you probably would prefer the lower number because it was likely caused by lower CPU prices.

Indeed, if AMD hadn’t largely emptied its piggy bank and took on a lot more debt to buy ATI, this terrible quarter wouldn’t matter all that much to non-investors, either.

It’s because AMD did empty the piggy bank and took on a lot more debt that this is a big deal even for those who couldn’t tell the difference between Wall St. and a brick wall.

It’s a big deal because AMD doesn’t have the money to run its businesses at current losses and also invest in the new/renovated fabs needed to stay in the race against Intel. It needs more outside money, either in the form of loans or new stock.

And frankly, if the AMD execs had decided, “Let’s make our company as financially unattractive and unlikely to get that extra financing as possible,” they couldn’t have done a better job than they did.

It’s like owning a baseball team that lost the last twenty games of the season, then announcing a big increase in ticket prices. The team may indeed desperately need more money, but it’s hardly the best time to ask disgusted fans for it.

For instance, in 2007, AMD’s two top priorities is to switch 65nm production to the next-generation processors, and start converting good old Fab30 from 90nm to 65/45nm production. If you don’t do the first, AMD falls even further behind Intel. If you don’t do the second, you have half of your manufacturing capacity stuck making increasingly obsolete and increasingly unprofitable 90nm CPUs.

This is going to take a few billion dollars. Right now, AMD probably has one billion in cash, and no prospect of making anything like the profits necessary to finance these short-term necessities anytime this year.

With the warning, AMD said it would cut its capital expenditures (i.e. the money to pay for these improvements) by a half-billion dollars, or 20%.

Even before these financial reverses, AMD warned that the rampup of its nextgen chips would be pretty slow, and cutting investment isn’t going to make that faster.

However, if AMD puts new-stuff-rampup first (which they will), that means cutting back even more on converting Fab30 to a modern facility will occur rather more slowly than expected. This means

Yes, even under the old plans, renovating Fab30 was going to mean lower capacity, which is why AMD brought Chartered into the picture.

However, Chartered just started making 65nm chips, and can’t be expected to ramp up 65nm or especially next-gen chips any faster than AMD. Indeed, it will probably be a good deal slower. Up to now, Chartered has been a trailing-edge second-source; AMD gets the process right at Dresden, then transfers the know-how over.

Also keep in mind that Chartered isn’t doing this for free. Chartered certainly is making a profit from this (they have to pay the fab bills, too), which means Chartered chips certainly cost AMD more than those made in Dresden. And if AMD is losing money on Dresden chips, they’ll lose even more on Chartered chips.

Due to all these financial woes, it’s not inconceivable that AMD could go belly up, and rather more likely that their future the next couple years will be somewhat to heavily crippled due to lack of money for both R&D and fabbing.

And that will affect the person buying a CPU, both in what he or she gets and what price he or she pays for it.

A Second-Half Savior?

The optimists are saying, “Just you wait. Our Barcelonas, Kumas and Agenas will fix everything.” Well, playing secrets might just be an annoyance when the company is doing OK financially, but when you’re burning money like a dotcom and you’ll need to raise a billion or two from people who think your stock is a POS, secrecy begins to verge on the criminal.

All we know now about these processors is that the quad-core Barcelona are supposed to do somewhat over 10% better than a Xeon X5355 in a spec_int benchmark, and 42% better in a spec_fp benchmark.

AMD didn’t even indicate whether the SPEC tests used were CPU2000 or CPU2006, nor submitted a score. (Perhaps conveniently for AMD, SPEC2000 scores are no longer submittable.)

Well, as ZdNet pointed out, they did, at least for the 42% figure, in China, and it’s an estimated CPU2000 score.

Rather amazingly, AMD did this while screaming that Intel was unethical for doing the same thing in one of their tests.**

But even if you give AMD every benefit of the doubt as to these two numbers, after you study the SPEC results for either test, you come to the conclusion that Intel will have to do little to match integer scores (if not with a faster QX, than with a Penryn) and AMD superiority in floating point will probably matter much more for workstations than the typical server (and Penryn will use SSE4, which would in time address any FP advantage).

More importantly, AMD has said absolutely nothing about how any of its other processors, the Agenas and Kumas are going to fare against C2Ds or Penryns. This is a matter of far more interest to those reading this, and to people in general.

One More Slick Trick?

It’s hard to believe that if AMD really had a barnburner, they’d be so . . . quiet about it while getting ripped apart in the stock market at the worst possible time. Intel certainly wasn’t with C2D, and sharing well ahead of time surely did them no harm.

What’s more likely is that K8L/K10 will be catch-up, or almost catch-up, that’s not going to drive interested parties or the stock price wild.

So what is AMD likely to do assuming they don’t have the real deal?

What they’ll probably do is to take a page from Apple. At the earnings conference call, there will be further hints about how awesome the nextgen is going to be, with a lot more coming at the shareholder’s meeting in May.

Then it’s off to Tunisia! There, something like 185 journalists (hmmm, wonder who’s paying for them) are supposed to see ten Barcelona Agena FXs in action.

Funny, the trip is under an NDA, so I guess the only thing the journalists will be able to say is how wonderful this is, just in time for the annual shareholders meeting!

When that comes along, the shareholders (and us) will get more of the same, a well, highly selective smattering of benchmarks. If these chips star in anything, it most likely will be floating point, so that’s what will be emphasized.

How representative are any numbers likely to be? Probably about as accurate as Apple’s, but hey, we’re talking survival here.

The goal will be to get the shareholders to approve the doubling of authorized shares. After that, do everything possible, by means fair and foul, to pump up the share price and either issue more stock at a decent price or get another loan before letting anyone outside of Green verify the claims.

Could I be wrong? Maybe, let’s hope I am. But based on long observation and past performance, I doubt it, and if they really don’t have the real deal, I really doubt it.

Will it work? I don’t know, AMD has been getting a lot of negative press lately, and some of it has focused on its lack of veracity. Then again, much of both the technological and financial media are, well, easily impressionable and credulous, and a trip to a place off the beaten track is a trip, so maybe the stunt works.

**(Then again, the man who did that, AMD’s chief propagandist, Henri Richard, is no doubt suffering from a crimped lifestyle since he hasn’t been able to sell most his AMD options for a 200-400K profit every month like he did last year. But he’s not the only one. President and Chief Operating Officer Dirk Meyer and Executive Vice-President Thomas McCoy also belong to the “sell them (OK, 80 or 90 or 100% of them) when you get them” club. Hector is a little bit better about hanging on to some.
It should be also noted that as the price of AMD stock has slid, so has the option price of the shares, so the execs won’t have to apply for food stamps. Yes, Intel execs sell a lot of shares, too, but they usually hold a lot more shares than their counterparts at AMD, and the option price has been a lot closer to the stock price for Blue than with Green.)

Ed


Be the first to comment

Leave a Reply