Goodbye, Advanced Micro Devices
Board: Advanced Micro Devices
I see zero chance of AMD surviving as an independent company without some kind of engineering miracle happening. I've heard that engineering got hit at roughly 30% layoffs, which will kill AMD. AMD might have survived by reducing sales and marketing and getting better products out there, but now they'll have salesman trying to sell inferior products with no better products on the horizon. If better products ever appear on the horizon, it will Osborne existing products. Engineering miracles happen about once every couple of decades; I don't hold out hope. Yes, SeaMicro plus Fusion could suddenly generate a solution that quadruples AMD's server share. But it's not the way to invest. I think AMD is gone in two years, tops.
Since blame is so American, I'll blame the ATI purchase. AMD paid $5B for ATI. AMD plus ATI is today worth a third of that. A merger seemed to make sense for both companies, but giving ATI shareholders a $5B bonus back then means that both companies are doomed now. Although others did more to directly kill AMD, the ATI purchase put AMD in such a bad debt situation that it led to divesting the fabs. Bonds at 5% might have been really cheap in 2006, but are crippling today. The only way to refinance those bonds depends on products that make people confident AMD will be worth the investment. The market didn't like the ATI purchase either, which hurt AMD's financial flexibility to issue more stock. So ATI was the beginning of the end.
I think AMD settled too cheaply with Intel. When the lawsuit was ongoing, Intel had to play nicely. Once it settled, and ARM arose to protect Intel against any claims that driving AMD out of business would create a monopoly, Intel has played hardball in every market. Before, Intel always left AMD the least profitable market segment to milk and survive. AMD has no market segments left. Although, to be fair, perhaps Intel was not intending to play so hard, and the macro economic picture hurt AMD worse than expected. Naw. Intel wanted zero chance of AMD hurting them ever again. AMD is (was) a *FAR* greater risk where Intel's margins are highests. Even if Intel swaps ARM for AMD, that's still apparently a good swap in the land of servers. A living lawsuit might or might not have gotten AMD more money to survive, but if AMD won (or seemed to be winning) they should have gotten better terms with Intel on what Intel would not do to AMD.
I blame Global Foundries. When they started actively pursuing other customers, it created the possibility that AMD would not be able to get all the wafer turns they needed. So AMD started pursuing other foundries, even though they were getting their best price from GF. Now GF is losing its biggest customer. GF tossed a hand grenade in anger at the other guy in the same lifeboat, and just now is realizing the boat is sinking. How culturally appropriate. We all know GF forced out Dirk, which was the beginning of the end, even though nobody says so publicly.
And I blame GF. Boy, do I blame GF. In order to support the fiction that Dirk was driven off for not pursuing tablets and smart phones aggressively enough instead of pursuing second sources, the Board of Directors, dominated by GF, forced AMD to put resources onto tablets. I believed, and stated here, that AMD would get into tablets by improving performance and shrinking lasers on Fusion parts. I *don't* blame GF for falling further behind in processes than Intel. That happens, it sucks, and Intel is the gold standard. But I am sure that a Fusion APU made at 14 nm would be the best performing tablet product out there with adequate battery life (and battery technologies are also improving). Think of high density smart phone screens powered by Fusion. Now who is going to make that part? Not engineering, which just lost 30% of its folks. Is Read going to do it all by himself? He's a suit, not an engineer.
And one more bit of blame for GF. Sure, they had to be pissed when AMD ordered fewer wafer from them and some wafers (generally not many) from other folks. Abu Dhabi had invested all those billions of dollars in buying GF assuming it would be an ongoing revenue source, not money pit. But AMD wasn't happy, either. The best way to reduce costs for AMD right now is order fewer wafers. At zero wafers, zero production costs! But at zero wafers, zero revenue. GF should have realized it wasn't that AMD was turning away from GF, but that everybody was turning away from PCs. Like it or not, GF was chained at the hip to AMD, and once GF started making things worse rather than better, both companies started drowning. < rank speculation> I bet IBM is going to pick up some GF pieces real cheaply
And I blame all the departures. Corporate memory has left the building. AMD used to be Chimpzilla, with no fabs. Fabless CPU companies survive by excellent and accurate communications between sales and ordering/manufacturing. Heck, fabbed CPU companies benefit from the same. If every wafer turned is a wafer sold, you keep inventory low. Costs can be much less than owning a fab in bad times, and times are bad. AMD just wrote off $100M worth of inventory. How does that happen when you own no fabs? Meyer and Rivet were the ones whose memory extended back to last century, and they were pushed out. I don't know what's going on inside AMD in this bad economy, but I bet every salesman, desperate to keep his job, padded his estimates based on sales that didn't happen. I mean, if the CEO is a salesman rather than an engineer, it makes sense that he'd push what he knows. Perhaps that's unfair; perhaps sales were bundled, with "I'll buy 50K PC APU and 50K Notebook APU and 10K Server CPU" and the whole sale fell through when the APU were delayed or didn't perform to spec. But from what I know of sales staff, I bet they were a big part of it, and Read doesn't realize, probably cannot realize, that sales isn't important when demand is low because products are behind.
By the way, you know the argument eachus makes about playing Russian Roulette with each design? Each time AMD changes focus, reassigns engineers, they are pulling the trigger. It's really hard to win at Russian Roulette if you take three times as many shots as your opponents.
Unlike some, I don't particularly blame Read. He was the wrong man for the job, but it's hard to know that until you're actually failing in the job. So he thought he could do it. He certainly could not say no to the BoD on anything; they put him in, and they obviously drove out far better people for the position.
I think AMD now exists solely as a takeover candidate. Samsung might want them. Sony might want them, but they've got issues. Microsoft and Apple both might want them. Even IBM might want them. If GF wants to stay alive, they might have to throw a few more billion dollars down the rabbit hole. But without a flush white knight undoing the engineering layoffs like yesterday, I think AMD is dead, except as a subsidiary of a friendly buyout.
Buying AMD now is pure speculation that they will get taken over. AMD can probably last 18 months based on cash flow and reserves. Intel has pushed AMD too hard and AMD broke. That also hurts Intel's margins in the short term; in the long term, well, who knows? Tick tock drove Intel too far ahead of AMD for AMD to survive without Fusion, and Fusion was late, and then resources were pulled off Fusion for pipe dreams. Nobody sat down and asked, "How can AMD survive for three years to get a new tablet product out there if we aren't also improving Fusion so our sales won't tank?" Okay, so maybe I blame Read some.
I still have a few thousand shares left, almost all of which could have sold at $90 (pre-split) in 2001 or at $40 in 2006. I did sell here and there along the way, so all my remaining shares are in the black. So are my mom's. Sorry, sis. At least I did tell her that it would be a good time to sell back in 2005-2006, and she chose not to. But I'm ready to wash my hands of AMD. Good luck, AMD longs. You'll need it.