AMD celebrates 500 million CPU sales
What price a billion?
Microprocessor designer and seller Advanced Micro Devices - the onetime chip manufacturer - is marking its 40th year in the chip racket business. today, the company said it has shipped more than 500 million x86 and x64 chips, Oh, and it also announced a shameless ultra-thin notebook giveaway through Twitter, hoping to convince people it's cool.
What makes AMD cool is not shenanigans like giving away four Hewlett-Packard Pavilion dv2z notebooks to Twitter addicts who listen in on AMD's feed and respond correctly to questions posed by AMD quiz-masters (details here. What makes AMD cool is rather good chip design and pushing the envelope on the x86 and x64 architecture in ways that you can bet that Intel would never had done on its own - or at least not at the fast pace Intel has been compelled to do a few times in its history because AMD was out in front with better chip designs.
AMD's historical albatross, like other chip makers, has been its inability to keep to its roadmaps. But even a cynic has to admit that since the Opterons were announced in 2003, AMD has hit most of its roadmap milestones - excepting the quad-core 'Barcelona' launch, which had a bug, and some sliding on the dual-core 'Santa Rosa' chips two years earlier.
The question now is whether removing itself from the chip making part of the business is going to make AMD better at hitting its goals or not. Even though AMD still has a sizable 34.2 per cent stake in GlobalFoundries, the former manufacturing arm of AMD that was spun out in March, the company gets to pretend that it is free to design and sell chips now that it's not encumbered by the substantial capital investments and technical challenges of moving to ever-smaller chip circuits.
But don't be misdirected. AMD is most certainly tied to GlobalFoundries, and its future is every bit as staked to the new Fab 2 foundry that will be built in Upstate New York after a groundbreaking this Friday. This wafer baker will use 300mm wafers and implement chips with 32 nanometer processes. It will cost an astounding $4.2bn, and it's set for volume production in 2012. The company says that this is the largest economic development project in the history of New York, unless you consider the nearly 100-year history of IBM.
The wonder of AMD is not that it can ship 500 million processors, but why it has been so difficult to make it 1 billion chips or more, giving AMD some profits (perhaps) and a larger piece of the market share for x86 and x64 chips (that should have been easier).
AMD bears some of the blame for this, with its sometimes choppy deliveries of new chips over its long history. AMD owes its entry into the x86 chip market to IBM, since IBM's supply chain policies back in the early 1980s when it entered the PC racket required it to have two sources for components. So AMD licensed Intel technology and became a second source, and it wasn't long before protracted legal and technical battles between AMD and Intel started.
It has taken more than a decade for AMD to shift from reverse-engineering Intel's x86 designs to setting the pace in x64 chip designs. But at this point, Intel and AMD are drawn pretty much even as Intel has pulled an AMD and reverse engineered the Opteron processors as its Nehalem family of chips. (Yes, that is an oversimplification. But it is basically true). And with AMD having lost money for 11 quarters in a row, Wall Street an investors are losing patience. Good designs should bring some money to the bottom line, but so far, AMD has not demonstrated it can do what public companies are supposed to do: make products and make money too.
If AMD is thinking then next 500 million microprocessors it sells will be easier than the first 500 million, it is in for a big surprise. This chip business only gets tougher as IT matures, and AMD has all the same risks that it had before with its fabs.
The difference is someone else - namely the government of Abu Dhabi - is footing a lot of the bill. AMD had better hope that GlobalFoundries gets some more customers to use that pricey fab so it stays in business and keeps competitive with Intel's wafer bakers. Getting IBM to shut down its own fabs in East Fishkill, New York after the 45 nanometer generation would be a good start. And considering how IBM's Systems and Technology group is not making any money these days, it is an obvious move for both IBM and GlobalFoundries. ®
Parallels between IBM's Systems and Technology group and Sun Microystems
Timothy Prickett Morgan writes, "IBM's Systems and Technology group is not making any money these days"
And what do they do?
"server and mainframe computing, offering UNIX and Windows-based offerings; ...Linux-based products. Its Microelectronics unit makes microprocessors and memory devices."
Wait a minute... people have been beating on Sun for declining revenue and lack of profits - and IBM with their Mainframe, POWER, and x64 systems is running into the same issues?
Anonymous posts, "So that division isn't making money right now, who is."
That is an interesting comment. Sun basically thought the same thing, after the post-Y2K recession, exhibiting itself as the "dot-com bust".
Is Mr. Morgan correct, in believing that the IBM's Systems and Technology group should leverage the $4.2bn GlobalFoundries fab (instead of building a new fab for a money losing organization) or make the same mistake as Sun did?
IBM won't close Fishkill
For one thing it is a Superfund site ready to be put on the list. They don't need the publicity or the cost. So that division isn't making money right now, who is. With all of the tax breaks that they get out of the State of New York. The political cost is way too high.
AMD could have done better.
"What makes AMD cool is rather good chip design and pushing the envelope on the x86 and x64 architecture in ways that you can bet that Intel would never had done on its own - or at least not at the fast pace Intel has been compelled to do a few times in its history because AMD was out in front with better chip designs."
x86 should have died. Too many complex legacy modes and instructions. You're right that Intel was abandoning the x86 to make inroads with new designs. It's unfortunate, although understandable, that AMD took the route it did by pursuing to extend the life of the x86 architecture. As a result, we're left with even more complex modes and longer electron paths due to the extra transistors required.
That being said, I've been a fan of AMD's prices. I haven't purchased AMD since the Phenom VT bugs however.