Intel announces AWS has become a client, Qualcomm likes its future tech, advances that as proof it's back in business

Intel has announced that Amazon Web Services has become the first customer for its foundry services packaging solutions, and that Qualcomm has become a partner of sorts in a silicon manufacturing process said to represent a step change beyond current Intel tech. Chipzilla revealed the new relationships at a Monday event titled …

[Pascal Monett]

"Gelsinger argued"

Of course he's going to argue, Intel failed it's 10nm proces for long enough to give AMD time to swamp the market.

With all due respect, Mr Gelsinger, nitpicking about how nanometers don't represent atoms is not going to change the fact that, for the second time in Intel's history, you are playing catch-up to your main competitor.

[Anonymous Coward]

This abuse of numbers is a classic marketing tactic for those whose technology is inferior to the competition's. Intel should know; they've seen it before. Calling a 10nm process node "Intel 7" and a 7nm node "Intel 4" is exactly the same thing as Cyrix's "Performance Rating", in which a 133 MHz processor was sold as PR166+. If you can't keep up with the competition, just use the same numbers they use and define them in the fine print. Cyrix had one thing going for them that Intel don't: their processors actually did outperform Intel's clock for clock; Intel's 10nm process doesn't provide its customers with any meaningful benefits over anyone else's 10nm process. It's just as hot and performs just as poorly. Whether or not you think features sizes are "meaningful", the simple fact is that Intel's processes deliver silicon that's about as capable, in terms of power and performance, as anyone else's at the same process node. "Intel 7" is not as good as TSMC 7nm and never will be. The numbers themselves may not be physically meaningful, but they are comparable and Intel are simply way behind. However Intel have something Cyrix never did: a seemingly infinite marketing slush fund with which to pay OEMs to keep building systems around their inferior technology. That payola is the only reason they're still in business at all.

The bottom line is that Intel plan to bring Sapphire Rapids to market as a 10nm part, while AMD will be shipping Genoa with (we assume) fourth-generation 7nm(+ or ++) cores. Not only did they miss the boat on EUV, they're now planning a fourth consecutive product family that will not be competitive with AMD's and by the time they start shipping 7nm parts -- branded as "Intel 4", in a gross insult to everyone's intelligence -- their competitors will probably have moved on to the next node themselves. Intel have realised they aren't going to catch up, ever. They tried to catch up only to fall even farther behind, then they tried the snow job ("these numbers aren't really meaningful anyway") which didn't really work despite the shocking number of gullible reporters who parroted the company line, and now they're down to their last play: lying through their teeth and hoping you'll be too stupid to notice it. Here's hoping that ends the same way it did for Cyrix.

[Anonymous Coward]

"the periodic table is exhausted"

For all practical purposes, the periodic table has already been exhausted. While it's probably possible to manufacture trace amounts of new elements with very expensive laboratory equipment, none of those elements is stable. Not even close, in fact. Even if the island of stability exists, there may be a couple of isotopes that are merely somewhat less radioactive than others; no model of which I'm aware posits that these are actually stable. No one wants alpha emitters, nor any kind of decaying atoms, in their silicon, and no one is going to pay millions of currency units for picograms of these exotic materials to put them to that use.

It's always dangerous to suggest that any avenue in physical science has been fully explored to its very end, but it does seem quite reasonable to suggest that for at least as long as Mr Gelsinger is alive, we're going to be stuck making computers from the same 80 stable elements humans have been using for millennia. Whether there is a heaviest element or it's possible to create stable superheavy isotopes are interesting questions but they will not be usefully answered before "Moore's Law" (really an observation about economics) peters out; indeed, it is doing so already. I don't envy Mr Gelsinger: he has taken on the impossible task of reviving a dead corporation that has lost its leadership position and much of its talent. No doubt reminding everyone who remains of Moore's founding role in the industry and wearing a brave face about his observation's -- and by extension the company's -- future is good for internal morale. But that's all this is. 3nm and 2nm work is already underway, and it's being done by Intel's competitors while they flounder around at the 7nm node those competitors have been using for years. If there are to be more nodes beyond, they won't involve superheavy elements, and it's doubtful they'll be implemented by Intel.