Intel's Death and Potential Revival

https://stratechery.com/2024/intels-death-and-potential-revival/

by feross • 2 years ago

25 21 2 years ago

xattt2• 2 years ago

Folks are really cheering on Intel to win, because no one wants to be left holding the bag. We all bought Intel because we all thought it was ol’ faithful, but they’ve really, REALLY shit the bed.

pavlov2• 2 years ago

What do you mean, holding the bag? Intel’s products have compatible alternatives from other vendors. If they went away, AMD would surely continue providing x86 compatible chips.

magicalhippo• 2 years ago

I assume holding the bag of stocks.

cheeseomlit• 2 years ago

They probably mean Intel shareholders 'holding the bag' as their once-valuable stock tanks. Which is probably a lot of us through 401ks alone

gjvc• 2 years ago

nonsense. Just buy AMD from now on, then ARM.

xattt2• 2 years ago

I meant stockwise. It just seemed like a sure bet a decade ago, like RIM the decade before that, and Kodak the decade before that.

bryanlarsen3• 2 years ago

That would be fine if and only if they both weren't dependent on TSMC. We need at least 3 viable sources of leading edge chip fab so that we're not down to one when one of the other 3 inevitably stumble.

Apple and Nvidia should be leading the charge here. They're the ones that are both vulnerable to a TSMC squeeze and have the resources to make a difference.

It appears that Nvidia is looking to manufacture the low end of their upcoming GPU's on Samsung, so they're doing at least something. But Apple appears to be doubling down on TSMC via exclusive contracts.

rbanffy3• 2 years ago

up to that point most new processors came with their own instruction set

Kind of. 8080 and 8085 share also the same ISA. Other families that did that were the 6502 (the 6510, the 6504), Z-80 (Hitachi's HD64180, which became the Z-180, Z-280, R800), and Texas' TMS9900, which was part of a whole family that spanned from the ill-fated TI-99/4 to the TI-990 minicomputers.

bryanlarsen3• 2 years ago

The 8088 and 8080 are not binary compatible, they're mostly source compatible.

rbanffy3• 2 years ago

I didn't say that - I mentioned that, previously, 8080 and 8085 shared the same ISA with different implementations.

klelatti• 2 years ago

Dylan Patel comes to essentially the same conclusions (and this is a better post I think):

https://semianalysis.com/2024/12/09/intel-on-the-brink-of-de...

The problem is there are two distinct objectives:

1. Maximise shareholder value

2. Maintain and extend Intel's (close to) leading-edge fab capability.

These are not necessarily consistent!

The Intel Board (bad as it is) is trying to deliver 1. and probably now doesn't care about 2. That's not what the US needs, so some sort of government intervention is inevitable if 2. is to be achieved.

And I'm not convinced that PG's IDM 2.0 strategy would have delivered either.

ableal• 2 years ago

Dylan Patel comes to essentially the same conclusions (and this is a better post I think): https://semianalysis.com/2024/12/09/intel-on-the-brink-of-de...

Thanks, good reading. Sharp retrospective and analysis, hopeful blueprint for the future:

"""

Intel Foundry won’t be laden with Intel’s lagging product team, Mobileye, or Altera. Intel Foundry will have one clear function, and it’s vital to national security and the future of America and the West.

Who should lead the charge here? Maybe a “Chip Czar” charged with restoring American logic prowess. We know someone with a great CV who has just become freed up for new opportunities…

"""

Farfignoggen• 2 years ago

Looks like the Next administration is more interested in rewarding billionaire donators with government positions regardless of if they are actually qualified for those positions! So look for even worse things to happen at those Technology Companies as the MBA types managing/running them only look for extracting from the Tech Companies the money and all other things be damned! (Que that Simpsons Schoolhouse Rock Parody of how a Bill Passes Congress and gets signed into law)

So How are all of you enjoying this new Guided Age of Technology Trusts and that new age that started in earnest beginning with the 1980 election and is really coming into its own with the election of 2024! Say hello to the Robber Barron Cabinet come Jan 20th 2024, and maybe 2029 will be a year not to dissimilar from the one that occurred one century before!

jmclnx• 2 years ago

I still do not really get this hand wringing over Intel.

The only thing I see are these institutional investors are very upset Intel is not like Nvidia. So they are complaining and these investors will eventually destroy Intel because their stock is not rising 500% per year.

So all this pressure on Intel since Nvidia became a thing is distracting Intel from focusing on their business. Causing more harm than good.

diziet_sma• 2 years ago

Intel products have been losing x86 to AMD in servers, desktop, and laptops. x86 itself is losing market share to ARM in the server and laptop space. Intel has no foothold in the fastest growing chip space (GPUs).

Intel foundry has fallen behind TSMC to the point the latest Intel laptop chip (Lunar Lake) was manufactured using TSMC. Recent attempts to attract external client s to fab using Intel Foundry have not met expectations.

Given all this, I think quite a lot of hand-wringing is warranted.

miohtama• 2 years ago

Let's not forget Intel share buybacks

https://www.calcalistech.com/ctechnews/article/bypn9cdrc

How Intel's $108 billion buyback gambit backfired—a cautionary tale for tech giants Intel’s focus on stock price over innovation left it playing catch-up in the AI race—and tech leaders should take note.

Buybacks artificially pumped the share price, which means any correction is/has been very brutal, more "bubble" like.

mikewarot• 2 years ago

Everyone assumes the next generations of chips that run AI are going to be horrifically complex and require superhuman levels of talent to optimize.

What if they're wrong, and there's a much, much simpler architecture that could be found, that could scale almost infinitely, and make better use of Dennard scaling?

DebtDeflation• 2 years ago

The fundamental flaw with Pat Gelsinger’s 2020 return to Intel and his IDM 2.0 plan is that it was a decade too late.

Correct, but if a decade ago was the right time to do it, then 2020 was the second best, and now is the third best.

Gelsinger’s plan was to become a foundry, with Intel as its first-best customer.

The only real alternative to this is to split Intel into two companies - a Foundry and a Products company - and effectively achieve the same thing via contracts rather than ownership.

I have a feeling that in 18-24 months we're going to be reading articles about how the best time to split the company was 2 years ago and now it's too late.

Intel's Death and Potential Revival

Related Stories