News Intel hires Morgan Stanley to protect itself against activist investors

[bit_user]

Accountants are the enemy of r+d and engineering as is short term expectation.

Investors come in two basic varieties, short term and long term. Both want to see profits. The long term investor sees long term growth as the benefit. The short term investor looks for a quick buck.

I recall seeing an article saying that Intel was next on the list for the most up to date lithography machines, these require investment but give the potential for good returns.
Intel can make good products, accountants held too much sway through the 2010s and development was stalled. “We are making good money, why change what we are doing?” There wasn’t real competition in the main market Intel served, x86 so.. why worry.

I wouldn't blame accountants for what are really senior executive level decisions about company strategy and priorities. I think the senior executives at Intel made a decision to cater to Wall St. and it bit them in successive fab delays. These delays started all the way back in 2014, with their 14 nm node. They just snowballed from there.

Things change, sure Zen 1 wasn’t brilliant the development has been consistent, arguably Intel has fallen behind.

The rise of AMD required Intel's node progression to falter while AMD and TSMC both surged ahead. Zen 1 and 1+ were slower than Skylake, but AMD was bold enough to go with 8 cores where Intel had been parsimoniously giving us only 4. It should also be noted that AMD was still on GloFo, at the time. Where things really changed is in the Zen 2 and Zen 3 era, which switched to TSMC and utilized chiplets to scale up to 64 core server CPUs when Intel was stuck at 28 (and then 40).

Intel has concentrated on accelerators which are nice in themselves but not general purpose. If a company doesn’t need the accelerator it doesn’t licence that part … no money for the inbuilt silicon. It looks good having it available but if it’s not used, no revenue.

AMD is like: "we'll just give you more cores and then you can use flexible software solutions instead of hardware engines". I'm paraphrasing there, as I don't know if they ever actually said that, but it's what they seem to be doing.

Especially when you account for all the delays, I think accelerators really bit Intel hard. We'll see if they pay off in the long run, but I'm not so sure it was a good bet.

Intel fabs making 3rd party silicon.. tough one here,

IMO, it's fine unless you're competing with Intel. I don't blame their competitors from staying far away from their foundries.

Spinning the fabs out completely could work, the required investment into that possible entity is huge, where is GloFlo now? Investments in technologies and techniques and maintaining those investments will be a drain until a reputation can be established,

GloFo had the reputation, but just took their foot off the gas with the 7 nm node, once they noticed it would be late. They've been milking legacy nodes and now probably stand to get utterly clobbered by excess Chinese capacity, in that market.

 

[thestryker]

Well, FIOS is still doing new buildout, even if the pace has slowed. It just came to my neighborhood in January.

They haven't added a new region since 2010 only building out in existing.

 

[stuff and nonesense]

GloFo had the reputation, but just took their foot off the gas with the 7 nm node

If I recall correctly the 14/12 node was a Samsung process implemented by GloFo. If that is the case they took their foot off the gas at 28nm.

 

[stuff and nonesense]

The rise of AMD required Intel's node progression to falter while AMD and TSMC both surged ahead. Zen 1 and 1+ were slower than Skylake, but AMD was bold enough to go with 8 cores where Intel had been parsimoniously giving us only 4. It should also be noted that AMD was still on GloFo, at the time. Where things really changed is in the Zen 2 and Zen 3 era, which switched to TSMC

Agreed, getting the cores out from under GloFo was a necessity. TSMC have been an incredible partner. TSMC seems to be half a step of AMD’s requirements perhaps due to Apple using the most cutting edge nodes a year earlier ?