News Intel 2Q25 Earnings

[511]

Then Intel is even more clueless than I thought.

Intel after 2010-11 has been the most clueless company they bought so many AI vendors / Altera / even got back into memory Business McAfee anyone it's a miracle sometimes how can someone survive after this much stupidity

 

[igor_kavinski]

McAfee

They got rid of that and now they have a dedicated NPU they could've used with McAfee for intelligent virus detection heuristics or whatever.

 

[gdansk]

They got rid of that and now they have a dedicated NPU they could've used with McAfee for intelligent virus detection heuristics or whatever.

Oh great, now we have the existence of the silicon waste unit arguing for integration with software waste.
McAfee is one of those purchases that Intel made which make no sense. Many of Intel's attempts at acquisitions are a strong argument for share buybacks. They were worse for Intel and its shareholders than if Intel had simply bought its own shares.

 

[igor_kavinski]

Oh great, now we have the existence of the silicon waste unit arguing for integration with software waste.

Waste+Waste ≠ Waste ?

Maybe?

 

[gdansk]

Waste+Waste ≠ Waste ?

Maybe?

If the silicon waste unit has a killer application, Intel won't have to write it. The independent software vendors will write it themselves and make money doing it.

 

[igor_kavinski]

If the silicon waste unit has a killer application, Intel won't have to write it.

They need to come out with something in the beginning to kickstart the "revolution". Otherwise, it will have the same fate as AVX-512.

 

[511]

They need to come out with something in the beginning to kickstart the "revolution". Otherwise, it will have the same fate as AVX-512.

AVX-512 is not a waste of silicon

 

[MS_AT]

Otherwise, it will have the same fate as AVX-512.

AVX512 had clear utility from day 0, so different story.

 

[igor_kavinski]

AVX512 had clear utility from day 0, so different story.

If they hadn't been disingenuous (mainly targeting it for enterprise) and pushed for its implementation in common OS functions, it could've become a mandatory requirement for Win11 and even Linux by now.

With McAfee under their own umbrella, an automatic use case for the NPU would've popped up immediately in the heads of the otherwise useless marketing people.

 

[NTMBK]

AVX512 had clear utility from day 0, so different story.

Yeah, it was useful, but Intel played stupid segmentation games with it. It's been around for 8 years now but penetration in consumer software is still practically zero. If they had been driven by engineering and made it universally available on all CPUs it would be ubiquitous by now, but sadly marketing won that fight and segmented the shit out of it.

 

[511]

Yeah, it was useful, but Intel played stupid segmentation games with it. It's been around for 8 years now but penetration in consumer software is still practically zero. If they had been driven by engineering and made it universally available on all CPUs it would be ubiquitous by now, but sadly marketing won that fight and segmented the shit out of it.

You know how idiot marketing is when even Xeons don't have AVX-512

 

[gdansk]

They need to come out with something in the beginning to kickstart the "revolution". Otherwise, it will have the same fate as AVX-512.

No. They made hardware, wrote drivers and OpenVINO. They did their job. Beyond that it's up to Microsoft and other software to find a use for it. Intel can differentiate their NPU by having good out-of-the-box experience but that's pretty much it. It was dictated from Microsoft for Qualcomm's benefit. Wasting more money than is needed on a distraction they're not even good at is prone to kill a cash-starved company.

The prime reasons for AVX-512 poor adoption isn't due to a lack applications rather a lack of a unified target. Intel shot themselves in the foot by trying to make it into a killer feature instead of trying to get a single target everywhere and to stick to it for all future designs.

 

[igor_kavinski]

Maybe Intel should now try to use the NPU for some of the missing AVX-512 instructions

 

[adroc_thurston]

Then Intel is even more clueless than I thought.

Dude Intel is the company that had a headstart in ML big iron race on everyone but Nvidia and completely (I mean wholly and irreversibly) fumbled the bag.
Nervana/Habana/GPU stuff actual, all dead.

 

[Thibsie]

Dude Intel is the company that had a headstart in ML big iron race on everyone but Nvidia and completely (I mean wholly and irreversibly) fumbled the bag.
Nervana/Habana/GPU stuff actual, all dead.

Yeah, yeah but regularly I wonder 'it couldn't be worse, could it?'
And each time there's more, lol.

'Dug so much they got to absolue zero. They still try to go deeper.' 🤪*

* sorry if this sound weird, this is translated from french.

 

[NTMBK]

Yeah, yeah but regularly I wonder 'it couldn't be worse, could it?'
And each time there's more, lol.

'Dug so much they got to absolue zero. They still try to go deeper.' 🤪*

* sorry if this sound weird, this is translated from french.

I gotchu fam, let me translate that into American:

 

[jpiniero]

Dude Intel is the company that had a headstart in ML big iron race on everyone but Nvidia and completely (I mean wholly and irreversibly) fumbled the bag.
Nervana/Habana/GPU stuff actual, all dead.

CUDA tho.

 

[511]

Dude Intel is the company that had a headstart in ML big iron race on everyone but Nvidia and completely (I mean wholly and irreversibly) fumbled the bag.
Nervana/Habana/GPU stuff actual, all dead.

Did the same with Mobile arm exists thanks to Intel as well

 

[Fjodor2001]

Who is going to prepay big money for newest 14A node, wait years praying it will actually work when they can do the same with TSMC and no exec will lose their bonus if suddenly TSMC misses deadlines?

Apple and NVIDIA:

Apple Is Reportedly Showing Interest in Intel's 14A Process for Future M-Series Chips; NVIDIA Could Be In Line For Adoption As Well

After a gloomy days for Intel, optimistic developments are being reported regarding its 14A, which could involve a partnership with Apple.

wccftech.com

Apple may be Intel’s last hope in the foundry business - 9to5Mac

Analyst Jeff Pu said that Apple could be among the customers considering Intel’s upcoming 14A process for future M-series chips.

9to5mac.com

 

[jpiniero]

Apple and nVidia:

Apple Is Reportedly Showing Interest in Intel's 14A Process for Future M-Series Chips; NVIDIA Could Be In Line For Adoption As Well

After a gloomy days for Intel, optimistic developments are being reported regarding its 14A, which could involve a partnership with Apple.

wccftech.com

That's just Wall Street shenanigans.

 

[Fjodor2001]

That's just Wall Street shenanigans.

Sure, it’s too early and no contracts are signed yet. But they should really need some second source, and then there are not many options.

 

[gdansk]

There is little doubt they're interested. But they aren't charities. Intel has to offer them working chips they think will be competitive and at lower price than TSMC.

Seems like a tall order.

 

[Doug S]

Intel doesn't need it, but saying it isn't a value add is an overstatement.
Google "site:reddit.com mediatek wireless".
Or "site:reddit.com apple broadcom wireless bootcamp"
Or "site:reddit.com atheros linux"

Intel wireless chipsets are a pretty safe recommendation. That Intel laptops almost always come with a good wireless chip is some value.

Is anyone choosing Intel laptops because they come with Intel wireless? No? Then it isn't a reason for them to keep that as part of the company!

Whoever they sell it to, unless they screw the pooch, will then became the new "pretty safe recommendation" and since it is generally delivered on a m.2 card if the laptop you buy has a crappy wireless card you can replace it with one from the new owner of Intel's wireless solutions group.

 

[gdansk]

anyone choosing Intel laptops because they come with Intel wireless?

The opposite. People avoid comparable AMD laptops because they often come with MTK junk.

And spinning it off from Intel is tolerable just don't say it has no value. Spun off they'll be a good acquisition target too.

 

[Fjodor2001]

There is little doubt they're interested. But they aren't charities. Intel has to offer them working chips they think will be competitive and at lower price than TSMC.

Seems like a tall order.

There’s also the geopolitical risk. What if the TSMC factories do not exist in X years, or are owned by China which sets an export ban for chips to the US.

How would Apple and NVIDIA deal with that if they have no second source. Perhaps better to at least have some second source, even if there’s a risk it’ll perform somewhat worse than the primary source?