Pat Gelsinger Writes to Employees on Foundry Momentum, Progress on Plan

[LittleBro]

Unsurprising. VW is closing down German factories too.

The energy cost in Germany is just way too high for advanced manufacturing anymore. And given their energy grid is run by clowns, that's not going to get better anytime soon.

This is not at all Germany's fault.
Intel asked German government for funding and they approved 6.8 billion €. Intel wanted even more (10 billion) after and the government declined.
The whole facility had estimated costs of 17 billion, so Intel dared to ask German government to fund more than half of the costs. Ridiculous.

Later Intel extended plans for plant in Germany in terms of building two plants in for total costs of 30 billion €.
Germany approved funding of 9.9 billion €, that's roughly 33% of the whole investment. Is that not enough for Intel?
Who is/was the real investor here? Intel or Germany?

Now, this amount of money will be redirected somewhere else:

Germany: Intel delays construction of Magdeburg factory – DW – 09/17/2024

US chipmaker Intel has delayed a construction project by two years amid cost-saving measures. German Finance Minister Christian Lindner called for funds earmarked for subsidies to be used to plug a federal budget gap.

www.dw.com

Intel hits pause on chip fabs in Germany and Poland amid financial crisis

Chipmaker lost $1.6bn in Q2, will spin off foundry business

www.datacenterdynamics.com

Poland plant was estimated to cost 4.6 billion €. It was meant to be an assembly and testing site, not chip manufacturing plant as was meant for Germany.

Anyway, postponing of plans in EU was more than expected after Intel lost 1.6 billion $ in Q2 2024. More layoffs are to be expected, I'm afraid.
Let's hope that upcoming Arrow Lake is a miracle and not will not really end like this.

 

[R0H1T]

This was always way too overambitious, especially coming after Covid. China's expanding their "homegrown fabs" at an unprecented rate, and India too, although we're way behind for now. Anyone thinking making 10~20 fabs at bleeding edge with some demand tapering, if not worse, was totally unexpected probably's sniffing Intel's cheap glue!

And yes, China could likely overtake the installed capacity of TSMC+Intel combined in the next few years.

 

[iameatingjam]

They went from 130k to 110k. Big yikes

Yeah, intel is bloated. I know they also make the chips, well did/plan to in the future/ whatever. Still, their employee count compared to AMD is striking.

 

[R0H1T]

Probably too much of upper/middle managers, and yes men trying to please the bean counters! I don't even remember the last big thing from Intel post AVX512 unless you count copying ARM's big.LITTLE as innovation?

 

[Assimilator]

And yes, China could likely overtake the installed capacity of TSMC+Intel combined in the next few years.

Not leading-edge nodes, though.

 

[redeye]

so stock buy back was a good idea?… not.

This was always way too overambitious, especially coming after Covid. China's expanding their "homegrown fabs" at an unprecented rate, and India too, although we're way behind for now. Anyone thinking making 10~20 fabs at bleeding edge with some demand tapering, if not worse, was totally unexpected probably's sniffing Intel's cheap glue!

And yes, China could likely overtake the installed capacity of TSMC+Intel combined in the next few years.

but Taiwan is a “part” of China… a distinction with a difference…
in the future (i guess) China does the “big” stuff, Taiwan does the “little” stuff, and INTEL does the stuff that can be done while playing around… “really guys, were not just foolin’ around we have press releases!… were really workin’…)

 

[Dragokar]

I think there's a good chance Germany isn't all that worried about this either, with their recent budgettary issues...

Depends on how and who is looking at it xD

 

[Count von Schwalbe]

No it's not. The extra money you pay a cloud provider is money you save by not having buy hardware. Or hire a network specialist. Or a security expert. Or a devops engineer. Or having to replace your entire hardware stack every 3 years because it's now longer fast enough. Or having to deal with a firmware update that bricks every SSD on every node in your production cluster and your site is down for a week.

Mostly economies of scale, IMHO. Get big enough you run it all yourself.

Still a data center I guess, but self hosted. Edge computing is a dying breed.

 

[_roman_]

Even reviewing games at 720p with a 4090 doesn't seem to be fooling people now.

That is a methodical choice. Feel free to think about why it's done. There are other gaming benchmarks also just for the raw numbers in the common resolutions.

 

[dyonoctis]

So literally every cloud application that already exists is "insufficient"?


They will if it's the only possible one.

You also can't ignore the fact that on-device processing is getting pushed for select A.I processing where cloud latency makes for a bad UX. Software companies are spending R&D money to take away some processing from the cloud :
Gemini Nano - Google DeepMind
It reminds me of the discussion about EV everywhere for everyone and only EV. Some technologies can be an upgrade for some people if the environment is favorable but for everyone else? This is going to be a downgrade.

Don't get me wrong, I'm not saying that nobody can benefit from cloud processing, just that I don't see it becoming a one-size-fits-all solution. Phones and tablets are actually getting insanely powerful for their size, while network infrastructures aren't really seeing blistering progress in speed/latency/range. It's getting better. But slowly. Faster network interfaces also have an impact on powerdraw for mobile devices. Total Cloud for office workers in a first-world country, is easy peasy, for a mobile device though? To me, it's looking like this will be hella expensive and a logistical headache.

The hybrid model just looks like the simpler solution

 

[ScaLibBDP]

In the letter Intel's CEO clearly informed that Intel is Not going to compete on AI training. Here is another news:

Intel's Falcon Shores Future Looks Bleak as It Concedes AI Training to GPU Rivals

Intel’s Falcon Shores Future Looks Bleak as It Concedes AI Training to GPU Rivals

Intel’s Falcon Shores future looks bleak as it concedes AI training to GPU rivals On Monday, Intel sent a letter to employees detailing its comeback plan after an abysmal second-quarter […]

www.hpcwire.com

 

[Wirko]

A subsidiary structure will unlock important benefits.

Subsidiary means hunting for subsidies ... how come we didn't understand that earlier?

 

[Steevo]

No new products at Intel, CEO writes strongly worded email expressing displeasure.

What is this garbage?