Thread regarding Intel Corp. layoffs

If a deep market pullback and economic slowdown is what is on the agenda next, then Intel will have yet again handled the node transition (more or less) correctly.

Imagine what happens (in a rapid slowdown) to all those memory companies who are racing to add massive capacity?

Of course, if instead the (post 2008 GFC) economy continues to roar, due to being run hot by the politicians, then yeah, Intel will be caught with insufficient capacity.

In the past, semis did not fare well when demand went through the roof, as that always, without exception, was followed by a demand collapse.

But hey, central banker money printer goes brrrrrr..

Excuses to cover bad decisions and bad executions. This is the norm in Intel to blame others to hide incompetence.

@eb I think the effect of the yield ramp was his point.

The company is doing the node transition, but the newer node (yield) does not yet justify ramping any more quickly.

Add to that the issue about how AI is being shoved at consumers, causing buyers to stay on the older nodes. You can be sure that Intel was unwilling to see that one coming.

The surprise may be due to forecasting but the real issue is probably more about a mismatch in forecasted demand. Pushing AI at consumers was d-mb.

@dn
OMG, in each of those previous successive ramp downs of old nodes Intel had a new node that were already producing high yields. And in each of those successive changes from ramping up new and ramping down old Intel was making money. Things will work themselves out are you F-ing off your rocker? The debt keeps stacking up, the ownership of the fabs is no longer solely Intel, and you might want to take a good look at the SCIP contract created with Brookfield for the sale of 49% ownership of those FABs because they're huge liabilities if Intel doesn't meet its commitments.

@dp No, the real reason was that they sat around in conference rooms talking, while looking at ppts.

It's all about the talking and ppt'ing.

Let us never talk about this again.

@c5 Any forecasting inaccuracies were mostly the result of biased roadmaps, combined with the fabs sandbagging and manipulating the forecast they were provided.

The actual generation of the forecast had issues but produced an output consistent with the inputs.

But the real issues were in Sales inputs into the roadmap process. Just outright lying, to each other and to everyone else. Most of those people are rightfully gone at this point.

@b3 Those people were low talkers and ppt pushers, so needed to go.

The reason why the old nodes were idled was because the roadmap and resulting forecasts suggested it was time to start shifting capital and people to the new nodes.

The company has done this many times before. Sometimes they get more demand than expected and can't deliver. Other times demand falls and it all works out.

It is always a challenge to get it right, and a slow yield ramp makes it even riskier.

As others have said, this will work itself out.

Also, there is no going back to having some 1000 people doing nothing but producing roadmaps and forecasts, because Intel can no longer afford to do that.

Work with what the company can afford to have.

@a7 The real problem is that they cut the people that run the old nodes. The equipment currently sits idle and is on the market for resale. This is why they've brought back so many people who were let go as GBs at a premium.

@b3 “Ridiculously accurate” doesn’t match what I saw. Forecasts, schedules, and LRPs were DOA because Intel couldn’t execute. If execution is broken, output planning is fiction.

@b2 By the way, by 'large number', I mean all but one person.
The group used to be dozens of people. The modeling alone took about a dozen people, and was both mind-numbingly complex and ridiculously accurate.

So there was room to make it all less perfect, at much lower cost.
AI is certainly doing that.

This is coming for all jobs, and those who are left are simply going to have to live with something that is cheaper and imperfect.

Resistance is Futile.

@b0 A large number of those who handle forecasting were either let go, left, or retired since Fall 2024.

There is a group that is trying to forecast using AI, and they have repeatedly proven incapable of forecasting anything new. AI (so far at least) is great at extrapolating but has not been built with enough inputs to forecast something which did not exist before.

The people who did that are almost entirely gone. Even the group is gone, including the entire mgt chain.

AI will eventually be able to do a great job, but not sure it will ever outperform what it replaced. At least the previous system could be explained to customers.

Read the articles about how the ASIC business is at a $1B run rate and growing by 50%.

Short of getting a whale contract, this custom silicon is helping provide proof points for potential customers and also helps IFS learn how to be an actual foundry.

These contracts are on 18a, and presumably will port over to 18a-p now that there is a PDK for that. 18a-p also seems likely to help grow the ASIC business.

The contracts include design, fabs and packaging, so are helping all aspects of IFS ramp up too handle meaningful customer volumes.

Not saying this cures the current 18a underutilization or does anything to help client demand move away from older nodes.

People do not want AI for much personal use, and certainly don't want their PC spying on them, so stop it already.

I agree with others that the stock could retest the gap it left behind at the start of this big rally (~$25), but I also like what appears to be a lot of business improvement in the second half of the year. The company is hinting that as the timing for the first big IFS customer, presumably for 18a-p, since that is the first industry std PDK for Intel..ever. Packaging deals are happening already. 18a yields will presumably be less awful, helping it to ramp less slowly.

The market may be about to help the stock in the effort to retest, and I'm ready to be long again.

Someday the stock will stop being so exciting, and I will then have to chase something else.

Look towards the group that handles forecasts. Once you see that, all explanations will become clear.

We have a serious failure with a single ELT member who has yet to be shown the door and until they are gone, this pi-s poor forecasting will continue. Just wait until wall street sees how much they miss this year - stock will be back in the 30's.

Trust me, we will have good chips tomorrow. Even better than the ones we couldn't supply when you needed them. Great marketing. Intel Su-ks!

It's like when the coffee stand runs out of coffee in the middle of the day. Intel makes chips. "Were out of chips". What the fu-k are you doing here???

This is due to the gap between producing on older nodes and ramping a new node.

This is nothing new, but the timing is unfortunate.

Once OR and F52 get yield improvements, then the wafer starts will get serious.

The difference in that capacity and ALL of the older fabs is striking.

Soon come the day when the older fabs are either sold or shut down, so don't be there..

Intel not being aware of market trends, not planning for capacity and not being able to scale for chips that customers actually want to buy says a great deal to any company thinking of using Intel fab as a service. It's not good.

@OP Fake News

The issue is that no capacity is being added to old nodes, and that is in no way a bad decision.

Got nothing to do with headcount reductions, other than the additional headcount that would have been needed to add more 7nm fabs. But that would be d-mb.

You call it a minor blip.🤷‍♀️ Others might call it complete incompetence.

Still have faith in our leadership?

Yes. This is just a minor blip on the path to success.

Sorry we cant make the chips you want. We got rid of the equipment and the people. But...build you chips here. lol