Thread regarding Intel Corp. layoffs

I don't believe OP is ready for the woodchipper.

LBT is in the process of changing what F means in IFS, and I don't think he is leaning towards Fork.

I think we can all agree that Step 1 is to Steal Underwear.

Still working on Step 2.

@ey Apparently UMC is considering working with Intel to produce 6nm, in addition to the 12nm they are working on in AZ.

SMIC is up to 7nm, so would also seem a contender.

As others have said, this is just one possibility. The company could just run the fabs to the end and shut them down, or do something else entirely.

@f0 Yes, the church wants to start using leading-edge wafers during communion.

Warning: do not eat wafers. They are highly toxic.

https://www.tomshardware.com/tech-industry/blacklisted-china-chipmaker-smic-becomes-the-worlds-second-largest-pure-play-foundry-by-revenue-outsells-globalfoundries-and-others

@ce Intel is already working with UMC, because they are only up to 22nm, so it would make sense for UMC to potentially be interested in buying the 14nm fabs.

Intel has worked with a number of the smaller foundries over the years, so really the deal could come from any of them. They know the capability and the issues with how the fabs are configured.

7,11,13 !
What happens when MBAs and Lawyers bankrupt us right before our eyes !

This thread is really exciting and borderline magical…. Until I did a quick numbers analysis and nothing pencils out. When you figure a realistic WACC and likely ROI, no second tier foundry will be able to pay to take on old Intel fabs.

@ce thanks! Very informative.
It seems like civil discourse is possible here.

@by There are a number of smaller foundries that would find the Non-EUV fabs to be leading edge for what they are doing.

Of course not TSMC, or Samsung. Maybe GF but there really are a number of smaller foundries that could be a potential fit.

The idea is that they sell the wafers to Intel until those older nodes run out of roadmap, and then retrofit to whatever they want.

This sort of thing has been done many times before. Even Intel back in the day got several fabs this way.

btw, Intel could do the same thing. The current strategy is to do no foundry on less than 18a, but if some customer (like UMC has done) wanted an older node then so be it. Converting those older fabs to a standard set of design rules and PDK would cost billions and require substantial time if Intel were doing it.

It would take an existing foundry less time to do the same conversion, because they would be making the fabs be configured like their existing fabs. For Intel it would be a challenge, because they have no comparable, industry standard fab, and those older fabs are nothing like the fabs that have 18A-P, which is the first industry standard fab for Intel.

But if a customer could work with the existing configuration, then Intel might just do that for one or more of those older fabs. This is kind of like how the first attempt at foundry operated.

Absent that, the best option is to sell the fabs right now to any foundry willing to take them on, then contract the wafers as needed till those older products run out of roadmap.

Not claiming this is what must happen, but it seems the most likely outcome. IFS needs to be stripped down to just what can be offered to external customers, to minimize the capital needs and overhead, and to lower the risk level to something the company can live with.

The current IFS situation is a burning platform, able over time to take out the whole company if not stopped.

@b9 If Intel couldn’t do this themselves, why would a potential buyer be able to do this?

Who would be such a hypothetical buyer?
There’s only GF, Samsung and a few smaller outfits which can build out their own capacity without Intel volumes.

TSMC certainly won’t be interested.

@ay for 1274, no customers

@b9 sounds like a job for SMIC. I bet they could get their stuff running at F28 without retooling too. I know a guy.

@ap The idea is to sell the older fabs while they are still running near capacity.

The buyer can sell back the wafers, and that covers the cost of the eventual retooling.

Intel improves its balance sheet and avoids eventually having to either shut down those fabs and/or pay some foundry to take them.

Workers remain employed, although some would be cut by the acquiring foundry.

So it is more or less a win all around, the best option for Intel and the workers.

Intel can't really offer those fabs for external use because the company would have to do the retooling and may never get that money back.

Intel Fabs are facing a market-based reality unlike anything they have ever faced. Expect the company to do what needs to be done, now that it has a CEO who is willing to do the right thing.

F28 is full capacity untill end of 2026.

@ah there is no long tail for Intel. Unlike TSMC, who are still running even 150mm fabs, Intel's strategy has always been to outspend the competition on the latest equipment and throw the old factories away. Until they got caught in their own trap for a decade with 10nm-- while the big brains in PTD exhausted the periodic table.

Who would buy the non-EUV fabs?

They aren’t compliant with foundry tooling, PDKs and IP libraries.
IBM had to pay GF $1.5B to offload their fabs.

Maybe something like that but Intel is broke and retooling them for foundry will be quite expensive.

@ah Intel has no plans to offer anything older than 18A-P to any external customer.

The non-EUV fabs are currently busy supplying current chips for Intel, but that will disappear quickly once 18A is up and running.

So the say is fast approaching when there will be 0 demand for those fabs.

If they sell them now to a lagging node foundry, then that foundry can make wafers for Intel, then convert the fabs to be more standardized with the industry, and go on to make automotive chips and such.

This does seem like an inevitability, at a time when the company is actively considering selling off all of IFS, or at least stopping or slowing development after 18A-P.

Some workers would be laid off, but that was going to happen anyway when those older fabs were shuttered, because that is the only other possible outcome.

Now is not the time for magical thinking. IFS is about to get real.

Haha, because the product side is so healthy? Good luck with that.

They will need those other fabs to produce older chips that will still be needed in the eco system. That is where the cash cow of foundry is when you become a more mature foundry. TSMC makes boo koo bucks selling the older tech.

Remember, No one expects the Spanish Inquisition.

@a4 True, this site is infested with trolls who chose to spend their time attacking Intel.

Not that it has any effect, because their trolls are boring and delulu.

That said, I would be surprised if IFS remains the same over the next year, let alone over the next few years.

I fully expect it to be split up, with the non-EUV fabs sold to another foundry. This will help pay for 14A and those fabs would eventually be shuttered anyway, as they wont ever be used for external business.

Once IFS is down to a more manageable size, then see what happens to TD vs the Fabs. I expect to see TD be downsized and to stop producing production wafers.

Then there is outsourcing and AI, which are a certainty just waiting to happen.

So within a few years foundry could easily be 50% of what it currently is. Getting it profitable is the goal and not much else matters right now.

If you can't apply external feedback quickly, and keep scrapping key external customer silicon, what other outcome do you expect?

This is very obvious considering the financial status of the company. Also roadmap does not have any fancy chips.

This place is full of Taiwan and AMD fanboys that will say anything, I'm sure 90% of posts here are not from Intel folks

Duh any new specific details?