With Intel’s latest layoffs, will the Ohio plant ever be built?

[Daniel Nenni]

Then-Intel CEO Pat Gelsinger, Gov. Mike DeWine joined by politicians and business leaders break ground ceremony for Intel’s new semiconductor manufacturing site, September 9, 2022, in Licking County, Ohio. (Photo by Graham Stokes for the Ohio Capital Journal / Republish photo only with original story)

Ohio’s top union leader is disappointed and concerned following Intel’s announcement of significant layoffs, which could affect the state’s multi-billion-dollar manufacturing plant.

The corporation refused to answer questions on how this could impact the state.

“The last 100 days have been very concerning for me,” said Mike Knisley, with the Ohio State Building and Construction Trades Council.
Thousands of union workers and billions of dollars — stalled.

“A lot of sweat equity into making sure that they were ready when Intel barked — we’re ready to move with this,” Knisley said. “That’s a little bit of a disappointment … I just don’t know in present time if that’s gonna be viable or not.”

WEWS/OCJ has been reporting for years on tech giant Intel’s massive semiconductor manufacturing facility — one that was supposed to be operational by now — and its financial hardships.

Knisely explained that workers are concerned now that the company has announced it will lay off a significant amount of its workforce, less than a year after cutting 15%. Bloomberg News reported 20%, but the company didn’t respond to comment on that number.

Intel’s stock has dropped by more than 50% in the last year, while the industry has grown by more than 120%. After spending 25 years on the Dow Jones Industrial Average, Intel was replaced on the index by Nvidia, a leader in artificial intelligence, in late November. The company also had mass layoffs in October, cutting 15,000 jobs.

An Intel spokesperson declined to answer questions about how this could impact Ohio but pointed to Intel CEO Lip-Bu Tan’s statement, which said, in part: “We are navigating an increasingly volatile and uncertain macroeconomic environment … We are seen as too slow, too complex, and too set in our ways—and we need to change.” Still, Ohio Gov. Mike DeWine remains positive that the plant will be built by Intel’s updated date — in the 2030s.

“They put now close to $8 billion into the ground,” DeWine said. “There’s going to be chips that are going to be made there, I will guarantee you that.” But he acknowledged the struggles the company is facing.

“What happens with Intel, what the future is with Intel — we certainly don’t know,” the governor said.

They are not the only major corporation cutting the work of thousands of Ohio workers. Microsoft recently stopped a $1 billion project to create multiple data centers in the state. The company did not respond to requests for comment.

“It would be a bad look for all of Ohio, including us with the trades, if everything went south and they just walked away,” Knisely said about Intel.

Still, the governor is telling Knisely not to worry.

“No company is going to invest as much money as they have already put into Ohio and walk away from it,” DeWine said.

 

[Daniel Nenni]

It was not discussed at the Intel event but personally I think Intel will focus on US based manufacturing moving forward so yes the Ohio fabs will be built. The other question is when will they be equipped? My guess would be in the 2030s as foundry manufacturing demand ramps up. If the semiconductor industry wants second source manufacturing these fabs will definitely be equipped, absolutely.

 

[siliconbruh999]

It was not discussed at the Intel event but personally I think Intel will focus on US based manufacturing moving forward so yes the Ohio fabs will be built. The other question is when will they be equipped? My guess would be in the 2030s as foundry manufacturing demand ramps up. If the semiconductor industry wants second source manufacturing these fabs will definitely be equipped, absolutely.

I think Madenburg will be canned and should be Intel doesn't have the demand to build this much should focus on filling their current fabs.

 

[Daniel Nenni]

I think Madenburg will be canned and should be Intel doesn't have the demand to build this much should focus on filling their current fabs.

I think TSMC really eclipsed the Intel Madenburg project. A JV with customers is definitely the way to go for international fabs. I think we will see more independent re-shoring semiconductor cuts in favor of these types of JVs with customers.

 

[dkr1986]

It was not discussed at the Intel event but personally I think Intel will focus on US based manufacturing moving forward so yes the Ohio fabs will be built. The other question is when will they be equipped? My guess would be in the 2030s as foundry manufacturing demand ramps up. If the semiconductor industry wants second source manufacturing these fabs will definitely be equipped, absolutely.

The construction at Ohio is actually going well per this. I track this Youtube channel for construction progress on Intel Ohio. I found that the construction progress has only accelerated or progressed well after the announcement of "delay", don't know why. I am guessing the construction will be completed but the fabs will be "shells" for future equipment move in. Or progress slowly or paused in future!

 

[XtorTourist]

The construction at Ohio is actually going well per this. I track this Youtube channel for construction progress on Intel Ohio. I found that the construction progress has only accelerated or progressed well after the announcement of "delay", don't know why. I am guessing the construction will be completed but the fabs will be "shells" for future equipment move in. Or progress slowly or paused in future!

The shell part of the buildout likely has hard time commitments at a few levels. I think some of it is related to meeting CHIPS funding requirements plus the state incentives. But I think the big one is that the construction companies have already planned out and committed their labor, supply chain, etc. to building the shell on a inflexible time frame.

 

[MKWVentures]

They should complete the shell but there is not a model showing they need to the capacity anytime soon. Some shells come back to life, so do not.... and shells do not grab top dollr when sold. Subsidies and litigation will most likely be as much of a decision factor as wafer demand.

the repeated delays and uncertainty were not a surprise. "If you build it, they will come" only works for ballfields in the midwest... not Fabs

Lets see if we can get Ireland and Arizona loaded before we worry about Ohio.

 

[Daniel Nenni]

Lets see if we can get Ireland and Arizona loaded before we worry about Ohio.

Ireland is 4nm and 3nm? They should be ramping up. I just hope that Intel can add some foundry business to that so they stay full. Intel 3/4nm are good solid nodes. Where is the NOT TSMC business here? My concern is that the Intel Foundry sales team are mostly Intel sales people and not foundry sales people. If I was Lip-Bu I would re energize the sale team/replace them. The process nodes are ready, now the rubber hits the road.

The best sales management I worked with was Mike Ello and Ravi Subramanian at Berkeley Design. They asked the hard questions and spent a lot of time in front of customers. There was no room for excuses and sales nonsense, it was a blunt and brutal and well documented fact finding process. It started out with "what do we need to do to get your business?" and generally ended with a purchase order or a new salesperson. Mike is CEO of Siemens now and Ravi is at Chief Product Management Officer at Synopsys. I can assure you they both run a tight ship. Intel Foundry needs to be a very tight and aggressive ship.

 

[MKWVentures]

Ireland is 4nm and 3nm? They should be ramping up. I just hope that Intel can add some foundry business to that so they stay full. Intel 3/4nm are good solid nodes. Where is the NOT TSMC business here? My concern is that the Intel Foundry sales team are mostly Intel sales people and not foundry sales people. If I was Lip-Bu I would re energize the sale team/replace them. The process nodes are ready, now the rubber hits the road.

The best sales management I worked with was Mike Ello and Ravi Subramanian at Berkeley Design. They asked the hard questions and spent a lot of time in front of customers. There was no room for excuses and nonsense, it was a blunt and brutal fact finding process. It started out with "what do we need to do to get your business?" and generally ended with a purchase order or a new salesperson. Mike is CEO of Siemens now and Ravi is at Chief Product Management Officer at Synopsys. I can assure you they both run a tight ship. Intel Foundry needs to be a very tight and aggressive ship.

Ireland F34 is supposed to be ramping up... It is Intel 3.... I think there is some question whether Intel 4 even has volume but I will leave that for Intel people to answer.. It is quite underloaded compared to commits to Apollo and plans. It should be the last finfet process and should have great opportunities for new products for a long time. The Fab is now 18months after initial production shipments so the technology should be mature.

I think your "not-TSMC" naming is perfect (I use it often giving you full credit). But the question is who is "not-TSMC" crowd. Is TSMC turning people away? is TSMC pricing much higher than Intel to some customers? Is TSMC refusing to partner with some people.

In a weird Irony, I talked to a small company who wanted to do a advanced process chip in partnership with a foundry. They said Intel was not really interested due to volume and that TSMC was more interested. Just anecdotal .... but an Intel foundry friend did not seem too surprised when I told him/slammed him about the comment.

 

[Daniel Nenni]

Ire

Ireland F34 is supposed to be ramping up... It is Intel 3.... I think there is some question whether Intel 4 even has volume but I will leave that for Intel people to answer.. It is quite underloaded compared to commits to Apollo and plans. It should be the last finfet process and should have great opportunities for new products for a long time. The Fab is now 18months after initial production shipments so the technology should be mature.

I think your "not-TSMC" naming is perfect (I use it often giving you full credit). But the question is who is "not-TSMC" crowd. Is TSMC turning people away? is TSMC pricing much higher than Intel to some customers? Is TSMC refusing to partner with some people.

In a weird Irony, I talked to a small company who wanted to do a advanced process chip in partnership with a foundry. They said Intel was not really interested due to volume and that TSMC was more interested. Just anecdotal .... but an Intel foundry friend did not seem too surprised when I told him/slammed him about the comment.

I would expect that Intel 4nm and 3nm are similar so 4nm fabs can be fitted for 3nm?

I think the NOT TSMC market is bigger companies who want a second source. Not just for the optics, for better negotiating with TSMC. At 3nm you had no choice because Intel Foundry 3nm and Samsung 3nm were not ready at the first design time. There was no negotiating opportunity. Today you have Intel 3nm and Intel 18a ready and waiting. It is time for the big companies to establish viable second source leading edge semiconductor manufacturing supply chain here in the US.

 

[ChrisGar]

In a weird Irony, I talked to a small company who wanted to do a advanced process chip in partnership with a foundry. They said Intel was not really interested due to volume and that TSMC was more interested. Just anecdotal .... but an Intel foundry friend did not seem too surprised when I told him/slammed him about the comment.

I don't think this is surprising because the overhead for TSMC to add a new/small customer isn't large. There is an ecosystem (including design methodologies & IP) and clear design & engagement rules.

At this point, I think Intel would have to babysit each customer ... provide them IP that may be new/immature ... and not have a well-defined design methodology (that has been proven across many designs).

 

[Daniel Nenni]

I don't think this is surprising because the overhead for TSMC to add a new/small customer isn't large. There is an ecosystem (including design methodologies & IP) and clear design & engagement rules.

At this point, I think Intel would have to babysit each customer ... provide them IP that may be new/immature ... and not have a well-defined design methodology (that has been proven across many designs).

Intel Foundry is establishing trust and that will pay off with big customers who return for big runs on 18A or 14A. So the focus should be on big customers that will fill the fabs versus small ones. Also, TSMC has an established process for emerging/small companies that Intel cannot compete with at the moment, my opinion.

 

[nghanayem]

Ireland is 4nm and 3nm? They should be ramping up. I just hope that Intel can add some foundry business to that so they stay full. Intel 3/4nm are good solid nodes. Where is the NOT TSMC business here?

Intel has time and again talked about how the design ecosystem, PDK, EDA, IP, and design ease of use where never intended for foundry on Intel 3, because they started development YEARS before foundry was even an idea in anybody's head. Heck just last week they were mentioning all of the hardship of reworking what would go onto be 18A into a foundry process, and that it won't be until 18A-P that Intel has something that is more flexible in its usecases. Also Intel announced their foundry ambitions in 2021 (1.5 years before i4/3 HVM start). If an Intel 3 customer was landed the day Pat became CEO (before Intel even redefined their process roadmap and retrofited what would go onto be Intel 3 and 18A to be foundry offerings) the earliest an Intel 3 foundry product could exist is 2025. Since an Intel 3 foundry customer wasn't landed until 2023, then 2027 minimum for first external Intel 3 designs. And probably even longer than that because of the previosuly mentioned issues.

My concern is that the Intel Foundry sales team are mostly Intel sales people and not foundry sales people.

That so absurd that it boggles the mind. I can't believe any company would be that incompetent. Looking at Samsung and GF, Intel seems to be landing customers faster than any other upstart foundry in this century (10+ advanced packaging, 1 Intel 16, 1 Intel 3, and 9 18A inside of 3 years). Now is that fast rate of acquisition because fabless customers are that hungry for an alternative, the HPC boom is working in Intel's favor, because Intel has been doing a pretty good job making this foundry transition, or some combination of factors; I couldn't tell you. Either way, I don't think Intel has a sales issue dragging them down. Realistically I don't think things could have realistically gone much better. Maybe Intel landed another whale or two if the 18A PDK 1.0 was done 6mo earlier, but that is the only major flub I can think of that doesn't rely on hindsight. It just takes time to build trust, time to make foundry compatible process technologies, time to retake their mantle as the technological heartbeat of semiconductor manufacturing, and 4-6 years for existing wins to translate into balance sheet results. To expect Intel foundry to be number 2 before 5N4Y was even finished was never going to be in the cards.

 

[nghanayem]

Intel Foundry is establishing trust and that will pay off with big customers who return for big runs on 18A or 14A. So the focus should be on big customers that will fill the fabs versus small ones. Also, TSMC has an established process for emerging/small companies that Intel cannot compete with at the moment, my opinion.

Even TSMC didn't get the big customers without first proving themselves with small designs.

 

[MKWVentures]

@nghanayem
"Intel seems to be landing customers faster than any other upstart foundry in this century (10+ advanced packaging, 1 Intel 16, 1 Intel 3, and 9 18A inside of 3 years)"

I this real? where is it listed that Intel has 9 18A customers. The metric should be product tapeouts by year and revenue. When do we start seeing any of these?

Side note/stupid question: When companies are taking wafers from Intel for checkout. Are they paying for them? I assume they pay for any wafer on a tapeout they did? I am trying to make sense of the insanely low revenue for Intel.

 

[count]

My own personal experience is that once a mega project like this stalls out, it is very rare that it can be resurrected.

I have seen multibillion dollar project abandoned, because the economic conditions that justify investments of that scale are typically a moment in time, and if you fail to seize that moment it will pass.

 

[nghanayem]

@nghanayem
"Intel seems to be landing customers faster than any other upstart foundry in this century (10+ advanced packaging, 1 Intel 16, 1 Intel 3, and 9 18A inside of 3 years)"

I this real? where is it listed that Intel has 9 18A customers.

Semiconductor Manufacturing Process | Intel 18A, 3, and 16

See how Intel is enabling tomorrow's cutting-edge semiconductor manufacturing process technologies to deliver a systems foundry for the AI era.

www.intel.com

Customers are confidently choosing Intel Foundry. The world’s two largest cloud-service providers have announced products using Intel 18A technology, part of nine total announced Intel 18A awards.

The metric should be product tapeouts by year

Why? It takes years to go from signing a wafer agreement to being mostly done with the design and tapping in your A0 step. Just over one year ago Microsoft said they recently started work to design an AI ASIC for 18A. It isn't a failure that Microsoft hasn't taped out yet, because there is no universe where their design is anywhere near done. Nobody is jumping down TSMC's throat because Apple started first N2 design work in 2020 and only tapped out in 2024 with HVM due near the end of 2025. There were some announcements of DoD chips before, but the first noteable 18A design win came in like fall of 2023 with big enough volumes to apparently justify a prepay. Even if we are talking best in world fabless design with A0 PRQ the earliest possible HVM on that is 27. The only fault of Intel's is not starting down the path of becoming a foundry years earlier so 18A foundry readiness and the ecosystem could have been ready closer to the Intel products ramp.

and revenue. When do we start seeing any of these?

Revenue is a trailing indicator and tells you nothing about momentum, growth, or Intel foundry's ability to deliver towards their objective of external foundry alone being bigger than Samsung internal plus external by 2030 with average foundry margins.

Side note/stupid question: When companies are taking wafers from Intel for checkout. Are they paying for them?

Why would they pay for them? I would think test wafers are provided because the cost will always be negligible compared to the value of a sale. After all there is no TSMC N2 or A14P revenue and I can garuentee you they are making plenty of customer or ecosystem test chips and plenty of customer tape outs.

I assume they pay for any wafer on a tapeout they did? I am trying to make sense of the insanely low revenue for Intel.

I mean the only customers they have for revenue right now is a small list of advanced packaging customers. The packaging substrate sales during the COVID capacity crunch are done, IMS earnings are no longer part of it, Tower said they are only going to start using NM fab11 for 65nm production in mid 2025, and Mediatek only just did their A0 qualification at end of last year (per themselves at IFS DC). And honestly if that is all they are doing, making as much money as Tower (who has 4 fabs and 3 manufacturing joint ventures) did in 2024 with just a couple of advanced packaging wins seems fairly solid.

But even if they change for test chips (customer or ecosystem) or customer tapeouts for the real products, I doubt that really changes the story. When we are talking $30M+ per quarter, a few wafers for customer A, a few for customer B, maybe a half lot for synopsis, etc. I can't imagine would ever really add up to anything substantial compared to the advanced packaging second sourcing Intel is doing.

 

[sit]

That so absurd that it boggles the mind. I can't believe any company would be that incompetent.

Yep, that be Intel. Even more confusing is the role of Sales/Marketing vs. Foundry Services.

 

[Daniel Nenni]

Looking at Samsung and GF, Intel seems to be landing customers faster than any other upstart foundry in this century (10+ advanced packaging, 1 Intel 16, 1 Intel 3, and 9 18A inside of 3 years).

I must have missed this announcement. At the Intel event only one foundry customer was named and that was MediaTek on Intel 16. I spoke to Intel directly about this as well.

I am wondering about the cozy relationship with MediaTek. They are a very Taiwanese company and the CEO is the former TSMC CEO. Maybe there is some bad blood there? But again, if you want good TSMC pricing sing the praises of Intel Foundry.

 

[ChrisGar]

Why would they pay for them? I would think test wafers are provided because the cost will always be negligible compared to the value of a sale. After all there is no TSMC N2 or A14P revenue and I can garuentee you they are making plenty of customer or ecosystem test chips and plenty of customer tape outs.

The mask cost of test wafer are significant. Common industry practice is the customer would have to pay for custom test chips. (but of course, this is all negotiable -- Intel could be cutting a deal)