GlobalFoundries downgraded by Citi amid competition - downturn concerns

[Daniel Nenni]

"While we expect the handset business (41% of 4Q23 sales) to rebound in 2024, we believe the auto business (17% of 4Q23 sales) could slow given excess inventory in the channel."

On Wednesday, GlobalFoundries Inc. (NASDAQ:GFS) experienced a downgrade in its stock rating by Citi from Buy to Neutral. The semiconductor company also saw its price target reduced to $56.00 from the previous $70.00. The adjustment comes as the company reported financial results that aligned with expectations but issued guidance significantly below the general market consensus.

The downgrade reflects concerns about the impact of the ongoing industry downturn on GlobalFoundries. According to Citi, the challenges facing the company are expected to persist, potentially delaying a return to peak earnings per share (EPS) for several years. The firm anticipates that GlobalFoundries will face increased competition, which could lead to pressure on both pricing and revenue growth.

GlobalFoundries, known for its specialized semiconductor manufacturing services, has been navigating a shifting market landscape. The company's recent guidance suggests that it is beginning to feel the effects of the broader industry slowdown, which has been a point of concern for investors and market watchers alike.

The reduced price target of $56.00 represents a notable decrease from the previous target, indicating a shift in expectations regarding the company's stock performance. Citi's analysis suggests that it may take GlobalFoundries some time to recover and achieve the level of profitability seen at its peak.

The news of the downgrade and the lowered price target comes as investors continue to monitor the semiconductor sector, which has been facing multiple headwinds. These include the cyclical nature of the industry, evolving demand patterns, and the competitive landscape, which now appears to be impacting GlobalFoundries' outlook.

Citi cuts GlobalFoundries stock rating amid downturn concerns By Investing.com

Citi cuts GlobalFoundries stock rating amid downturn concerns

www.investing.com

 

[Daniel Nenni]

I still think Samsung should buy GF. It would change the face of Samsung Foundries (for the better) and get Samsung access to a whole new customer base. GF 14nm is licensed from Samsung so there is no problem there. It will give Samsung a bigger world wide foot print and Samsung has the cash to do it.

Samsung really needs to change the foundry culture of "first to a new technology and last to yield".

I no longer think Intel should buy GF. The Intel deals with UMC and Tower seem to be the roadmap. Buying GF would not help that at all.

Thoughts?

 

[blueone]

I still think Samsung should buy GF. It would change the face of Samsung Foundries (for the better) and get Samsung access to a whole new customer base. GF 14nm is licensed from Samsung so there is no problem there. It will give Samsung a bigger world wide foot print and Samsung has the cash to do it.

Samsung really needs to change the foundry culture of "first to a new technology and last to yield".

I no longer think Intel should buy GF. The Intel deals with UMC and Tower seem to be the roadmap. Buying GF would not help that at all.

Thoughts?

Sounds like a good idea.

 

[nghanayem]

I still think Samsung should buy GF. It would change the face of Samsung Foundries (for the better) and get Samsung access to a whole new customer base. GF 14nm is licensed from Samsung so there is no problem there. It will give Samsung a bigger world wide foot print and Samsung has the cash to do it.

Interesting idea. Foundry culture wise GF definitely seems ahead of Samsung. Technologically speaking I don't know if that would do a whole lot for Samsung. Due to the unique needs of their IDM they have a good portfolio of value add technology for their lagging nodes. The way the report things makes it hard to exactly tell, but it also seems like Samsung's external foundry business is predominantly (at least from a wafer starts perspective) using the old DRAM/logic fabs to mostly spit out specialty versions of these old nodes. If Samsung already has this portfolio and the old fabs in Korea and Texas spitting out these trailing products, then I don't think GF has much if anything to offer Samsung. Presumably there has to be a better/cheaper way of redefining their foundry culture than buying GF and putting TC in charge of the whole operation? But I won't pretend that I have the business sense to know what that "better way" is.

Samsung really needs to change the foundry culture of "first to a new technology and last to yield".

I think that is an execution issue rather than a strategic issue. If 3GAE launched real products in 2022 with QCOM and the like running real products on 3GAP in 2023, then I think Samsung could get away with PDKs that are rough around the edges and yields that were "meh". On the leading edge there are definitely folks who I think would be capable or willing to put up with that if they could have a narrow window of power-performance parity or maybe even a slight lead over Apple (even if they had to give up density and design ease of use to do it). Similar deal with 7LPP. If 7LPP was in 2018 then Samsung could have simpler design rules and similar PPA (at the cost of cycle time and capacity). Then on N+2 nodes like 8LPP when they get to mature yields and have the PDKs more polished they can always leverage their unique position to undercut TSMC.

Of course the problem is you aren't getting meh yields and a narrow window of process performance parity with Apple so the whole thing falls apart. I'm curious what others think, but I think the advantage that Samsung could offer to their customers if they were first to a technology with reasonable yields could work. While I know it isn't a perfect example, but I think of something like intel 90nm. Even if intel was a year later than ROW rather than a year early, intel 90nm would have still been a really awesome node just because they were years ahead of the rest when it came to uni-axial strain (IBM and co. would only be able to do bi-axial at 65nm and not have uni until 45nm).

 

[Daniel Nenni]

I remember when 14nm came out, Samsung had a very good process and won many customers including some Apple business. Samsung then had a very bad experience with 10nm yield and customers went to TSMC. Samsung then came back with a strong 7nm node but 5nm was again a challenge. Samsung 3nm is another bad node. Today TSMC has picked up all of the key customers at N3 and it looks like N2 will be the same.

Given the past, and now with Intel in the foundry mix, my advice to Samsung is to change up your game plans because the status quo will not work. Acquiring GF is one way, an expensive way, but it would be the quickest.

 

[Daniel Nenni]

Semiconductor industry ‘needs another quarter or so’ to bring inventory down: GlobalFoundries CEO​

https://www.cnbc.com/video/2024/02/13/semiconductor-industry-needs-another-quarter-or-so-to-bring-inventory-down-globalfoundries-ceo.html

 

[coldsolder215]

I don't see how GF's culture of "new technology is overrated and why is everyone leaving" can help anyone in 2023. Just set up hiring booths at the exits of any one of their ancient facilities, no need for the headache of acquiring those facilities with that management. Their black pilled lawfare against TSMC, IBM, and by extension the entire industry indicate they can no longer monetize their IP by manufacturing it. They had something to offer 5 years ago but conditions weren't ripe before the collapse of Intel/Samsung relative to TSMC and the new yellow peril underpinning CHIPS. Maybe if our politicians weren't all sundowning libertarians fixated on lost wars and insider trading, they could've stepped in and turned it into an IMEC-like entity to seed the CHIPS initiative. Instead, government bureaucrats are doing it from scratch and there's nothing left for GF but to let it wither and part it out for pennies on the dollar.

 

[delong.height]

I still think Samsung should buy GF. It would change the face of Samsung Foundries (for the better) and get Samsung access to a whole new customer base. GF 14nm is licensed from Samsung so there is no problem there. It will give Samsung a bigger world wide foot print and Samsung has the cash to do it.

Samsung really needs to change the foundry culture of "first to a new technology and last to yield".

I no longer think Intel should buy GF. The Intel deals with UMC and Tower seem to be the roadmap. Buying GF would not help that at all.

Thoughts?

yeah, me too. But I think there will be a samsung-intel union at sometime in future. Not sure if I'm biased, I feel like Samsung is like old intel. It slowly but surely falling behind in advancing new process tech.

 

[Barnsley]

I don't see how GF's culture of "new technology is overrated and why is everyone leaving" can help anyone in 2023. Just set up hiring booths at the exits of any one of their ancient facilities, no need for the headache of acquiring those facilities with that management. Their black pilled lawfare against TSMC, IBM, and by extension the entire industry indicate they can no longer monetize their IP by manufacturing it. They had something to offer 5 years ago but conditions weren't ripe before the collapse of Intel/Samsung relative to TSMC and the new yellow peril underpinning CHIPS. Maybe if our politicians weren't all sundowning libertarians fixated on lost wars and insider trading, they could've stepped in and turned it into an IMEC-like entity to seed the CHIPS initiative. Instead, government bureaucrats are doing it from scratch and there's nothing left for GF but to let it wither and part it out for pennies on the dollar.

Are the non-ex Chartered Semi FABS in GF having the same high handed management, communication skills, as those ex-Chartered Semi FABS Singapore?

Its slowly changing here , now again there communication go back to old way when they were Govt entity and like to speak to suppliers like they something you trod in

 

[cliff]

GF has received government funding. There is no way they should be allowed to merge with Intel, TSMC, Samsung, UMC, nor even the suppliers ASML, AMAT, etc.

GF and Micron are one of the few companies that should get funding... and they need to stay in the US.

What's wrong with 14nm?

 

[Daniel Nenni]

GF has received government funding. There is no way they should be allowed to merge with Intel, TSMC, Samsung, UMC, nor even the suppliers ASML, AMAT, etc.
GF and Micron are one of the few companies that should get funding... and they need to stay in the US.
What's wrong with 14nm?

GF 14nm was licensed from Samsung. The other processes GF has came from elsewhere as well. South Korea could very well need a silicon shield so Samsung investing heavily in the US would not be a bad thing. I do see your point, GF/IBM has close ties to the US Government which could be a problem. Not insurmountable though. By the way, GF is still majority owned by the United Arab Emirates, correct? Not really a US company.

 

[cliff]

GF 14nm was licensed from Samsung... so what? They have an operating facility now with a great process, correct? Why is Japan funding 28nm and 16nm?

Why is Russia making 130nm?

Why was India begging for 28nm?

Majority owned by the United Arab Emirates.. very disturbing and must be dealt with. This should not be allowed if US tax $$$ go to GF.

I am disturbed by losers and critics. We should create. I totally agree with their CEOs vision and expanding on 14nm and SIPs and exotic additions (MRAM, MEMs).

Create. Build. Be engineers, not quitters.

 

[cliff]

As far as UAE ownership, I thought GF went public to reduce the UAE holding percentage.

Edit: I could argue that GF14 should be split off to being another company. The 14nm facility should not be allowed to merge or be run by any entity run by people outside of the US.

 

[Daniel Nenni]

GF 14nm was licensed from Samsung... so what? They have an operating facility now with a great process, correct? Why is Japan funding 28nm and 16nm?
Why is Russia making 130nm?
Why was India begging for 28nm?
Majority owned by the United Arab Emirates.. very disturbing and must be dealt with. This should not be allowed if US tax $$$ go to GF.
I am disturbed by losers and critics. We should create. I totally agree with their CEOs vision and expanding on 14nm and SIPs and exotic additions (MRAM, MEMs).
Create. Build. Be engineers, not quitters.

GF has never been successful developing CMOS process technology. They followed IBM and Samsung at 28nm and failed. They tried to develop 14nm and 7nm and failed. The fabs they have were either acquired or the technology was licensed from a competitor. I don't know how their FD-SOI products are doing but they licensed that technology as well. So what has GF really created? Seriously, GF spent billions of dollars and this is what is left? GF did go public but it is still majority owned by a foreign government.

GF will do fine monetizing existing technologies but they are a boutique semiconductor foundry frozen in time. So why invest CHIPs Act money in GF? Micron yes, Intel yes. Micron should be first in line. We need memory technology in the US, absolutely.

 

[cliff]

Stopping 7nm early was smart. They recognized quadruple patterning was insane. Good move. Stick to double patterning.

They failed. Who doesn't initially. Have you ever developed anything? You keep going. You should not be encouraging quitting.

Do you think continuing to represent a nand gate with 80 FETs is a great idea? ASICs (boutique chips) at 14nm are far more efficient and faster than FPGAs

Frozen in time? Are you kidding me? They are adding exotic memory, correct? How about the advancement of packaging (interposers, etc) around this fantastic technology.

 

[nghanayem]

Stopping 7nm early was smart. They recognized quadruple patterning was insane. Good move. Stick to double patterning.

It was smart because they were late and subscale. There is no other reason to stop otherwise, you limit your TAM too much and are forced into low ASPs. Also their 7LPP didn’t use quad metals. Only intel has ever attempted that (on the logic side anyways since SAPQ isn’t too hard for memory applications). Like TSMC, GF stayed at the limits of double patterned metals for 7”nm”.

Do you think continuing to represent a nand gate with 80 FETs is a great idea? ASICs (boutique chips) at 14nm are far more efficient and faster than FPGAs

Same thing with an 8LPP or N7 ASIC while also smoking GF14 cost per fet and power performance…

Frozen in time? Are you kidding me? They are adding exotic memory, correct? How about the advancement of packaging (interposers, etc) around this fantastic technology.

What company isn’t doing 2.5D and MRAM? TSMC, fuji, Samsung, reneasas, skywarer, intel, UMC, they all have active development in those areas or full support on some of their portfolio. Heck in the case of fuji, reneasas, TSMC, and tower they already have hybrid bonding for 3DIC products on the market today. Even IDMs like Sony have played in the 3DIC space for over a decade.

 

[blueone]

ASICs (boutique chips) at 14nm are far more efficient and faster than FPGAs

On this point we are in agreement. Even at 28nm I suspect ASICs are superior to FPGAs, and the latest design tools make ASICs more approachable with small teams. But FPGAs can be programmed with software engineers in C++, which is an intoxicating attraction to some system designers. I think FPGAs have that special combination of high cost, high power consumption, high heat, slow logic compared to ASICs. They are great for prototyping and design debugging though. Financial companies and Microsoft disagree, and continue to use them for production systems.

 

[cliff]

Mr. Blue: Of course FPGAs are popular and needed, but so are ASICs.

Mr. Ng: Who is knocking 6nm? But why does every foundry need to be leading edge? If you aren't number 1, then you pack up and close? There are lots of customers that don't have $50M NRE budgets. Why is NXP, etc making chips on 16nm?

My point is that there is nothing wrong with 16-12nm processes. I believe GF's plan is a good practical plan and should stick with it. They should NOT be allowed to merge with Samsung, Intel, etc. No way.

 

[cliff]

As far as 28nm... correct, a 28nm ASIC will blow away an FPGA.

Note: Automation makes 28nm obsolete compared to 16-12nm. IanD made a good case for GF22, so we continue to support that.

 

[Fred Chen]

Isn't GlobalFoundries more focused now on its 22FDX SOI and 12nm FinFET?