Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/news-samsung-reportedly-reviews-chip-business-considers-halting-pyeongtaek-and-u-s-taylor-investments.22248/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

[News] Samsung Reportedly Reviews Chip Business, Considers Halting Pyeongtaek and U.S. Taylor Investments

XYang2023

Well-known member

2025-03-10 Semiconductors editor


News
samsung-taylor-texas-office-624x413.jpg


According to a report from ijiwei, citing The Korean Economic Daily, Samsung Electronics has been examining its chip design and foundry businesses closely, considering a potential restructuring. The review may include an executive reshuffle and workforce reallocation, as the report notes.

The report suggests that this business review marks Samsung’s first major internal audit since the establishment of the Management Diagnosis Office in November in 2024.

As reported by The Korean Economic Daily, sources suggest that Samsung’s Management Diagnosis Office launched a comprehensive evaluation of its System LSI division, which focuses on chip design, in January. Following the completion of this assessment, the office is expected to extend its scrutiny to the company’s foundry operations.

Samsung’s business review is anticipated to bring substantial changes to the System LSI division. The report mentions that, citing sources, Samsung is considering transferring its Exynos system-on-chip (SoC) business from the System LSI division to the Mobile eXperience (MX) division to better align with Samsung’s overall smartphone strategy.

Regarding its foundry business, Samsung plans to evaluate the feasibility of suspending investments in its Pyeongtaek plant in Korea and Taylor plant in the U.S. The company’s $37 billion US fab in Taylor, Texas, has been delayed, with production now set for 2026 instead of its original timeline, the report adds.

As noted by the report from The Korean Economic Daily, with TSMC recently unveiling a USD 100 billion expansion plan in the U.S., analysts cited in the report express growing concerns over whether Samsung can secure enough major clients to justify its investment in the U.S..

Samsung’s foundry business posted more than 2 trillion won (USD 1.4 billion) in operating losses in the fourth quarter of 2024, as noted by The Korean Economic Daily.

According to TrendForce, Samsung Foundry remained the second-largest foundry in the third quarter of 2024, holding just a 9.3% market share, while TSMC dominated with nearly 65%.

 
Last edited by a moderator:
The problem with Samsung SoCs: you can't buy them.

They used to be the more available at the market when there was like 20+ phone SoCs makers on the market, but then at the first signs of success, their sales turned lazy. If you you are not a billion dollar business, they will not even talk to you.
 

Samsung’s system chip, foundry business under close scrutiny for overhaul​


A researcher in a Samsung Electronics chip cleanroom

A researcher in a Samsung Electronics chip cleanroom

Samsung Electronics Co., the world’s largest memory chipmaker, has put its system chip and foundry businesses under tight scrutiny for a possible business overhaul, including an executive reshuffle and employee relocation, as it faces an uphill battle against rivals such as Taiwan’s TSMC.

The move underscores the urgency of bolstering competitiveness in a sector that has been the key focus of Chairman Lee Jae-yong’s long-term strategy but has struggled to gain ground against industry leaders, analysts said.

The South Korean tech giant’s Management Diagnosis Office in January began a thorough business review of its System LSI division, which specializes in chip design, people familiar with the matter said on Thursday.

Once a careful review of the division is over, the office plans to look into the firm’s foundry, or contract chipmaking, business, sources said.

(Graphics by Dongbeom Yun)
zoom_icon.png

(Graphics by Dongbeom Yun)

The business review marks Samsung’s first major internal audit since the company launched the office last November as part of its groupwide efforts to revive its struggling businesses.

Sources said the review signals a critical evaluation of Samsung’s strategy in non-memory semiconductors – an area the chairman, known as Jay Y. Lee in the international community, designated as a future growth engine back in 2019 under Samsung’s “System Semiconductor Vision 2030” initiative.

PERFORMANCE CONCERNS

While Samsung has made substantial investments in its system semiconductor business over the past few years, results have been mixed.

Its Exynos 2500 mobile processor failed to secure a spot inside the flagship Galaxy S25 smartphone, launched in January.

A Samsung employee examines a chip wafer

A Samsung employee examines a chip wafer

Samsung’s image sensors, the global market share of which is less than 20%, continue to lag behind Japan’s Sony.

The Korean chipmaker’s foundry business has struggled for years to compete with Taiwan Semiconductor Manufacturing Co. (TSMC), which maintains a dominant position in high-end semiconductor manufacturing.

As of the fourth quarter of 2024, Samsung, the world’s No. 2 foundry player, had 8.2% market share, compared with TSMC’s 67.1%.

The lack of progress has prompted Samsung to take a deeper look at its operations.

The management audit office, led by former Samsung SDI Co. CEO Choi Yoon-ho, has been tasked with evaluating the challenges facing Samsung Electronics’ key businesses and identifying strategies for improved competitiveness.

The office’s involvement is viewed by industry watchers as a step toward restoring a central corporate control structure akin to the former Samsung Future Strategy Office, which was dismantled in 2017.

TSMC is the world's largest foundry player

TSMC is the world's largest foundry player

HEADWINDS

A key challenge for the System LSI division has been its inability to expand its customer base beyond the company’s own divisions – a situation highlighted by the Exynos 2500 mobile application processor’s exclusion from the Galaxy S25.

Similarly, while Samsung has made inroads in automotive semiconductors by supplying automotive Exynos chip to Hyundai Motor Co. for the automaker’s infotainment system, it lost a key bid to Qualcomm for Hyundai’s self-driving chip supply contract.

In image sensors, Chinese smartphone makers, including Oppo, Vivo and Xiaomi, have increasingly turned to domestic suppliers, limiting Samsung’s market expansion opportunities.

Samsung has also faced difficulties in developing custom AI semiconductors – a critical area of future growth.

Its joint AI accelerator project with Naver Corp., known as Mach-1, has nearly stalled, and a second initiative, Mach-2, targeting Microsoft and other US Big Tech firms, has yet to yield results.

These setbacks have raised concerns about Samsung’s ability to compete with specialized players such as Broadcom and Marvell in the application-specific chip market.

Samsung's flag seen outside its headquarters in Seoul

Samsung's flag seen outside its headquarters in Seoul

FOUNDRY BUSINESS

Samsung’s foundry division, also set to undergo a management review, has been struggling with low production yields in advanced manufacturing nodes.

The company was the first to adopt gate-all-around (GAA) transistor technology in its 3-nanometer process, but low product yields have prevented it from securing major customers.

Samsung’s Shell-First Strategy, a custom-tailored new operation strategy, is also struggling as the chipmaker has failed to secure enough foundry clients.

Samsung’s foundry business posted more than 2 trillion won ($1.4 billion) in operating losses in the fourth quarter of 2024.

The company’s $37 billion investment in a US fab in Taylor, Texas, has also faced delays, with production now pushed back by two years to 2026.

With TSMC recently announcing a $100 billion expansion plan in the US, concerns are mounting over whether Samsung can attract enough high-profile customers to justify its investment, analysts said.

Samsung faces an uphill battle in advanced AI chip competiton

Samsung faces an uphill battle in advanced AI chip competiton

ORGANIZATIONAL RESTRUCTURING, PERSONNEL RESHUFFLE

Samsung’s business review is expected to lead to significant changes within the System LSI division.

One possibility under consideration is handing over the Exynos system-on-chip (SoC) business from the System LSI division to the Mobile eXperience (MX) division, aligning it more closely with Samsung’s smartphone strategy, sources said.

In the image sensor segment, the company may pivot from competition for high-resolution products and focus instead on products tailored for autonomous vehicles and robotics, they said.

Samsung’s System LSI recently set up a team tasked with securing a deal to supply image sensors to Apple Inc. as early as next year.

Samsung Electronics' chip cleanroom

Samsung Electronics' chip cleanroom

For the foundry business, Samsung plans to assess the feasibility of its paused investments in its Korean plant in Pyeongtaek as well as Taylor.

The chipmaker also aims to work out a strategy to improve product yields in advanced nodes and secure AI chip manufacturing contracts, sources said.

To achieve these goals, Samsung may reshuffle executives and researchers at its System LSI division, including foundry workers, they said.

 
Samsung continues to make TSMC look brilliant. Samsung even makes Intel look good even though both missed AI. How does South Korea miss AI?!?!?!
 
Samsung continues to make TSMC look brilliant. Samsung even makes Intel look good even though both missed AI. How does South Korea miss AI?!?!?!

It did not miss AI, it consistently missed to reach cost parities in all its <40nm nodes, thus it has many unprofitable legacy fabs.

And it did it because Lee Jr. always bet on trying to take the leading position regardless of cost. "I want to be the new Iphone, or bust", and thus it's bust.

TSMC never chased the leader, it chased revenues.
 
TSMC never chased the leader, it chased revenues.

This is true. Intel and Samsung are IDM foundries. They grew up with a "build fabs and customers will come" culture. TSMC builds based on customer forecasts and pre pays.

Hynix isn’t missing it. The new Korean darling for now

Exactly. How does the number one semiconductor memory making country miss AI logic?!?!?!?
 
It did not miss AI, it consistently missed to reach cost parities in all its <40nm nodes, thus it has many unprofitable legacy fabs.

And it did it because Lee Jr. always bet on trying to take the leading position regardless of cost. "I want to be the new Iphone, or bust", and thus it's bust.

TSMC never chased the leader, it chased revenues.
Well TSMCs cautious approach is one good reason they are what they are it allows customer to build their Biz on them it reduces uncertainty for customers.

Leadership comes with risk TSMC doesn't take risk.
 
Hynix isn’t missing it. The new Korean darling for now

Memory prices are rocking to the bottom, and both Hynix, and Samsung have retreated to enterprise SSD market, where they have high contractual pricing, and lock on very high capacity controllers (no one in enterprise buys mainland).

But both Hynix, and Samsung are significantly lagging technologically with latest generation memory exactly because they got lazy living off safe enterprise SSD sales. Samsung V9, and Hynix V9 are both late, and not bringing as much performance upgrade vs. their previous generational advances.
 
Last edited:
what process does Samsung run in the US now and about how many WSPM?

Remember when people were worried we wouldn't have enough people for all the US fabs? (Intel, Samsung, Micron, TSMC) This will be remembered as the Era of the mythical Fab in the US.

Can you put pickleball courts into an abandoned Fab Shell? Cannabis farms? Wegovy/Ozempic manufacturing plant? :ROFLMAO::LOL:
 
Samsung has successfully employed “chicken games” in the past, but this strategy only works when its technology is leading and costs are the lowest.

In the foundry business, TSMC holds the technological lead and maintains lower costs. For other sectors such as LCDs, NAND, and smartphones, the “chicken game” strategy is less effective, as Chinese companies backed by the CCP are not easily intimidated.

DRAM is the foundation of Samsung’s business. Samsung should deprioritize its foundry business reassign its top talent back to DRAM to secure its future.
 
Chinese companies backed by the CCP are not easily intimidated.

No, they just go bust, or keep making substandard low yield wafer at loss, never progressing further, if their product is so significantly behind the market.

Nobody would buy a 3 generation old NAND chip, even if sold at a super deep discount.

This is why I keep telling that all Western Block semis made monumental mistake chasing low volume, niche, high profit stuff, when commodity gets eaten by subsidised, volume oriented followers from communist countries, where subsidy doesn't need to be that huge to tip the scales.

If Samsung, TSM, GF had sane prices, and high capacity on <40nm, mainland semis would have nothing to eat. They wouldn't have had enough commercial clients to kickstart the self-sustaining business, to keep with node downscaling.

Node discount with age is basically an auction for who can pay the most for wafer for a given performance-price ratio.
 
Last edited:
what process does Samsung run in the US now and about how many WSPM?

Remember when people were worried we wouldn't have enough people for all the US fabs? (Intel, Samsung, Micron, TSMC) This will be remembered as the Era of the mythical Fab in the US.

Can you put pickleball courts into an abandoned Fab Shell? Cannabis farms? Wegovy/Ozempic manufacturing plant? :ROFLMAO::LOL:
Samsung runs 14nm derivatives in Austin and I believe it’s a fairly large fab.

 
Samsung Foundry operates in the US in Austin with a single high volume 100,000 WSPM fab. Leveraging first gen 14nm FINFETs as well as older planar technologies.
There is a dividing line in the foundry market with Samsung Austin, and Samsung generally, on the low-volume, niche side. A single-fab site makes you a niche player, and not really part of the Foundry big-leagues.
Samsung Austin has a niche group of customers, in RF, chipsets, automotive, with a largely mobile focus. Like Global Foundries, Samsung Austin is profitable but business is down due to working through inventories from the pandemic era. There is enough business within the niche, but there is a need to come to grips with ambitions vs. reality.
The Austin fab is 29 years old, with a few gap years here and there for conversion from 8 inch to 12, and from memory to logic. The fab expanded multiple times. While Taylor is the future, Austin produces wafers at a competitive cost, with world-class yields, on many mature nodes.
 
Samsung Foundry operates in the US in Austin with a single high volume 100,000 WSPM fab. Leveraging first gen 14nm FINFETs as well as older planar technologies.
There is a dividing line in the foundry market with Samsung Austin, and Samsung generally, on the low-volume, niche side. A single-fab site makes you a niche player, and not really part of the Foundry big-leagues.
Samsung Austin has a niche group of customers, in RF, chipsets, automotive, with a largely mobile focus. Like Global Foundries, Samsung Austin is profitable but business is down due to working through inventories from the pandemic era. There is enough business within the niche, but there is a need to come to grips with ambitions vs. reality.
The Austin fab is 29 years old, with a few gap years here and there for conversion from 8 inch to 12, and from memory to logic. The fab expanded multiple times. While Taylor is the future, Austin produces wafers at a competitive cost, with world-class yields, on many mature nodes.
I wouldn't exactly call Intel chipsets niche or low volume, but that is probably a reasonable charcterization of a lot of the Austin customers beyond Intel after Qualcomm, Apple, and NVIDIA moved to more advanced Samsung and TSMC processes.

Funny to think that after Qualcomm abandoned ship after SF5E and the confusingly named 4LPX (which was related to SF5 family but not SF4 family) that Intel is likely Samsung's largest external foundry customer by a wide margin.

On an unrelated note does anyone know if Samsung offers their CMOS image sensor process technology to fabless image sensor houses to compete with Tower and TSMC? Or are they more like OnSemi, SK-HYNIX, and Sony and their image sensor process technology is only for the internal design house? When I have looked through their website in the past I haven't seen any mention of it. But I feel like I might have seen something about it from early early Samsung foundry days.
 
Back
Top