Comparing Samsung, Intel, and TSMC

[Martian2]

There is a misconception among the public that size equals technological lead. This is not true in the case of Samsung, Intel, and TSMC.

Samsung's sales for this year are $65.6 billion.
Intel sales total $61 billion.
TSMC sales are about $30 billion.

Samsung is outspending Intel and TSMC combined in capital expenditure (capex) for this year.

This has led the mainstream media to erroneously conclude that Samsung dominates the semiconductor industry. This is an incorrect conclusion. Samsung has made no headway in manufacturing complex RISC-based logic chips for other companies.

For RISC-based logic chips, TSMC has dominated the worldwide market with 60% market-share.

Intel holds the patents for the x86 CISC-based logic chips.

Samsung dominates the commodity market for DRAM and NAND memory chips, which are the simplest to manufacture (due to the repeating storage elements).

For 2015, TSMC had $26.4 billion in sales from manufacturing logic chips. In contrast, Samsung only had $2.6 billion from manufacturing Exynos processors for its own smartphones. When you compare logic chip revenues, TSMC beats Samsung by 10-to-1.
----------

Samsung Electronics Recruited Foundry Experts from Competitors Yet Again | BusinessKorea

 

[Martian2]

Is Samsung afraid of Chinese DRAM and NAND competitors?

Look at Samsung's unbelievable $26 billion in capital expenditures (capex) for this year. See chart below from IC Insights.

During the previous seven years, Samsung spent an average of about $10 billion in capex per year. Why the massive jump in capex?

In my view, Samsung is trying to crush the new mainland Chinese DRAM and NAND companies. Samsung's strategy is pretty simple. Flood the market with cheap DRAM and NAND chips to cause financial losses for the Chinese memory chip companies.

I don't think Samsung's old playbook (of destroying the competition) will work this time. The Chinese companies are subsidized by the Chinese government. Chinese State-Owned-Enterprises (SOEs) are big buyers of DRAM and NAND memory chips. The Chinese government can always pressure the SOEs to buy some Chinese-manufactured memory chips. While the Chinese memory-chip companies won't make any money, they also won't go bankrupt.

This kind of aggression from Samsung will slow down the progress of the Chinese memory-chip companies. However, in the long term (5 to 10 years), I still think the Chinese companies will start taking market-share away from Samsung in the critical Chinese market.

"IC Insights estimates that Samsung’s $26 billion in semiconductor outlays this year will be segmented as follows:

3D NAND flash: $14 billion (including an enormous ramp in capacity at its Pyeongtaek fab)
DRAM: $7 billion (for process migration and additional capacity to make up for capacity loss due to migration)
Foundry/Other: $5 billion (for ramping up 10nm process capacity)"
----------

Will Samsung’s 2017 Semi Capex Deliver Knockout Blow to Competition? | IC Insights (November 14, 2017)

"Samsung’s current spending spree is also expected to just about kill any hopes that Chinese companies may have of becoming significant players in the 3D NAND flash or DRAM markets. As our clients have been aware of for some time, IC Insights has been extremely skeptical about the ability of new Chinese startups to compete with Samsung, SK Hynix, and Micron with regards to 3D NAND and DRAM technology. This year’s level of spending by Samsung just about guarantees that without some type of joint venture with a large existing memory suppler, new Chinese memory startups stand little chance of competing on the same level as today’s leading suppliers." (second to last paragraph)

 

[Martian2]

Chinese Memory Chip companies (with Taiwanese participation) vs. Samsung

Starting next year, there is going to be an economic war in memory chips between the Chinese memory chip companies (with Taiwanese help) and Samsung.

There are three important Chinese memory chip companies.

Yangtze River Storage Technology (YMTC) has just completed its 32-layer 3D NAND memory chip (see Electronics Weekly November 15, 2017 article). Mass production is scheduled for Q2 2018.
The acting chairman of YMTC is Charles Kau. "Kau is a former chairman of Inotera, a Taiwanese memory [company] that has been acquired by Micron. And about 50 Taiwanese engineers are said to have joined YMTC." (See EE News Analog May 8, 2017 article)
Basically, Mr. Kau brought over the Taiwanese memory chip experts to mainland China.

Fujian Jin Hua Integrated Circuit has invested US$5.3 billion in a new DRAM fab. "Jin Hua has [Taiwan] UMC's help in developing production technologies, with the first-generation 32nm node to be ready for volume production in the third quarter of 2018, the report said. Monthly production capacity at Jin Hua's DRAM fab is estimated at 60,000 12-inch wafers initially." (See DigiTimes November 23, 2017 article)
UMC is Taiwan's second-largest semiconductor foundry with annual sales of US$4.6 billion.

The third large Chinese memory chip company is a joint-venture between GigaDevice Semiconductor and Hefei RuiLi IC Manufacturing. (See Electronics Weekly November 1, 2017 article)
GigaDevice Semiconductor is a Chinese company that joined with other Chinese companies to buy America's Integrated Silicon Solution Inc. for $731 million in 2015. Integrated Silicon Solution Inc. is a fabless designer of DRAM memory chips.
Integrated Silicon Solution Inc.'s co-founders (Mr. Yeu Han Kong and Mr. Shueh-Mien Lee) and chief executive officer (Mr. Yeu Han Kong) are both ethnically Chinese. It is unclear whether they are Taiwanese-Americans or Chinese-Americans.
----------

Yangtze River completes 32-layer 3D NAND development | Electronics Weekly (November 15, 2017)

----------

Chinese 3D-NAND flash coming this year? | EE News Analog (May 8, 2017)

----------

China firm starts equipment move-in at DRAM fab, says report | DigiTimes (November 23, 2017)

----------

Third China DRAM contender gears up | Electronics Weekly (November 1, 2017)

 

[Martian2]

TSMC has a dominating 55.9% ($32 billion) of worldwide semiconductor foundry revenues in 2017.

TSMC remains the worldwide leader as a semiconductor fabricator for other companies, such as Apple, Qualcomm, Broadcom, Nvidia, AMD, etc.

No one cares about Samsung in the foundry business, because Samsung mostly manufactures its own Exynos smartphone processors. Samsung does manufacture some Qualcomm chips, because Samsung pressured Qualcomm into placing foundry orders as a condition for Samsung to continue using Qualcomm Snapdragon processors in Samsung smartphones.

----------

Top Ten Foundries 2017 | Electronics Weekly (December 1, 2017)

 

[Martian2]

TSMC spends $10 billion in annual capital expenditure for foundry. Samsung spends only $5 billion.

In an apples-to-apples comparison, TSMC's $10 billion annual capital expenditure (capex) on logic-chip foundry manufacturing is double Samsung's $5 billion. Clearly, TSMC is the investment leader (and worldwide market-share leader) in cutting-edge logic chip manufacturing technology.

After licensing IBM logic-chip process technology in 2004, Samsung has been in the foundry business for 13 years.

"Mar 5, 2004 - A separate agreement with IBM will give Samsung license rights to 90nm CMOS logic technology. Samsung plans to introduce IBM's proven logic technology to its highly integrated System-On-Chip (SOC) product line such as HD-TVs, DVDPs, and mobile applications. Based on Samsung's SOC business ..."

Samsung has made little to no progress in the foundry business. At some point, the only sensible decision is to give up on its foundry goals. The structural bias against Samsung Foundry will never go away. Nvidia, Broadcom, Apple, AMD, etc. are not going to place large orders with Samsung Foundry, which will strengthen Samsung Electronics.
----------

TSMC 'will snatch Qualcomm orders from Samsung in 2018' | Nikkei Asian Review (December 22, 2017)

"TSMC, Samsung and Intel are now the only chip companies in the world capable of manufacturing chips with transistors smaller than 10nm.

Rising expenditure

For all of 2017, TSMC said its capital expenditure reached $10.8 billion, and the company plans to spend more than $10 billion in the next few years. This year, Samsung will have spent some $5 billion in its foundry business, out of its total spending of $26 billion on semiconductors -- mainly DRAM and NAND flash memory chips -- according to IC insights.

TSMC is the global leader with 56% share of the market in making chips for others, known as the foundry market, while Samsung, the world's No.1 memory chipmaker, aims to capture a higher share of the foundry sector. Samsung had some 7.7% share for all of 2017, according to Taipei-based research firm Trendforce, and is looking to boost its foundry share to 25% by 2022."

 

[Martian2]

Samsung Chaebol has feet of clay

Samsung is a large industrial conglomerate (aka chaebol). It is the South Korean version of Japan's Zaibatsu/Keiretsu.

In the 1980s and 1990s, Fujitsu and Sony were strong. They used their financial advantage to garner vast worldwide marketshare in camera film and electronics.

The problem with Chaebols/Zaibatsus/Keiretsus is the long-term wastefulness in the use of capital.

Samsung has a very simple business plan. TSMC makes a profit of $10 billion per year. Samsung thinks it can take TSMC's market position by offering lower prices.

The logic is straightforward. Lower prices for leading-edge semiconductor logic-chip manufacture. Slowly starve TSMC of revenues and profits. This will reduce TSMC's investments in R&D. Use Samsung Chaebol's financial resources to win a war of attrition and push TSMC into a second-rate semiconductor fabricator.

The problem is that Samsung's business plan is stupid and not working. Samsung Electronics competes against other companies. If TSMC had higher prices, most companies would still contract with TSMC and not Samsung.

TSMC probably offers lower prices, because it has a superior library of design tools, higher manufacturing yield rate, and economies of scale.

TSMC is the market leader with 56% worldwide market share. Samsung only has 8% worldwide market share, which includes its own Exynos chip production. If you count Exynos as an IDM (integrated device manufacturer) and not foundry (which by definition means third-party semiconductor fabrication) then Samsung's share is only about 4%.

In a rational world, Samsung would have stopped manufacturing leading-edge logic chips. Samsung has the same costs for R&D and facility construction as TSMC. TSMC has over 40 customers for 7nm logic-chip production. However, Samsung only has its own meager Exynos line.

Samsung is flushing billions of dollars down the drain with no meaningful return on capital.

That's not the way capitalism works. You're not supposed to engage in non-productive economic activity. The stubborn Japanese subsidized their zombie Keiretsu companies until most of corporate Japan went bankrupt after 20 years.

We're seeing something similar in Samsung. It is waging a losing economic war against TSMC. TSMC is the dominant third-party semiconductor fabricator in the world. TSMC's profits are eye-popping. Samsung can keep dreaming about taking TSMC's place, but it can't land orders from large customers.

Memory-chip profits are paying for Samsung's logic-chip adventure. However, China is ramping up its own memory chip production and we'll see how long the Samsung memory-chip profits can continue.

In capitalism, you pursue an economic activity where you have an advantage. TSMC has proven it can crush Samsung in logic-chip manufacturing year-after-year. Just like the stubborn Japanese Keiretsus, Samsung keeps investing billions in a business with no profit. That's the definition of insanity. I predict Samsung will follow the Japanese Keiretsus into oblivion in about 20 years.

When a company keeps wasting its capital, it is only a matter of time before the competition overtakes Samsung. There are two prominent examples. Firstly, TSMC is destroying Samsung in foundry. Secondly, Huawei has displaced Samsung's former position as the top seller in the Chinese smartphone market (which is the world's largest by volume).

Samsung would be a much stronger company if each business division was forced to stand on its own. If each business was profitable and growing then there's a positive economic contribution from every business category. Instead, we see Samsung wasting its financial resources in a futile attempt to drive TSMC out of its market leadership.

The Japanese have already tried Samsung's plan. It didn't work. Fujitsu gave up on logic-chip manufacture at 28nm.
----------

TSMC to outpace Samsung in 7nm volume production in 2018 | DigiTimes (January 4, 2018)

"Taiwan Semiconductor Manufacturing Company (TSMC) is ready to enjoy a sure win over Samsung Electronics in the race of 7nm process volume production in 2018, as it has secured orders from over 40 customers for fabricating chips on the advanced process for mobile communication, high-performance computing and AI (artificial intelligence) applications. Both Apple and Qualcomm are among TSMC's major customers, with the foundry house contracted to fabricate all the A12 processor chips for the 2018 new-generation iPhone devices, according to industry sources."

 

[rambro]

Samsung spends too much on marketing