Semiconductor Manufacturing
International Corporation (SMIC) is China's flagship semiconductor foundry. It’s a leading player in challenging TSMC, Samsung
and Intel in the semiconductor technology race. Founded by the intensely driven Richard Chang,
SMIC ruthlessly hired from its Taiwanese rivals, outcompeted its Chinese ones, and
quickly established itself as the top dog in China's burgeoning semiconductor industry. In this video, let's look at SMIC - China's
fiercest semiconductor foundry player. SMIC was founded in January 2000 by Richard Chang.
Chang is an interesting fellow. Born in Nanjing, his family fled to Taiwan along with a
million other ethnic Chinese migrants. The waishengren, as they are
called, grew up in Taiwan but held deep roots to the Chinese mainland.
I did a video about them earlier. Many of the waishengren, disoriented
and feeling unwelcome in Taiwan, emigrated to other locales like the United States.
Chang was one such person. After graduating from the prestigious National Taiwan University, he
got a master's degree from the State University of New York in Buffalo. Then he and
his wife went to work for Texas Instruments in Dallas. TSMC founder Morris Chang
(no relation) also worked at TI for over 25 years. At TI, Richard Chang oversaw foundry work in East
Asia for twenty years. He did well there and took early retirement in 1997. Returning to Taiwan,
he started a foundry of his own - Worldwide Semiconductor Manufacturing Corporation
and eventually sold it to TSMC in 2000. Richard soon then itched for another opportunity
and China came calling with cheap land, water and electricity. He crossed
the Strait and started SMIC with $1.6 billion of investment from Goldman
Sachs and a few other private equity firms. Chang and SMIC soon found themselves in
fierce competition with Grace Semiconductor. Grace was founded around the same time
as SMIC by two political heavyweights. The first was Winston Wang, son of powerful
Taiwanese industrialist Wang Yung-ching. Wang Yung-ching founded one of Taiwan's biggest
chemical empires, Formosa Plastics Corp. The second heavyweight was Jiang Mianheng,
son of Chinese paramount leader Jiang Zemin. Naturally having such connections
would put Grace at the very forefront of the Chinese semiconductor foundry industry.
SMIC knew that it would have to compete extremely hard in order to distinguish
itself from the competition. They did. As it turns out, being a rich guy's son
does not help much in competitively running a semiconductor foundry. For Richard Chang
and his team, this was their second startup and they knew how to get things done. In contrast, Winston Wang had little experience in building
and running leading edge independent foundries. Grace could not keep up. In the same time Grace
took to get its first foundry to mass production, SMIC had two in Shanghai, a bunch more in the
hopper, and was already prepping for their NYSE IPO. Grace settled into a number two
position and the Chinese government soon shifted their favor to SMIC in terms
of subsidies, cheap loans, and more. By 2004, less than 5 years after its founding,
SMIC had ascended to become the fourth biggest player in the independent foundry industry.
Only TSMC, UMC, and Singapore national champion Chartered Semi were ahead of it.
And it was rapidly eating up market share. Two big things you need when running a
startup foundry is a fab to make the chips and a lot of talented people to run it. Otherwise, I'm not sure you might have a feasible
business. You might have a Hongxin. SMIC's business model at the time is not all
that different from an NFL or NBA sports team. A foundry needs an expensive multi-billion
dollar fab to do their work. Likewise, a sports team "needs" an expensive multi-billion
dollar arena. Neither of those organizations can raise or risk the billions of dollars to
self-finance one. So, get someone else to do it. Unable to tap bank financing, SMIC thus went
to Chinese municipalities around the country and got those local governments
to finance a fab's construction. SMIC would then move into the fab on a contract
basis to operate it. The local government would get a fee against the future revenues and the
opportunity to build a "semiconductor cluster". Kind of like how NFL teams
like to say that on game day, local businesses benefit from all the increased
traffic around their billion dollar stadium. You should Google those claims to see
how it turned out for those local cities. As SMIC ascended up the chain and the leading
edge process got more and more expensive, these capital expenditures became onerous
even for the richest local governments. SMIC began performing a type
of regulatory arbitrage, shopping their services to cities around
the nation looking for the best deal. For example, in 2005 SMIC was
running two factories in Shanghai. Then the next year they engaged
with the cities of Chengdu and Wuhan to start foundries there too. Two more joint
ventures at a cost of some $1.5-3 billion each. It allowed SMIC to grow extremely fast while
keeping their actual invested capital low. In a previous video, I likened advanced
semiconductor foundry work to baking. And like baking, accumulated, unwritten
experience is critical. For the life of me, I can't make a decent cake even if the steps
in the recipe are right in front of me. This is a broad generalization, but
foundry work is in many aspects an art that cannot be easily boiled down into steps. Not everything is or can be written down - so
it is critical to have with you the best chefs. Both Grace and SMIC aggressively recruited
overseas talent to fill their ranks, including Taiwanese from TSMC and UMC. Being a
startup, these companies lured such employees with the promise of stock option riches. At
the time, salaries on the Chinese Mainland were far below that which can be found in the
US or Taiwan. A senior manager in China for instance would get paid a quarter or a third
of what that same person could get in the US. Grace handed out stock options but
SMIC gave out them out like candy. For certain key TSMC employees, SMIC offered
80,000 shares and stock option equivalents. Such amount of shares at the $17.50 US IPO
price in 2004 would be worth $1.4 million. Not to say that those people actually got
their hands on all of that $1.4 million. I have worked at startups myself and it is very
common for that 80,000 to quickly get diluted as the company goes through
funding rounds and the like. Beyond the potential riches, people
gravitated to SMIC for two other reasons. First, many of these engineers were ethnic
Han Chinese living and educated overseas and wanted to give back to their
country. Of SMIC's founding 1,000 engineers, nearly 40% of them were Chinese
citizens returning to the motherland. Second, many of them felt that they had hit the
glass ceiling in an American or Taiwanese company. Big companies have often established
ranks of who gets promoted, and there is often more qualified people than
open spots. We should also mention the possibility of racism against Asians and Chinese in
the United States - the bamboo ceiling. To join SMIC would be to step into a
Silicon Valley-like environment and get in on the ground floor on something exciting. To build something for the future and
glory of China. Hard to say no to that. By 2003, TSMC realized that Grace and especially
SMIC were growing faster than it should have. For example, SMIC had managed to
ramp up on 0.18 micrometer process in just 12 months. This is without a prior track
record of success in doing such things before. This was implausible and implied the loss of
private, extremely valuable trade secrets. Thus in that same year as SMIC
prepared to go IPO on the NYSE, TSMC launched a pioneering intellectual property
theft lawsuit against its mainland rival. This lawsuit was filed in California and
interestingly enough TSMC asked for a jury trial. It would not be resolved for
many years, but when it did, the jury found that SMIC had indeed
infringed on TSMC intellectual property. The lawsuit went back and forth. But when
the dust finally settled six years later TSMC won. They received a 10% stake in
SMIC. Richard Chang resigned in 2009 after attempting a counter-suit
against TSMC that failed. At the same time, TSMC took to battling
SMIC directly on its home turf. In 2004, TSMC won a public debate in Taiwan for
permission to enter the Mainland market. In 2007, TSMC's first fully owned
foundry began operations in Shanghai. This appears to be a page out of their
playbook. In my video about Singapore's Chartered Semiconductors, I mentioned
TSMC founding a semiconductor foundry joint venture with the goal of battling
Chartered on their home market. It helped to weaken the overall market enough
to decelerate SMIC's once-torrid growth. Throughout this entire saga, SMIC lost money.
The company had been running on subsidies, loans and the money it received from
investors but the company could not consistently turn a profit on its
own. There are a few reasons for this. First, there were the upstarts. SMIC was
the largest Chinese foundry but it was not alone. Other Chinese cities
saw what was happening in Wuhan, Beijing and Chengdu and wanted in on the action.
They began creating small SMIC's of their own, saturating the market and bringing
down profits for everyone else. One upstart takes place in the city of Ningbo,
the place of TSMC founder Morris Chang's birth. Ningbo Zhongwei was founded in 2002
by a team of mid-level TSMC managers. They bought used equipment from TSMC and
invested $150 million of taxpayer money to build a leading fab capable of
outputting 40,000 wafers a month. The fab reached the 10,000 wafers
a month benchmark behind schedule. The company could not reach the necessary
scale as TSMC's equipment was very old and required expensive servicing
($100-150 per hour maintenance). Eventually the company folded and it sold
their factory to electric car maker BYD. Another reason has to do with the product
being sold. In order to make sure some of its early fabs got online as fast as possible,
SMIC devoted them to creating commodity DRAM. The fabs indeed got up and running really
quickly, but making a commodity product meant low margins and intense competition. It is
the same issue that Chartered dealt with. You need to ascend the value chain and get to the leading
edge. That means a lot of investment into R&D. Too many startups. Too much supply sloshing
around. Not enough differentiated product. And of course some of the most economically challenging
years with the global financial crisis. Where can you get the money to invest in the R&D for better product if you are not
making money from your products? Is the Chinese government just going to have
to constantly pump billions into your company? I guess they can, but how long can you put
money into a company that doesn't turn a profit? Singapore owned Chartered for 22 years and
for the majority of those years Chartered churned out losses despite becoming the third
largest independent foundry in the industry. Just because you have a lot of money doesn't mean
you like rolling it up in a cigar and smoking it. That was what's happening with SMIC so
long as it did not own and create the technology behind the processes
themselves. They were trapped. Over time, local city governments like Wuhan
and Chengdu began to feel uncomfortable about these seemingly unending operating
losses. After Richard's resignation, Chengdu and Wuhan terminated the
partnerships. Chengdu sold its fab to Texas Instruments and Wuhan decided
to go it alone with Xinxin Semiconductor. SMIC hired a new CEO in 2010 and sought to dial
down on growth in order to achieve profitability. If Richard Chang was Uber's Travis Kalanick
- the vicious, driven founder determined to take over the world - then his successor
David Wang was more like Dara Khosrowshahi - trying to turn the company
around and make a profit. David Wang did not last long. He resigned after
being losing a vote of confidence a year later. Tzu-Yin Chiu led the company as CEO until 2017, whereupon he helped stabilize the
ship and turn out a very small profit. He retired in 2017 and was succeeded by
co-CEOs Zhao Haijun and Liang Mong Song. Liang is infamous in Taiwan for jumping
ship from TSMC to Samsung and helping them reach the leading edge process node. It seemed
that he had been doing the same with SMIC too, leading a team of 2,000 R&D engineers
to ready the 7nm process node. He realized just how important it is
to research, create and thus own the process technology itself. Not to take shortcuts
like Chartered did in partnering with others. Per recent reports, it seems
like they are getting close to high volume production with it. This is great
news for the Chinese semiconductor industry and represents a return to form
for SMIC after many long stumbles. SMIC remains the premier foundry on the China
mainland. Their recent billion dollar IPO on the Hong Kong and Shanghai markets after a 2019
NYSE delisting makes it clear. The Chinese national government is starting to invest real
resources in actually developing an indigenous semiconductor space from the ground up. SMIC will
likely be its centerpiece for making it happen.
