What Happened to Singapore's TSMC?
Or, "How Hard It Is To Compete in the Semiconductor Foundry Space"
Author’s note: If you want to watch the video, it is embedded below
Author’s note: I think Chartered Semiconductor demonstrates the best of Singapore’s approach towards developing industries within its economy. They managed to create an industry for itself in this area, and that is what we should focus on here—not the fact that Chartered was sold off. It demonstrates the difficulty of the industry and how ruthlessly its leaders play the game.
If you are interested in what this channel does and want to support the work, feel free to check out the Patreon. The Early Access Tier lets you watch videos up to 2 months before they released to the public. That includes videos on companies like BOE Technology and Reliance Jio. Nifty, I reckon.
In September 2009, Abu Dhabi's sovereign wealth fund agreed to buy Singapore's Chartered Semiconductor Manufacturing for $2.5 billion SGD or $1.8 billion USD. Abu Dhabi would merge Chartered's operations into its GlobalFoundries venture.
Chartered had been the third largest independent foundry in the market. It was Singapore's only big indigenous player in its burgeoning semiconductor industry. The acquisition marked the end of a 22 year journey of challenges and financial losses.
In this video, I want to talk about what was supposed to be Singapore's TSMC. How it started, where it went, and how it ended as an independent entity.
Upscaling into Semiconductors
In a previous video, I talked about some of the strategies Taiwan used in order to launch its semiconductor industry. The gist is that they used public-private arrangements that spun off small private entities to commercialize raw technologies. Some of these companies would fail. Some of these companies wildly succeeded. Most are just okay, which isn't bad because they can absorb employment.
Singapore's industry development strategy leans heavily on using its business-friendly environment to attract large multi-nationals to establish a presence on the island. You start getting spillover effects, transfers of technology and skillsets coming over from foreigners to Singaporean locals. As the theory goes.
I am kind of reminded of those whalefalls at the bottom of the ocean. When a whale dies, its body sinks to the sea floor. Soon thereafter, animals arrive to feast on the meat. Before long, you have spawned an entire ecosystem. Where before there was nothing, now there is life. Okay, weird metaphor but it gets the point across.
In pursuit of this goal, Singapore considers its well-educated, hard-working population as one of its most critical assets. This is in contrast with other countries like Saudi Arabia, Canada, and Australia, which boast vast natural resources like oil, forests, and iron ore. Singapore boasts vast reserves of human capital.
Like Taiwan, Singapore has had a presence in electronics assembly since 1969. It makes up one of the economy's big industrial exports alongside refined petroleum products and fabricated metal products. The country especially did well in disk drives and memory - Singapore at the time was the world's disk-drive capital.
But as time passed, new competitors were emerging in the space and it was clear that this could not remain the status quo. The global marketplace is intensively competitive. Multi-national disk drive makers were beginning to decamp for places like Malaysia and Indonesia, chasing the next low cost locale.
Singapore's economic planners saw chip making as the natural next step for the country's electronics industry. It is not labor intensive but rather requires highly skilled and educated workers.
With Chartered Semiconductors, Singapore sought to climb the value chain and nurture a semiconductor national champion of their own.
Singapore has had success growing indigenous national champions even in competitive areas like airlines. Singapore Technologies Group (STG) is one of those champions. It started as a defense equipment supplier, growing into one of Singapore's biggest companies. STG was selected to spearhead Singapore's entrance into the semiconductor space.
STG kicked off what would eventually become Chartered Semiconductors in 1987 - the same year of TSMC's founding - with a technology transfer agreement between itself and US fabless semiconductor maker Sierra Semiconductor.
STG had originally hoped that it could count on the support of Sierra and US multinational company National Semiconductor. But National Semiconductor dropped out and Sierra was not a big customer. But STG saw the success of TSMC's innovative new independent foundry business model and restructured Chartered to copy it.
Chartered thus spun off its Integrated Circuit design house and moved forward as a pure foundry. Chartered moved forward, rebranded as Chartered Semiconductor Manufacturing or CSM.
In 1999, they went public on the NASDAQ and Singapore stock markets. A year later, the company would be restructured as a majority government-backed entity directly within Singapore's sovereign wealth fund Temasek.
Growth and Business Strategy
At Chartered's founding, Sierra transferred over its 3.0 micron process technology and trained 100 Singaporean technical staffs in how to use it. This process technology was way outdated and it needed to be updated to the cutting edge as fast as possible.
Almost immediately, CSM began an in-house R&D lab in order to improve the process tech they got from Sierra. Highly motivated to advance as fast as possible, they pushed ahead quickly.
1990, they achieved 1.5 microns.
1991, 1.2 microns.
1992, 0.8 microns.
1993, 0.6 microns.
A year later in 1994, CSM announced a new deal with Toshiba to use their 0.5 micron process technology. They rapidly expanded their production facilities throughout the 1990s with five foundries on some 60 hectares in Northern Singapore. Plans had been drawn up for up to 25 such foundries in the near future.
Meanwhile, CSM began to expand its product assortment - seeking to capture as much of customers' business as possible. The more of a complete, turnkey solution that they can offer to their customers, the more locked-in their customers can be with CSM.
The Singapore government put its muscle behind the company by offering multi-national companies ample tax incentives and subsidies to come and lodge their manufacturing facilities there. Government funds helped companies pay for the costs of training staff. State-run universities started programs to graduate electrical engineers.
The same tactics that helped Singapore get its start in the memory industry, government officials practiced to get leverage in semiconductor fabrication. The government's muscle helped to get CSM to third place in the foundry industry behind Taiwanese companies TSMC and UMC.
So what happened to CSM? Why did they agree to a buyout? They essentially got caught in a pincer movement that eventually led to the company being squeezed out of the foundry market.
Mainland China's Low-cost Startups
The first has to do with the entry of low-cost semiconductor foundries from Mainland China. I talked about this a little bit in a previous video.
In 2000, a few Chinese independent foundries entered the market. China had recently entered the World Trade Organization, giving birth to the biggest export machine in world history.
The most notable of these Chinese startups was Semiconductor Manufacturing International Corporation, or SMIC. Founded and led for many years by a former TSMC vice president, the company rose to become China's foremost independent foundry.
SMIC took advantage of substantial government subsidies and cheaper costs of living at the time to disrupt the market from the bottom up.
The same process technology licenses available to CSM were also available to them too. Thus, they climbed the technology process tree just as fast. Since the tech all came from IBM or the like, vendors could easily switch to the lowest cost provider.
The same tactics of cheap land, tax subsidies, engineering talent, and more to lure semiconductor entrepreneurs and MNCs, China can offer just as much as Singapore. But on an even larger scale.
And then TSMC decided to bring the battle to CSM on its own home turf. In 1998, TSMC executed on a joint venture with frequent partner Philips Electronics to start SSMC in Singapore. TSMC casually threw down $2 billion SGD over 3 years to build a fab there with trailing-edge technologies - just good enough for many customers in the Singaporean semi market.
CSM discovered that they could not offer a product differentiated enough from their cheaper Chinese and Taiwanese competitors. Ergo their prices had to go down, eating away at the gross margins of everyone in the business. You can say that gross margin is a rough financial indicator of just how special your product is and how much people are willing to pay for it.
In 2005, CSM made just 11% gross margin on their product. 2003, they did -17.5%. During that same period TSMC's gross margins were 44%. The company needed to go up the ladder, get to the leading edge and differentiate their product in some other way.
But suddenly that ladder vanished beneath them.
The Arrival of the Samsung Giant
In 2005, Samsung announced that they would enter the standalone foundry business, bringing their immense resources and expertise to the fray. They would leverage both to quickly rocket to second place ahead of CSM and UMC. I did a video about them earlier.
Samsung Electronics was already one of the biggest electronics and semiconductor companies in the world. They owned the largest share of an oligopoly in the digital memory market as well as the biggest handset business at the time. They knew what they were doing, and they had a captive customer for their product (despite whatever they might say about Chinese walls between the various divisions).
So while the Chinese competitors ate away at the lower rungs of the foundry market like Pac-man, Samsung stole the business of servicing high end customers looking for either diversification, additional capacity, or better pricing from TSMC. Worse yet, Samsung kicked off a war that CSM could not compete in.
Capital Starvation and Hard Times
Samsung's entry into high end foundry work directly threatened TSMC. Samsung not only brought formidable engineering prowess drawn from their memory business but also billions and billions of dollars.
TSMC then decided that it was time to get real and hit the gym. Rather than using their profits to boost their stock by buying back shares or paying more dividends, they poured money into fabs, R&D, and more.
The semiconductor world is highly cyclical. There are good years that are great and bad years that are really bad. The early 2000s saw consecutive bad years. Too many foundry players and not enough demand.
It is not just about ponying up billions, but also doing so during these hard times. TSMC had the wherewithal and the patience to invest billions to pick up failing fabs and continue on their massive foundry work without pause. Same with Samsung.
Meanwhile, staying on the leading edge just kept getting more expensive. In 2005, a leading edge fab cost $3 billion to build, six times higher than what it would have cost in 1995. Fifteen years later in 2020, TSMC's leading edge N5 fab in Tainan cost $17 billion. Its next generation N3 process fab (also in Tainan) is estimated to cost $19.5 billion.
CSM tried to lighten the financial load by partnering with other organizations like IBM to develop new edge process technologies. But doing so drastically slowed down product development and they fell behind the curve anyhow. Not to mention that the tech partner would then license out their learnings to competitors within a few years.
Thus CSM found itself caught in a pincer movement. Not enough revenue to fund their work. Not enough subsidies to stay in the game. Too far behind in the tech curve to catch up. But can't stay where they were. The Singapore government was willing to spend money. They allocated hundreds of millions of dollars to CSM and its dreams of having a national champion. But hundreds of millions falls far short of billions.
By the time the financial crisis came around, CSM and its investors realized that they had lost. The Singapore government stuck around for 22 years of financial losses, but it was time to let go.
Abu Dhabi then came calling with a bold idea. They wanted to build a global national champion in semiconductors. They would use their voluminous oil profits to buy out a number of foundries in search of a global giant. In doing so, the people of the UAE can learn valuable STEM skills and business savvy. It's a classic journey by other Gulf states to head off potential turmoil after the oil fields run out.
GlobalFoundries offered $2.68 SGD for CSM's stock in September 2009. The price was just 2 cents higher than the closing price. Since majority holder Temasek backed the deal, the acquirer didn't feel the need to offer much of a premium.
For what it is worth, GlobalFoundries never managed to execute on their business strategy. GloFlo as it was colloquially called could not reach the necessary scale for escape velocity. It is more than just a money thing.
Just around the time of GloFlo's founding, TSMC and Samsung as the market leaders executed on a number of necessary technical moves to lock in their premium leading-edge customers at a very deep level. It's hard to explain, but it basically means that if you wanted to make a chip with the most advanced processes post 2010-ish, you can only use TSMC or Samsung.
Unable to keep up with the tech race and now locked out of the premium market, GloFlo in 2018 ended their 7nm development to focus on specialized processes. Internet commenters immediately began second-guessing the move.
Conclusion and aftermath
Singapore set out on its journey with CSM in order to build its own TSMC. They had a lot going for them. They had an established strategy, deep connections with powerful Multi-Nationals, and a well-educated work force. With this, they sought to build a thriving semiconductor industry on par with what Japan, Taiwan or South Korea had.
CSM failed and that is unfortunate because the Singaporean government is extremely competent, patient, and effective at what they do. They deserved success, but success is never guaranteed. Entering into such a market, it is like running in a race with no ending. There is no finish line to celebrate, because your competitors keeps pushing it further down the road.
There are also unsettling questions about how much the Singaporean natives actually benefitted from the massive investments the government threw into CSM. A significant percentage of the jobs, including on the senior level, went to highly paid expat foreigners. That did not much change throughout the 22 years of losses. How much did the ordinary Singaporean benefit? These are difficult things to think about and consider when reviewing the results. But of course, hindsight is 20-20.
But I think the overall Singaporean semiconductor sector continues to do alright. It remains a significant part of the economy, creates jobs for Singaporeans, and makes a profit for its partners. That's a good thing and something we should not ignore. Home runs are home runs for a reason. They are rare and hard to make. Doing a bunch of doubles and singles can be perfectly fine.