Discover more from The Asianometry Newsletter
Taiwan's Fabless Giant
Author’s note: If you want to see the video, you can start below:
TSMC and Foxconn get all the media attention. But MediaTek is as treasured and valued by the Taiwanese investment community as the others. This company’s journey to becoming an important player in the chip design industry is special. Its business model of selling reference designs for others to modify and resell is pretty unique. I see a lot of it here in East Asia, but rarely saw it in the States.
These deep dives into Chinese and Asian companies are fun and I hope to able to do more of them. If you want to support the channel, I recommend you check out the Patreon. I am moving the Newsletter to a weekly cadence, partly because so much of my recent writing has not been made public yet. It is available to Early Access Tiers until it finally comes up on the release schedule.
TSMC’s founding in 1987 would launch two new types of semiconductor firms: Foundries and design firms. Foundries and the complexities of semiconductor manufacturing have been taking most of the headlines. If you are interested in that, you can check out my TSMC YouTube playlist, linked below in the description. But the design firms fabless are just as vital. They are the customers who keep the foundries fed.
Taiwan also has a thriving fabless semiconductor ecosystem. In this video, we are going to look at the largest of those fabless companies: MediaTek or MTK. I will use both names interchangeably.
Over the years, the Hsinchu-based company built a place for itself designing and fabbing mobile phone chips for developing markets. Qualcomm is the industry leader, and they make the sexy super-fast chips that gadget enthusiasts lust for. But MediaTek has been doing quite well for itself.
## Out Of UMC
MediaTek began in 1997 as a spinoff startup. It had been a part of United Microelectronics Corporation or UMC. UMC was Taiwan's first indigenous semiconductor company and had been operating as an integrated device manufacturer. This means it designed, fabbed, and sold its own chips.
As I said in my video about how Taiwan created its semiconductor industry, UMC came out of the same research institute as TSMC - ITRI. UMC had done okay at first but TSMC leapt out of the gate with its innovative new business model. UMC decided that in order to keep up, it needed to pivot and also adopt the independent foundry business model.
But customers could not accept UMC as an independent foundry unless they knew that UMC would not compete with it. So to gain customer trust, they decided to spin off their semiconductor design operations as multiple companies. Thus was founded MediaTek, which had once been UMC's R&D department for multimedia devices. Ergo the name.
Current Chairman Ming-Kai Tsai had been a design manager at ITRI and joined UMC in 1983 as its Director of R&D. Over the years, he climbed the ranks and eventually oversaw the development of UMC's memory and multimedia business units. So he was the natural choice to lead this small multimedia startup.
From CDs to DVDs
True to its multimedia device roots, MediaTek began with the opportunity afforded by the growing CD-ROM, DVD, and Blu-Ray market. Those devices needed controller chips to work. That chip market was currently dominated by Japanese suppliers.
MediaTek defeated those incumbents by offering a cheaper, highly integrated system on a chip. They contracted with independent foundries like UMC and TSMC to build this chip. Since the foundries took care of the manufacturing, it allowed MediaTek to focus on the design and selling. This is the crux of the fabless semiconductor model. It is also, as it turns out, very profitable.
Taiwan-based CD-ROM manufacturers like BTC and Lite-On adopted MediaTek’s solution first and soon thereafter Hitachi, Sony and LG. They would win a 50% share of the CD market.
MediaTek followed up their CD success with DVDs. Their big breakout hit in that market was supplying chips to Apex Digital, which made a cheap DVD player that sold very well in America. Sony and Philips then signed on for their own DVD players.
Just four years after MediaTek's spinoff, the company was generating about $200 million in annual revenue. It was already Taiwan's second largest fabless semiconductor firm. But Tsai envisioned his company doing more than just DVD players and set upon a brand new market: Mobile phones.
Mobile Phones and the Turnkey Model
MTK began studying the mobile phone market in 2001, long before the iPhone. So most phones back then were simple flip phones. I had one a long time ago. They formally entered the market three years later, soon capturing 14% global market share.
MediaTek's style of business is to attack entrenched players in markets after they mature. Their go-to-market strategy for doing so is to offer a "turnkey" solution to their customers. That means that MediaTek offers for a particular device category a "skeleton", called a reference design.
These MTK turnkey reference designs might be low-priced, but boast high quality features like touchscreens, 5G connectivity and good cameras. They also come with well-tested software. To help their customers best modify the reference design, MediaTek also offers a development SDK as well as technical training.
Customers buy these designs and modify them to best suit their business strategy. How so? It depends on the customer. One smartphone maker specialized in phones for farmers. They modified their phones to have extremely loud speakers so that the farmers can hear them ring while in the fields. Or they can decide not to modify anything, slap on a brand, and just sell the whole design as is.
MTK's reference designs were very popular globally, with many unique features requested by customers for their own local market. For instance, Brazilians like to switch between different telecoms, so their phones needed to have as many as 4 SIM slots.
With MTK, a phone vendor can now start their own phone brand without having to reinvent the wheel. It cut the go-to-market time from 9 or 12 months to just 2 or 3 months. Furthermore, the financial startup cost would be just half a million USD.
This "white labeling" business practice is not unfamiliar to Taiwanese electronics vendors. For example, Acer's UniLoad system - best described as a skeleton desktop computer - was a modular system that sold very well in Latin America. It drove much of Acer's growth in the 1990s.
The iPhone and the smartphone era it launched would upend the traditional desktop computing paradigm. MediaTek quickly realized that they needed a smartphone offering of their own.
In 2012, the company rolled out their first Android designs. They had actually started three years earlier in 2009 with Windows Mobile, but nobody wanted that. It was a big flop and actually caused a dip in company revenue in 2011.
One example design is Mediatek MT6589, a quad-core chip manufactured on TSMC's 28-nanometer process and released in December 2012. Phones using this ARM core would be priced at $150-$200 off-contract, but worked as well as the higher-end phones of the era (i.e. the Nexus 4 or iPhone 4S).
Many of MTK's designs went to China. It is the dominant mobile chipset in China, with nearly 40% market share. The Chinese handset market back in those early years were rough and tumble. Price and features was the main differentiators and an ecosystem of cheap, low-end mobile phone makers sprouted to fill the need.
MTK's chipsets powered many of those early Oppo, Xiaomi and ZTE phones. It meant that MTK was in a position to sell picks and shovels to the gold miners. When those domestic smartphone makers took the China market away from foreign brands like Samsung and LG, MTK benefited handsomely.
MediaTek's fabless focus and its suppliers' manufacturing prowess meant that its chips were often very good. The company wants their chips to be disruptors in the market. They bring once premium features to the low-cost market and eat their way up to challenge the market incumbents.
MediaTek’s reliance on the China market means that this Taiwanese company depends very much on friendly relations with the mainland. That means a few potentially thorny situations.
Here is one example that occurred in November 2015. That month, Zhao Weiguo, CEO of Tsinghua Unigroup, visited Taiwan and lobbied the government to amend its law to allow it to acquire MediaTek. MediaTek, along with many other Taiwanese semiconductor makers, in 2015 was struggling from a weak economy.
Tsinghua had already purchased MTK’s biggest mainland rival - Spreadtrum. Merging with MTK would have created the world's largest fabless semiconductor maker at the time and a true rival to Qualcomm.
Careful of angering its valuable Chinese customers, MediaTek issued a cautious press release saying that they were amenable to negotiations. But the Taiwanese government would eventually decline to open up its semiconductor makers to Chinese acquisition.
Luckily for MediaTek and the Taiwanese government, Tsinghua soon fell to its own debt issues and so nothing came of it. MediaTek would recover from its woes and more. In late 2020, industry analysis firm CounterPoint found that MediaTek had been the world's leading smartphone chipset vendor in Q3 2020, beating Qualcomm for the first time.
Taiwan's semiconductor and electronics makers are critical to its overall economy. There are strong Taiwanese companies in other fields such as Giant for bicycles or Formosa Plastics for plastics, but the heart and soul of Taiwan's economy is in electronics.
Taiwan's fabless design firms and foundries work together in a harmony to push ahead and challenge integrated device manufacturers like Intel.
The foundries need the input of their customers to push them and keep them on their toes. Each fabless designer benefits from the combined investment and knowledge from many other designers like themselves. It helps keep them at their very best and at the cutting edge of the ultra-competitive electronics space.