Latest Thread
TSMC Is Becoming the Global Chipmaker It Didn’t Aspire to Be
A little over three years ago, Taiwan Semiconductor Manufacturing Co. was among the world’s most geographically concentrated technology giants with almost the entirety of its capacity within a 300-mile radius. Now, it is on the verge of becoming one of the most globally diversified chipmakers. This wasn’t the plan.
A new facility near Dresden, Germany, is set to begin operations in 2027, the Hsinchu-based company said Tuesday. Coupled with current plans, TSMC will have factories in five countries spread over three continents, rivaling the sprawl of rivals Intel Corp. and Samsung Electronics Co. These overseas plants add to the significant operations it has in Taiwan and the two existing sites in China. (For more than 25 years it has also owned a fab near Portland, which though profitable is small and not seen as a company success story.)
Having all its manufacturing close to home has always been an advantage for the made-to-order chip foundry. The tight relationship between research and development, and factory operations, where engineers can easily shuffle between production lines, helped TSMC become a fast-moving supplier in a high-stakes industry. Dotting the world with fabs risked diluting this advantage.
But then TSMC’s true global expansion kicked off in May 2020 with the announcement of a new facility in Arizona, a project which was enhanced two years later to include a second plant at the site, taking total investment in the Southwestern state to $40 billion.
A venture with Japan’s Sony Group Corp., unveiled in 2021, took TSMC in a new direction. Instead of owning a factory outright, Sony Semiconductor Solutions Corp. will take a 20% stake in a factory being built in Kumamoto. Automotive components supplier Denso Corp. later signed on to take a stake of over 10%. (Fun fact: that plant is closer to Shanghai than Tokyo).
Dresden is a continuation down that path of working with clients to jointly own facilities, largely to supply the growing demand for components used in automobiles. TSMC will invest up to €3.5 billion ($3.8 billion) for a 70% share of newly formed European Semiconductor Manufacturing Co. Robert Bosch GmbH, Infineon Technologies AG and NXP Semiconductors NV each take 10%, and total capex is expected to be around $11 billion, with the money coming from equity, debt and German and EU funding.
Since its founding by Morris Chang more than three decades ago, TSMC eschewed equity partnerships in favor of maintaining full control over its operations, and thus its destiny. But the global winds have changed, and its new leaders, Chairman Mark Liu and Chief Executive Officer CC Wei, have had little choice but to adapt. TSMC’s balance sheet is solid, its cash flow is stable, and its credit rating is high. It doesn’t need clients nor governments to hand it money in order to pay for these new facilities.
What it does need, though, is buy-in. These remote factories at locations many time zones from home require firm orders as well as a solid commitment from third parties motivated to ensure the company’s success. Having the likes of Sony, Infineon and NXP on the ownership list ensures they have skin in the game, while government involvement should help secure political and economic support.
Suddenly, TSMC goes from being an under-the-radar Taiwanese supplier solely focused on a coterie of semiconductor clients, to a global entity with multiple stakeholders across numerous national and local jurisdictions. It’s already proving to be a difficult adjustment.
Liu last month announced the delay of its Arizona opening by about a year. Time spent navigating local regulations and a struggle for talent, including among vendors, means TSMC won’t kick off operations there until 2025. Last week, the company signed an agreement with Arizona Governor Katie Hobbs to follow a worker safety program that’s stricter than federal rules, a sign that TSMC needs to keep adjusting to a changing regulatory landscape. Continued concerns about pay and conditions among local workers means a labor dispute could flare up at anytime, a situation uncommon at home in Taiwan.
Also of surprise is the escalating scale of divergence between costs in the US and Taiwan, which will likely force the chipmaker to charge clients like Apple Inc. and Nvidia Corp. significantly more for products made in Arizona. The Japan plan appears to remain on track for production late next year, with a high chance a second fab will be added to the project.
Yet despite the $60 billion to be spent in total by all parties, the new facilities will account for no more than 10% of global capacity. And not all fabs are created equal; the best stuff will remain in Taiwan for the foreseeable future, with Dresden and Kumamoto both deploying much older production technology — which is fine because automotive chips don’t need anything more modern.
Still, these foreign partners have no reason to complain. Clients are getting a stake in, and access to, precisely the factories and know-how they need. Governments, meanwhile, can tell their constituents that they’ve been successful in luring the world’s most important technology company to their shores.
TSMC is also a winner. Just five years ago, the company warned investors that the European Commission was looking into concerns about “alleged anti-competitive practices” in relation to semiconductor sales. The US Fair Trade Commission was also showing interest, it was reported at the time. Nothing came of these probes, but it would be particularly awkward for regulators in either jurisdiction to now accuse TSMC of being a predatory tech giant when its management has bent over backwards (and spent billions of dollars) to set up shop on their turf.
These overseas plants also dampen the constant drumbeat among rivals that TSMC is overly concentrated in one place, and that governments and chip customers need to look elsewhere. Now, the company is giving them that “elsewhere.” Half the world gets a piece of TSMC, and in return all the chipmaker had to do was lean into globalization.
I really wish that our own chip production would spend more time promoting and looking for export markets.
On a further noteChina AI & Semiconductors Rise: US Sanctions Have Failed
Huawei, SMIC, 7nm, H800, Domestic AI Capabilities, ASML, MediaTek, Qualcomm, Apple, RF Capabilities, Potential Sanctionswww.semianalysis.com
Every time Westerners talk about how the Chinese cannot "innovate", I have to conclude that they think the Chinese as a "race" have physiological hurdles stopping them from doing this, because otherwise I cannot wrap my head around how they might think this applies when there's no material basis that would stop them from doing so.
Now you'd think this to be a very imbecilic attitude from the lowest rungs of western society, but as I have gleaned time and again from reading historical texts regarding specific dominating zeitgeists, idiotic temporal ideological blind spots and biases have almost a universal radiation in the populations of different cultures and these biases tend to influence policy in a somewhat surreptitious manner in the decisions of unsuspecting rational actors.
He s right.Peter Zeihan claims that Huawei's 7nm Chip Isn't a Big Chinese Breakthrough, Do they guys think he has a point or is he full of shit?
When it comes to PRC, he is usually full of shit. But I will still listen.
You right.He s right.
But we should not forget Taiwanese (another branch of Chinese) had 2 nm chip....
The article I posted is written by either semiconductor analysts or former ASML employees and condensed matter physicists or based on their data and they know the architecture and have tested it extensively and they think it's legit and good and on par with the best Qualcomm had to offer two years ago. If we consider bias, they might be even a bit biased towards the west, but they try to be as objective as possible. What the fuck does Peter Zeihan know about semiconductors?Peter Zeihan claims that Huawei's 7nm Chip Isn't a Big Chinese Breakthrough, Do they guys think he has a point or is he full of shit?
Full of shit. It is a big breakthrough which is why the US opened a probe to find out how the Chinese succeeded.Peter Zeihan claims that Huawei's 7nm Chip Isn't a Big Chinese Breakthrough, Do they guys think he has a point or is he full of shit?