This week might go down in history as one of the most memorable moments of the global semiconductor industry. We were writing down our thoughts as events were unfolding. When we discussed the content for this week's blog, this came into our mind: why not share our train of thought with our readers? So here we go.
The WHAT, WHY and HOW of the White House Semiconductor Factory Plan (May 12th, 2020)
The WHAT part isn't complicated: Wall Street Journal reported that the White House officials are talking with Intel and TSMC to build semiconductor factories in the US
The WHY part is also straight forward: to have better control over the manufacturing of advanced semiconductor products
However, the HOW part is much more complicated, with more questions than answers. The challenge for TSMC is geopolitics:
TSMC is the largest foundry in the world, acting as a contract manufacturer, and has clear leadership in the most cutting edge chips. All the leading processor design companies rely on TSMC to build processors for them, which includes Apple, Nvidia, Qualcomm in the US, and Huawei in China.
When asked about their plan, chairman Mark Liu said: " If TSMC were to build a factory in the US it would be because of consumer demand, but currently it would not be cost-effective to do it." TSMC currently operates most of its factories in Taiwan and has seen no issue serving customers all over the world. TSMC's mission statement is "being everyone's foundry," it is clear that the company wants to stay that way, keep everything strictly a business decision and stay away from geopolitics.
But how long can they keep doing that given its critical role in the supply chain? How would China react if TSMC does build a cutting edge factory in the US and how would the US react if TSMC doesn't? These are the questions that don't have clear answers now.
The challenge for Intel is more about the business model. In the letter sent to Defense Department, Intel's CEO Bob Swan said: "We think it is the best interest of the US and of Intel to explore how Intel could operate a commercial US foundry...realizing a sustainable, commercially viable foundry that is necessary to meet US government security and infrastructure needs."
Unlike TSMC which manufacture processors for others, Intel mostly only manufactures its own chips. The company has been running a foundry business for a long time but wasn't able to scale it up even when it had technology leadership.
The most important reason is the potential conflict of interest. How could companies be comfortable having Intel making their core products when Intel is selling competing products itself? How could the company build a commercially viable business if the major customers are having these concerns?
The company already has most of the factories in the US anyway and building a new factory in the US is not going to make these concerns go away. Does the company have to spin off its semiconductor manufacturing operation into a separate entity to build a commercially viable business? How could the company compete with TSMC, which has been running the foundry business for 30 years and now have technology leadership vs Intel for the first time?
TSMC's Plan to Build a New Fab in Arizona Came Out Earlier Than We Expected (May 14th, 2020)
But after reading through the details, we can't help but think it is an effort to remain neutral, keeping both US and China happy.
The planed capacity is 20,000 wafer per month, this is similar to TSMC's existing fab located in Nanjing, represent ~2% of total capacity. This new capacity is a nice addition but nothing groundbreaking.
More importantly, the planned factory will be manufacturing chips on 5nm, with production starting in 2024. 5nm is the most advanced technology today, but according to TSMC's roadmap, the company will start 3nm production in 2021/2022, and possibly 2nm by 2024. That put this Arizona factory 2 generations behind by the time it goes into production, hardly the most advanced factory the White House was hoping for.
TSMC is still trying hard to remain "everyone's foundry" in a challenging environment, but it is hard to predict how the political environment will change going forward. That is the biggest risk the company is facing as we see it.
Esoterica's statements are not an endorsement of any company or a recommendation to buy, sell or hold any security. For full disclosures, click here.