Steven Zhou, CEO, Moov Technologies
Suppose a company, Company X, is looking to acquire an additional fabrication centre (fab); for demonstrative purposes, assume a 300 mm fab. Company X could be an integrated device manufacturer (IDM) that foresees lasting supply chain disruption due to the impact of recent events on global trade or a foundry looking to capitalise on growth in a particular subsector of the market. Regardless of the reason, Company X has a few options for expanding their fabrication centre footprint, namely build, buy or partner.
To understand the relative cost of building versus buying, here are a few recent (2018–19) pricing examples:
- United Microelectronics Corporation (UMC) acquired 100 percent ownership of a foundry venture between itself and Fujitsu for US$520 million. The venture includes a 300 mm fab in Kuwana, Japan, as well as a design centre and HQ.
- ON Semiconductor is to acquire a 300 mm fab in East Fishkill, New York, US, from GlobalFoundries for $430 million.
- Sony plans to spend $918 million on building a 300 mm fab for image sensors in Nagasaki, Japan.
- Bosch is spending approximately $1.1 billion on building a 300 mm fab for silicon carbide automotive chips in Dresden, Germany
- .Texas Instruments has earmarked a site in Richardson, Texas, US, for its second 300 mm fab for analogue wafers, which is expected to cost $3.1 billion.
Looking at these figures, one generally comes to a couple of conclusions, one true and the other potentially misleading. First, the cost of acquiring a fab or building a fab varies. This is true and may reflect factors such as projected output of the fab, process technology, location, etc. Second, acquiring a fab is significantly less expensive than building one. This is potentially misleading because the figures do not necessarily reflect upgrades and/or conversions that a fab’s new owner may be required to undertake in readiness for production.
Building or buying for under $1 bn
Is building a fab limited to companies possessing a $1 billion capital expenditure (capex) budget? Not necessarily. Tooling expenditure accounts for about 80 percent of the cost of building a fab1. For the vast majority of fabs, the secondary market for semiconductor manufacturing equipment (SME) represents a significant opportunity to decrease capital investment by procuring used or refurbished tooling. Based on proprietary data and projections from Moov—a global marketplace for preowned semiconductor tooling—the secondary SME market is worth $50 billion globally, of which only $5–$8 billion per year is currently being moved. This is why Moov is on a mission to raise awareness among manufacturers that this highly advantageous yet unrealised acquisition channel exists.
There are several misconceptions about the secondary SME market that should be dispelled. First, many assume it is mostly 200 mm tooling that is available on the secondary SME market as a result of the industry having transitioned from 200 to 300 mm wafers. As a corollary, there is a misconception around availability of 300 mm tooling on the secondary SME market because the industry has yet to embrace a widespread transition to 450 mm. However, based on an analysis of the most recent 10,000 listings on Moov’s marketplace, the general assumption that only tooling compatible with 200 mm or smaller wafer sizes is available on the secondary market is incorrect. In fact, Moov saw roughly twice as many listings for 300 mm tooling than 200 mm tooling, as well as listings for 450 mm tooling.
Another misconception about the secondary SME market is that the most valuable tooling is not available used. Again, this is incorrect. Looking at the split of front-end versus back-end process tooling for the most recent listings on Moov’s marketplace, front-end tooling accounts for 71 percent and back-end tooling accounts for 29 percent. Furthermore, the most expensive tooling categories of photolithography, diffusion and etching account for nearly 10, over 5 and 11 percent of these listings, respectively.
It is therefore clear that Company X can find tooling on the secondary SME market to equip its new 300 mm fab, but what are the benefits of doing so? Initial savings on capital investment is the obvious reason to buy used. Indeed, Moov sees companies save 80–90 percent on used/as-is tooling and 40–50 percent on refurbished tooling. Translated to overall capex savings for Company X’s new fab (based on 80 percent of investment going toward tooling), this equates to an overall cost reduction of 64–72 percent for buying used and 32–40 percent for buying refurbished. While these are rough estimates intended to demonstrate a point, the opportunity of buying used tooling for all but bleeding-edge fabs is significant.
A better business model
The potential business benefits made possible by the secondary SME market expand beyond initial capex savings. Building a fab means that Company X is able to make decisions on capacity, wafer size, process technology, location, etc. based on the market conditions prior to construction. However, fab construction takes around two years to complete and market conditions may change significantly by the time Company X is ready to begin production.
As well as decreasing costs, buying used or refurbished tooling significantly decreases delivery lead times. Lead times on new tooling can run from 12 to 24 months, but end-to-end procurement and logistics on Moov’s marketplace, for example, can be completed in a matter of weeks. A shorter procurement to production timeframe means less risk of market dynamics changing substantially.
The most significant cost efficiency improvements for leading edge fabs are achieved by reducing bottlenecks2. For example, if a machine breaks down during production, Company X might wait 12–16 months for a new replacement, but during this time, it is operating at reduced capacity and therefore reduced profit. However, the secondary SME market offers a fast alternative to replacing a machine or its parts.
Company X can seek to maximise revenue by predicting demand, which means increasing capacity. Additionally, the company can endeavour to be conservative in predicting demand, since underutilised fabs erode profit margins. The secondary SME market offers Company X the opportunity to respond quickly to changes in demand, for example, by increasing capacity to capitalise on new revenue opportunities without hurting margins or by selling idle tooling. Instead of attempting to solve the unknown in initially predicting capacity, Company X can easily shift capacity after production begins.
Summary
While it is feasible for Company X to save hundreds of millions of dollars in initial capex investment by taking advantage of the secondary SME market, the greater value will still be in the potential for the company to respond nimbly to changes in the global semiconductor market. The value of this flexibility cannot truly be known until business models that capitalise on the secondary SME market emerge.
Moov Technologies
References
1Jamison, M. 300 mm wafer fab contamination control [presentation]. HDR Architecture. Available at: http://bit.ly/39mQIXu
2Heck, S., Kaza, S. and Piner, D. (2011). Creating value in the semiconductor industry. McKinsey on Semiconductors, number 1, Autumn 2011, p.12.