How to Spend $39 Billion
Most CHIPS Act funds are already spent, we just don't realize it
Welcome to SemiLiterate, a guide to the chip industry through the lens of public policy.
BLUF: This post summarizes known and likely recipients of CHIPS Act funding. It finds that over 50% of the $39 billion earmarked for semiconductor manufacturing incentives is already spoken for based on announcements of new factory construction or existing factory expansions.
Introduction
The United States Innovation and Competition Act of 2021 (USICA) passed in the Senate last week. The CHIPS Act, one of the many provisions in the USICA, directs $52 billion of funding towards the semiconductor industry. $39 billion of this funding is to be spent exclusively on building new semiconductor factories (“fabs”), and the Commerce Secretary has publicly estimated there will be 7-10 new fabs that come out of this effort. This post compiles a list of likely recipients of these funds and the implications of this funding.
Current Provisions
With Senate passage of the CHIPS Act, the semiconductor industry is getting closer to its $52 billion pay day. The latest cheat sheet from the Senate indicates:
$39 billion will go towards incentives to build new fabs:
$19 billion in FY22, $5 billion each year FY23-FY26
$2 billion of this is earmarked for “legacy chip production funding”
$10.5 billion will go to the Commerce Department:
$2 billion for a National Semiconductor Technology Center
$2.5 billion for a National Advanced Packaging Manufacturing Program
$500 million for “other related R&D programs”
Follow on funding in FY23 ($2 bil.), FY24 ($1.3 bil.), and FY25 and FY26 ($1.1 bil.)
$2 billion for Department of Defense chip needs, allocated in $400 million increments over 5 years.
$500 million for an “International Technology Security Fund,” allocated in $100 million increments over 5 years.
Expected Recipients
There are already many major semiconductor factory projects in the U.S. that have broken ground, or will break ground shortly, which will receive the lions share of this funding. Starting with the more-obvious ones who are likely going to receive the maximum $3 billion incentive:
TSMC (Arizona) - $3 billion incentive
Samsung (State TBD) - $3 billion incentive
Intel (Arizona) - $3 billion incentive
Micron (Virginia) - $3 billion incentive
Texas Instruments (Texas) - $3> billion incentive
GlobalFoundries (NY) - $3> billion incentive
There are also many companies who are eligible for CHIPS Act funds that will get less than the full $3 billion incentive (they’re not going to build an entirely new chip factory, but maybe they will upgrade existing factory lines from 200mm wafers to 300 mm wafers, for example). These companies in particular might target the pot of money earmarked for “legacy” chip production “essential to the auto industry, the military, and other critical industries.” Below is a list of these companies, the states where their current fabs are located, and thus where they’re likely to build/expand operations using CHIPS Act funds:
SkyWater Technology Foundry (MN or FL) - $1< billion
Qorvo (NC or TX) - $1> billion
Microchip (AZ or OR) - $1> billion
ON Semi (ID or OR) - $1> billion
Analog Devices (CA) - $1> billion
Honeywell (MN) - $1> billion
Broadcom (PA or CO) - $1> billion
NXP Semiconductor (AZ or TX) - $1> billion
Infineon - (TX or CA) - $1> billion
TowerJazz - (CA or TX) - $1> billion
CREE - (NC) - $1> billion
MACOM - (MA or CA) - $1> billion
Implications
That’s a lot of money to send out the door in a very short period of time. In fact, the back-of-the-envelope numbers indicate over 50% (~$20 billion) of the CHIPS Act funds for semiconductor manufacturing incentives are already spoken for, and it isn’t even law yet. That’s actually a pretty conservative number and there are a couple of other things that could revise this calculation upwards quickly:
Unexpected Investments:
Kioxia (Japan), SK Hynix (South Korea), UMC (Taiwan) and Renesas (Japan) are all major semiconductor manufactures that are theoretically eligible for these incentives but have not indicated any interest in expanding their U.S. presence. If they were to do so, any one of them would be able to take advantage of up to $3 billion in funding and draw down the pot of funds even further.
Preferential Funding for DoD-affiliates:
That Axios graphic linked to above actually undercounts the number of fab lines in the U.S. substantially because it lacks all of the in-house fabs operated by big defense contractors: BAE Systems, HRL Laboratories, Northrup Grumman, Raytheon, SRI International and TSI Semiconductors are all accredited for providing Trusted Foundry Services (pdf) by DOD/DMEA. There will be substantial competition for DOD earmarked funds which may put pressure on the larger pot of money.
Fuzzy Eligibility:
If these funds are made available to manufacturers of semiconductors other than just integrated circuits, there could be further increased demand: companies like Vishay Intertechnology and Rogue Valley Microdevices which make MEMS (microelectronic mechanical systems) might get in on the funding too.
There are also several universities that operate fabs. If the small fabs at places like MIT Lincoln Labs and University of Florida are eligible for funding, this pot of money goes away fast.
Next Steps/Conclusion
The United States Innovation and Competition Act of 2021 is now headed to the House. It seems relatively safe to assume that the top-line number of $52 billion will remain mostly unchanged as this area/industry enjoys a moment of bipartisan support. But, there are only a few weeks before summer recess and the legislative calendar is already full, so the timing of its passage and actual signing in to law remains unknown (September, maybe?).
The CHIPS Act focus on fabrication over innovation is understandable but unfortunate. Most industry-watchers do not believe that Intel or TSMC is going to invent the future of the semiconductor industry. Yet this Act directs the vast majority of its funding to those (and similar) companies. Hopefully these large companies will use their CHIPS Act funds to see the CMOS roadmap/Moore’s Law to its very end and reinvest some of their profits in in-house R&D/workforce development that perpetuates their competitiveness in the short to medium term.
However this Act does not guarantee any long term competitiveness for the U.S. industry. For that to happen, the other Commerce Department programs like the NSTC and advanced packaging initiative will have to strategically allocate a relatively small pot of money to create the incentives, commercialization roadmaps, and workforce necessary to invent the future of a non-CMOS semiconductor industry. A later post will focus on how Commerce could most effectively stand up these programs to increase the probability the innovations they fund meaningfully increase the U.S. industry’s competitiveness.