Tensions Mount as Major Chip Players Jockey for US Funding

News
By
published

Intel, TSMC and others are sniping at each other in meetings and filings.

Intel fab in Arizona.
The construction site for Intel's Fab 52 and Fab 62 in Chandler, Ariz. (Image credit: Intel)

Chipmakers are out for blood, building up intensive lobbying operations to angle for federal subsidies under the CHIPS and Science Act, designed to bulk up the semiconductor manufacturing industry in the  United States. A report from the New York Times depicts a fight with Intel and other companies headquartered in the U.S. against TSMC, which is headquartered in Taiwan, and its clients such as AMD.

The sniping has come from both sides, the Times reports. Intel asked government officials how much money from U.S. taxpayers should be going to its competitiors, which are based outside of the country, and suggested that American technology and intellectual property could end up elsewhere. Additionally, the report names Intel and other U.S. companies like GlobalFoundries and SkyWater Technology as wondering whether foreign firms' U.S. factories would be able to continue producing chips "in the event of a crisis int heir home country," likely referring to the idea of a potential conflict between China and Taiwan.

But competitors have suggested that Intel is a risky bet, and that Intel has a ways to go to catch up to its rivals. TSMC has committed to a $40 billion investment in fabs in Arizona, which are set to come online in 2026 with 3 nanometer-class nodes (Apple is a rumored customer). The company, which also counts AMD as a major client, wrote in a filing that "preferential treatment based on the location of a company's headquarters is not an effective or efficient use of the grant."

In a March filing, AMD suggested that competing companies could use the money to build factories, but not immediately make them operational with the equipment needed to make chips .

"Any facility receiving federal assistance must be operational upon completion of construction," AMD wrote. "A facility that sits idle or is held in reserve for demand increases should immediately forfeit any federal funds." This is likely in reference to Intel, which is attempting to make its name as a foundry service. 

An intel spokesperson didn't comment on AMD's jabs, but the Times says that they defended CEO Pat Gelsinger's "smart capital" policies, suggesting that Intel could build shells for factories and then get them up and running to match the demands of the market. But Intel suggested that while that strategy is being followed in fabs in Arizona, Ohio and New Mexico, the plan is to have working factories, not "just building shells."

And that's all without universities and other research institutions trying to get in on the action. The Times reports that President Joe Biden's administration is planning on releasing the "ground rules for applications" next week, with grants possibly starting this spring.

No matter how the U.S. government decides to dole out money, it will have a challenge in kickstarting a domestic semiconductor industry, which would include getting enough workers, raw materials, research and development, manufacturing capacity across the country. TSMC founder and former CEO Morris Chang previously jabbed to Congresswoman Nancy Pelosi that "fifty billion dollars – well, that’s a good start."

For more, including reporting on TSMC's relationship to the U.S. military and a variety of other organizations jockeying for grants, check out the Times' full report.


Andrew E. Freedman

Andrew E. Freedman is a senior editor at Tom's Hardware focusing on laptops, desktops and gaming. He also keeps up with the latest news. A lover of all things gaming and tech, his previous work has shown up in Tom's Guide, Laptop Mag, Kotaku, PCMag and Complex, among others. Follow him on Threads @FreedmanAE and Mastodon @FreedmanAE.mastodon.social.

See more CPUs News
18 Comments Comment from the forums
  • Great way to make everyone else cautious of what US will do. I'd be really reluctand to go into any kind of serious business with US based companies after any such statement, let's add economy war with China. They likelihood of agreements going the wrong way when things don't work out how they wanted is just too high.
    Reply
  • bit_user
    If the goals are clear and the bidding process is well-managed, competition should deliver better value for the taxpayers' investment.

    The danger is that we'll build a bunch of fab capacity that will sit idle and decay. The original idea seems to be that we want to sustain a certain amount of fab capacity in the US, and taxpayers need to understand and support the kind of long-term commitment that could entail. I worry this last bit of legwork has barely begun, and yet the lack of such support is already starting to jeopardize the initial investment, as companies like Intel seem to be looking for ways to avoid or hedge against building over-capacity.
    Reply
  • PlaneInTheSky
    Why would the US taxpayer fund Taiwanse chip companies like TSMC ?

    Is the Taiwanese taxpayer funding US chip companies ? I think not.
    Reply
  • Giroro
    Intel and TSMC have plenty of money and bloat to build their own factories. If these companies are too shortsighted to diversify their supply chains, then we only need to enable their hubris to reach it's natural conclusion.

    We should be investing in lean start-ups and new technology. Anybody who has a realistic shot at disrupting the monopoly. Once there is a diverse and competitive marketplace, then the domestic supply chain problem will solve itself.
    50x $1 Billion companies have a far, far, far greater chance of pushing forward national interests than feeding a groaning pig in the muck.

    In the very least, Intel and TSMC need to finally be forced to prove why exactly they keep saying a single fab ~$10B should cost more than some entire industries, let alone individual companies. Then we should cure that problem at it's source, before throwing good money after bad. All we have at this point is a vague hope that we can temporarily alleviate a single symptom of a deeply rooted cancer.

    Just for fun, here are some companies that a government could buy for $50Billion, which would provide a far better return on investment.
    All of Global Foundries ($35.11 Billion market cap)
    STMicroelectronics ($44.3B)
    Microchip Technologies ($44.49B)
    SK Hynix ($49.19 B)
    Controlling interest in Micron ($64.5B cap)
    Atmel ($3.45B)
    Keysight ($28.7B)
    Foxconn ($46.95B)
    Baidu (49.46B)
    Motorola Solutions ($44.41B)
    Panasonic ($20.5B)
    Toshiba ($13.72B)
    Honda ($43.68B)
    US Silica ($830M)
    Amphenol ($46.17B)
    Halliburton ($32.91B)
    BAE Systems ($33.19B)
    Nintendo ($46.28B)Realistically, there's fewer than 350 companies on earth which could not have been bought for $50 Billion.
    Reply
  • bit_user
    PlaneInTheSky said:
    Why would the US taxpayer fund Taiwanse chip companies like TSMC ?
    The funding would be for TSMC fabs built on US soil.

    According to the article, Intel or their representatives argue that without all the IP based in the US, such fabs are in jeopardy of having their operations disrupted, should something happen at TSMC's HQ. That's a legitimate concern, if the idea is to build a resilient silicon supply chain, and one that can and should probably be mitigated against.

    PlaneInTheSky said:
    Is the Taiwanese taxpayer funding US chip companies ? I think not.
    US companies aren't building fabs it Taiwan, last I heard.
    Reply
  • Giroro
    bit_user said:
    If the goals are clear and the bidding process is well-managed, competition should deliver better value for the taxpayers' investment.

    The danger is that we'll build a bunch of fab capacity that will sit idle and decay. The original idea seems to be that we want to sustain a certain amount of fab capacity in the US, and taxpayers need to understand and support the kind of long-term commitment that could entail. I worry this last bit of legwork has barely begun, and yet the lack of such support is already starting to jeopardize the initial investment, as companies like Intel seem to be looking for ways to avoid or hedge against building over-capacity.
    It shouldn't be the role of the government/taxpayer to spend an infinite amount of direct funding to sustain a failed business model on life support, forever.
    It should be the role of the government to break up monopolies, protect the "little guy" from exploitation of the legal system, and foster a regulated marketplace where the the industries we need can profit and thrive on their own merits.

    It doesn't make sense to throw money at any company/companies who have openly stated they are only interested in doing the bare minimum to milk the taxpayer for free money, instead of actually reinvesting that funding to build out a profitable domestic business, which could sustain itself.
    Reply
  • Eximo
    Giroro said:
    Just for fun, here are some companies that a government could buy for $50Billion, which would provide a far better return on investment.
    All of Global Foundries ($35.11 Billion market cap)
    STMicroelectronics ($44.3B)
    ---


    Many of those don't really make silicon directly (most order from Samsung, TSMC, etc, just assembly and PCB fab and the like, or further outsource assembly to companies like Foxconn.

    And all of the chip fabs rely on ASML for the lithography and production hardware they need to make high end chips.

    Basically taking the companies that already have expertise and funding them to buy additional equipment and set up new production lines quickly vs taking a company with zero expertise and waiting 15 or 20 years.

    You may also be forgetting what tends to happen to factories over time. Global Foundries was sold off into its own entity. If a US based fab proves unprofitable, they may sell it and it could be acquired by a local company that could take it in a different direction.

    I won't say these companies will necessarily pay appropriate taxes, since they really won't, but there is an economic benefit regardless.

    #1 Local jobs, which leads to income tax
    #2 increased sales of electronics, which leads to sales tax
    #3 Chips made on local soil might encourage local based assembly, more jobs (rather than build chips, ship to other places, assemble, re-import though short term that is very likely)
    #4 Maybe, but not likely, reduced prices
    Reply
  • bit_user
    Giroro said:
    We should be investing in lean start-ups and new technology.
    Not to diminish their role, but it's not lean startups that crank out the cutting-edge wafers, nor could they ever amass the expertise, IP, and capacity in a useful time horizon.

    Semiconductor fabrication is an extraordinarily capital-intensive undertaking, and it's about scale. While startups could contribute IP, techniques, and technology, they will never make up the backbone of semiconductor fabrication in the US or anywhere else.

    Giroro said:
    50x $1 Billion companies have a far, far, far greater chance of pushing forward national interests
    $1B is hardly enough to stand up a single wafer production line, and then where are you going to get the expertise and IP to design the process for it to run?

    Giroro said:
    In the very least, Intel and TSMC need to finally be forced to prove why exactly they keep saying a single fab ~$10B should cost more than some entire industries, let alone individual companies.
    It's not just them. Look at Samsung, SK Hynix, Micron, and Global Foundries. If you really cared to understand the costs involved, the information is out there.


    Giroro said:
    Just for fun, here are some companies that a government could buy for $50Billion, which would provide a far better return on investment.
    All of Global Foundries ($35.11 Billion market cap)
    Okay, that's great if we want to run the entire tech industry on Ryzen 2000-class CPUs, because 12nm is where Global Foundries stopped their node development.

    Also, why do you want to nationalize any of these companies? Capitalism has worked well, so far, but the problem we're facing is that US-based semiconductor makers are up against state-backed competition abroad.

    Giroro said:
    STMicroelectronics ($44.3B)
    SK Hynix ($49.19 B)
    Foxconn ($46.95B)
    Baidu (49.46B)
    Panasonic ($20.5B)
    Toshiba ($13.72B)
    Honda ($43.68B)
    BAE Systems ($33.19B)
    Nintendo ($46.28B)
    Good luck. These are foreign companies.

    Giroro said:
    Microchip Technologies ($44.49B)
    Controlling interest in Micron ($64.5B cap)
    Atmel ($3.45B)
    Keysight ($28.7B)
    Motorola Solutions ($44.41B)
    US Silica ($830M)
    Amphenol ($46.17B)
    Halliburton ($32.91B)
    I don't see how this solves the problem. You're just throwing names around. The names you listed which even do chip production are mostly on older nodes and specialized towards mixed-signal, etc.

    Giroro said:
    Realistically, there's fewer than 350 companies on earth which could not have been bought for $50 Billion.
    The Congress decided to invest in securing our domestic semiconductor production capacity, and you want to spend the money on magic beans? Even if we do get a giant beanstalk out of them, that won't help us if/when TSMC goes offline.
    Reply
  • bit_user
    Giroro said:
    It shouldn't be the role of the government/taxpayer to spend an infinite amount of direct funding to sustain a failed business model on life support, forever.
    It's not infinite, and governments (the US and others) support other vital industries like agriculture.

    You just have to think rationally about the alternatives.
    Spend less than 0.1% of GDP to support domestic semiconductor production.
    Be plunged into a chip shortage that would make the last couple years look tame, by comparison.
    Option #3 might be to stand up an international supply chain initiative, but that would still take financial backing and provide fewer benefits for the US economy.

    Giroro said:
    It should be the role of the government to break up monopolies, protect the "little guy" from exploitation of the legal system, and foster a regulated marketplace where the the industries we need can profit and thrive on their own merits.
    Yeah, but it can walk & chew gum, at the same time.

    Giroro said:
    It doesn't make sense to throw money at any company/companies who have openly stated they are only interested in doing the bare minimum to milk the taxpayer for free money, instead of actually reinvesting that funding to build out a profitable domestic business, which could sustain itself.
    Agreed. That's why they need to put forth credible proposals and there need to be benchmarks and milestones to continue receiving the money. The amount of money they'll get isn't enough to 100% fund the fabs we want. They'll still be picking up most of the tab.
    Reply
  • cyrusfox
    Giroro said:
    Intel and TSMC have plenty of money and bloat to build their own factories. If these companies are too shortsighted to diversify their supply chains, then we only need to enable their hubris to reach it's natural conclusion.
    The issue chips act was set to resolve was an unbalanced supply chain that is viewed as a security risk. What caused the supply chain to become so unbalanced... Foreign subsidies. It wasn't long ago you could get a whole fab built on government dime in Korea or Singapore.

    The CHIPS funding as well as the European chips bill is to incentivize local fabrication. Intel and others have successfully lobbied this as a national security issue. While that parts debatable, they have succeeded in the near term as TSMC comes to Arizona and Intel is building in Germany and Ohio. I expect Korea, Taiwan, and other countries will look to compete to retain fab market share and I personally expect a glut of fab capacity. How are they going to populate all this clean room space with tools and people? All this capacity coming online to softening demand is not a profitable picture, perhaps demand picks up as they come online but its going to be challenging to fully utilize all this new capacity.
    Reply