Chips act should an investment, not a grant

[Arthur Hanson]

With Gelsinger threatening to scale back plans for an Ohio fab if not passed, the "Chips Act" should strictly be an investment and not a grant. On this note, it would be far better to invest in TSM in the form of stocks or warrants, than the welfare that Gelsinger wants. If the Chips Act is truly an investment, the taxpayer should have an ownership interest, in whatever company receives taxpayer funds that most have worked hard for. If it is welfare, it should be called the "Throw Money at Failure Act". This is how all government subsidies should be treated, not just in the tech sector. Any facts, thoughts or opinions are appreciated.

 

[lilo777]

With Gelsinger threatening to scale back plans for an Ohio fab if not passed, the "Chips Act" should strictly be an investment and not a grant. On this note, it would be far better to invest in TSM in the form of stocks or warrants, than the welfare that Gelsinger wants. If the Chips Act is truly an investment, the taxpayer should have an ownership interest, in whatever company receives taxpayer funds that most have worked hard for. If it is welfare, it should be called the "Throw Money at Failure Act". This is how all government subsidies should be treated, not just in the tech sector. Any facts, thoughts or opinions are appreciated.

Are we at it again? Lululemon is a better investment than TSM. Should US Government invest in them instead? In general, governments do not act as investors. There are some exceptions when it comes to sovereign funds but US is a minor player in this domain with just $240B combined in these funds:
APFC / NMSIC / PWMTF / SIFTO / IEFIB / PSF / PUF / ATF / NDLF / LEQTF / CSF / WVFF (source). But then again, for these purposes the funds would be much much better off investing in Lululemon:

Who needs TSM (or INTC to this matter) when they can invest in LULU?

CHIPS act has nothing to do with investing (at least not in the stock market investing sense). It has everything to do with state security and, for obvious reasons, TSM (and other foreign companies) have much lower priority for US government than US companies.

 

[blueone]

I disagree. Giving the USG an equity position in US companies is a really bad idea. The Chips Act is intended to give US companies the incentive to build fabs in the US when it is clearly substantially cheaper to build and operate them in Asia, and works mostly by issuing tax credits. Personally, I'm not in favor of government subsidies, in this case because I believe much of the extra cost of building fabs in the US is a self-inflicted by regulations and taxes. Nonetheless, the US Congress seldom fixes the root causes of problems and tries to implement what promoters believe are easier to sell workarounds. IMO, The Chips Act and the Fabs Act are workarounds, intended to equalize the cost factors between the US and Asia enough to, well, bribe in a way, companies into building fabs in the US rather than in more profitable, lower cost countries. Europe seems to be in a similar position, as does Japan.

Personally, I do not understand the controversy over these two Acts. State and local governments in US have been giving factories of all sorts huge incentives to locate in their states and cities. Intel, for example, gets property tax breaks in Oregon because if they didn't it would be really uneconomical to locate fabs there. And now that they did over the years Oregon is the home of Intel's largest site, with over 22,000 employees. Intel is there because Oregon, Washington County, and Hillsboro made it economically possible, and there's cheap power, lots of water, and cheaper land than California. Now Ohio has decided to be even more accommodating than Oregon. Ohio also has a much greater population and larger universities than Oregon. The tax breaks the states and local governments give to companies get approved because they're acting in their constituents best interest. The progressives in the US Congress, and the other big mouths in socialist politics like Robert Reich, are all up in arms over credits at the national level that already occur at lower levels. It also fascinating that these same progressives and socialists are all in favor of subsidies for what's ever on their political agendas, like electrics cars (which go mostly to the relatively wealthy), and solar and wind power projects (ditto on recipients), being the two most expensive examples.

As a nation we have to decide if we want fabs built in the US or not. If we do we'll have to pay, or fix the underlying reasons why it is expensive to build here quickly, and I think that's impossible.

 

[lilo777]

I disagree. Giving the USG an equity position in US companies is a really bad idea. The Chips Act is intended to give US companies the incentive to build fabs in the US when it is clearly substantially cheaper to build and operate them in Asia, and works mostly by issuing tax credits. Personally, I'm not in favor of government subsidies, in this case because I believe much of the extra cost of building fabs in the US is a self-inflicted by regulations and taxes. Nonetheless, the US Congress seldom fixes the root causes of problems and tries to implement what promoters believe are easier to sell workarounds. IMO, The Chips Act and the Fabs Act are workarounds, intended to equalize the cost factors between the US and Asia enough to, well, bribe in a way, companies into building fabs in the US rather than in more profitable, lower cost countries. Europe seems to be in a similar position, as does Japan.

Personally, I do not understand the controversy over these two Acts. State and local governments in US have been giving factories of all sorts of huge incentives to locate in their states and cities. Intel, for example, gets property tax breaks in Oregon because if they didn't it would be really uneconomical to locate fabs there. And now that they did over the years Oregon is the home of Intel's largest site, with over 22,000 employees. Intel is there because Oregon, Washington County, and Hillsboro made it economically possible, and there's cheap power, lots of water, and cheaper land than California. Now Ohio has decided to be even more accommodating than Oregon. Ohio also has a much greater population and larger universities than Oregon. The tax breaks the states and local governments give to companies get approved because they're acting in their constituents best interest. The progressives in the US Congress, and the other big mouths in socialist politics like Robert Reich, are all up in arms over credits at the national level that already occur at lower levels. It also fascinating that these same progressives and socialists are all in favor of subsidies for what's ever on their political agendas, like electrics cars (which go mostly to the relatively wealthy), and solar and wind power projects (ditto on recipients), being the two most expensive examples.

As a nation we have to decide if we want fabs built in the US or not. If we do we'll have to pay, or fix the underlying reasons why it is expensive to build here quickly, and I think that's impossible.

As usual, it's the partisan divide that is a culprit. The congress (the democrats) put some of their priorities in their version (like worker training) whereas the senate (the republicans) is against it. Here is a good article that explains the differences.

 

[blueone]

And Intel just announced the delay of the groundbreaking for the facilities in Ohio. Let the partisan yelling begin.

 

[Daniel Nenni]

Without the chip shortage narrative the politicians will find something else shinney to chase. Bringing non competitive fabs to the US will not solve our problem. TSMC is now motivated to build fabs around the world due to the geopolitical landscape so the US should keep pushing that angle and get TSMC to build more fabs like they are in AZ. The US should be in a good bargaining position with Taiwan for goods and services.

TSMC is probably against the CHIPS act anyway. Less CAPEX for Intel means more wafer orders from Intel.

 

[prime007]

To summarize a Washington Post article (paywall)...
There is strong support in Congress for the semiconductor subsidies, and for increased spending on the National Science Foundation and other research efforts, but agreement breaks down over other policies.

Congressional aides said it is likely the final bill will more closely resemble the Senate legislation because it passed with bipartisan support, while the House bill had just one Republican supporter, Rep. Adam Kinzinger (Ill.)

House Democrats have had to make concessions along the way on trade and climate provisions that they included in their bill, said the person familiar with Tuesday’s congressional leadership meeting.

The House bill’s expansion of the Trade Adjustment Assistance Program, which provides aid to workers who lose jobs as a result of offshoring and other adverse effects of foreign trade, is a particular nonstarter for Republicans, congressional aides say.

Another provision sparking debate would require the federal government to screen and at times prohibit certain U.S. investment in China. The measure, proposed by Sens. Robert P. Casey Jr. (D-Pa.) and John Cornyn (R-Tex.), has some bipartisan support in both chambers but has nonetheless “been one of the more contentious issues to reach agreement on,” said Stephen Ezell, vice president for global innovation policy at the Information Technology and Innovation Foundation.

Todd Tucker, director of industrial policy and trade at the Roosevelt Institute think tank, said the House bill includes important provisions aimed at protecting U.S. supply chains from external shocks, such as the pandemic, which caused widespread shortages of medical goods.

Among other steps, the bill would establish an Office of Manufacturing Security and Resilience at the Commerce Department with $500 million in appropriations, tasked with tracking the availability of goods and services in real time and promoting critical manufacturing in the United States and allied nations, Tucker said.

Source (paywall): https://www.washingtonpost.com/technology/2022/06/22/chips-act-funding-congress/

 

[hist78]

With Gelsinger threatening to scale back plans for an Ohio fab if not passed, the "Chips Act" should strictly be an investment and not a grant. On this note, it would be far better to invest in TSM in the form of stocks or warrants, than the welfare that Gelsinger wants. If the Chips Act is truly an investment, the taxpayer should have an ownership interest, in whatever company receives taxpayer funds that most have worked hard for. If it is welfare, it should be called the "Throw Money at Failure Act". This is how all government subsidies should be treated, not just in the tech sector. Any facts, thoughts or opinions are appreciated.

Like many things we do today, we search them around the world (including US) to find the best and the worst practices from other companies and other countries. In this Chips Act I don't think we learned enough from the best practice.

Why we expect injecting free taxpayers money into US semiconductor industry without any strings attached will have great success? IMO, it's likely those huge grants recipients will be the the same old faces. Some of them will perform nicely as they are doing today. And some of them will perform badly, just like they are doing today.

Free taxpayers money will not alter the quality and integrity of those companies' executives and board of directors. But their quality and integrity will ultimately determine the outcome.

I think Arthur is posing a very good thinking direction: are we in such hopeless and helpess situation that nothing we can do to ensure the outcome of this billions dollar spending? If there is nothing we can do, then is this whole Chips Act a wishful thinking act?

 

[blueone]

As usual, it's the partisan divide that is a culprit. The congress (the democrats) put some of their priorities in their version (like worker training) whereas the senate (the republicans) is against it. Here is a good article that explains the differences.

Thanks. One of the better articles I've read on NPR, because it doesn't have any discernible bias to it. I really dislike the "directorate" focusing on computing technology leadership, or the "technology hubs", whatever they are. Since when has the US government provided leadership thinking in anything?

 

[hist78]

Without the chip shortage narrative the politicians will find something else shinney to chase. Bringing non competitive fabs to the US will not solve our problem. TSMC is now motivated to build fabs around the world due to the geopolitical landscape so the US should keep pushing that angle and get TSMC to build more fabs like they are in AZ. The US should be in a good bargaining position with Taiwan for goods and services.

TSMC is probably against the CHIPS act anyway. Less CAPEX for Intel means more wafer orders from Intel.

During the recent annual shareholders meeting, TSMC Chairman Mark Liu stated that the status of Chips Act will not alter TSMC fab construction in Arizona.

I believe the delay of Chips Act might not be the only reason for Intel to postpone the Ohio fab groundbreaking. It's worth discussing in another thread.

 

[blueone]

Without the chip shortage narrative the politicians will find something else shinney to chase. Bringing non competitive fabs to the US will not solve our problem. TSMC is now motivated to build fabs around the world due to the geopolitical landscape so the US should keep pushing that angle and get TSMC to build more fabs like they are in AZ. The US should be in a good bargaining position with Taiwan for goods and services.

TSMC is probably against the CHIPS act anyway. Less CAPEX for Intel means more wafer orders from Intel.

I doubt Congress will be happy with just bringing TSMC fabs to the US. TSMC fabs are still Taiwanese fabs, and most of what I've read implies Congress is interested in expanding US-based chip manufacturing capabilities. If China invaded Taiwan could TSMC fabs continue to function independently? I doubt it.

Anyway, Intel and Samsung are promising SOTA fabs. I can't believe anyone in Congress or on their staffs are smart enough to really determine if either company has the actual capability to succeed.

 

[Arthur Hanson]

Lululemon is not a critical company, you cannot

Are we at it again? Lululemon is a better investment than TSM. Should US Government invest in them instead? In general, governments do not act as investors. There are some exceptions when it comes to sovereign funds but US is a minor player in this domain with just $240B combined in these funds:
APFC / NMSIC / PWMTF / SIFTO / IEFIB / PSF / PUF / ATF / NDLF / LEQTF / CSF / WVFF (source). But then again, for these purposes the funds would be much much better off investing in Lululemon:

View attachment 809
Who needs TSM (or INTC to this matter) when they can invest in LULU?

CHIPS act has nothing to do with investing (at least not in the stock market investing sense). It has everything to do with state security and, for obvious reasons, TSM (and other foreign companies) have much lower priority for US government than US companies.

Are we at it again? Lululemon is a better investment than TSM. Should US Government invest in them instead? In general, governments do not act as investors. There are some exceptions when it comes to sovereign funds but US is a minor player in this domain with just $240B combined in these funds:
APFC / NMSIC / PWMTF / SIFTO / IEFIB / PSF / PUF / ATF / NDLF / LEQTF / CSF / WVFF (source). But then again, for these purposes the funds would be much much better off investing in Lululemon:

View attachment 809
Who needs TSM (or INTC to this matter) when they can invest in LULU?

CHS act has nothing to do with investing (at least not in the stock market investing sense). It has everything to do with state security and, for obvious reasons,L TSM (and other foreign companies) have much lower priority for US government than US companies.

Lululemon is not a critical company to the world economy and that is totally off the subject. I gladly accept input, just make sure it applies to the tech sector and or is critical to the world economy, which Lululemon is not.

 

[lilo777]

Lululemon is not a critical company, you cannot



Lululemon is not a critical company to the world economy and that is totally off the subject. I gladly accept input, just make sure it applies to the tech sector and or is critical to the world economy, which Lululemon is not.

You can't have it both ways. If the government is investing to make profits, Lululemon is as good investment as they come. If they are "investing" into country's security, they are going to give money to US companies, not TSMC. Your logic is simply neither here nor there. US government is not investing in "technical sector". They are investing in US technical sector. Why do you keep suggesting that they need to give money to TSMC simply escapes me.

 

[hist78]

You can't have it both ways. If the government is investing to make profits, Lululemon is as good investment as they come. If they are "investing" into country's security, they are going to give money to US companies, not TSMC. Your logic is simply neither here nor there. US government is not investing in "technical sector". They are investing in US technical sector. Why do you keep suggesting that they need to give money to TSMC simply escapes me.

DoD and several federal agencies encouraged/invited TSMC to start a fab in Arizona. That's based on a very simple reason: TSMC is a critical part of the supply chain that provides essential semiconductors to US national security related applications.

There are plenty of DoD suppliers who are foreign companies. From DoD's point of view, as long as those vendors are from friendly countries, it is important to have them eventually to manufacture the products on US soil. DoD doesn't plan the weapon systems based on vendors' roadmaps or good intentions. DoD needs the real products to be delivered on time with the features promised. This realistic view of supply chains and weapons system development has served US very well since the end of World War II.

TSMC Arizona fab is a typical case of executing this tried-and-true policy.

I think the original TSMC subsidies was planned to be included the defense budget. But due to the nature of the wonderful democratic system and with 535 representatives and 100 senators' input, it grew to the so-called Chips Act and it became a legislation nightmare.

TSMC already said regardless of the status of Chips Act, TSMC Arizona fab will be moving forward.

If Intel believes the taxpayers money from Federal government (Ohio already committed $2.1 billion) is a make or break factor to build the Ohio fab, they should stop right there and cancel the whole project.

It's a clear indication that this Intel Ohio fab is not sustainable in the long run. It can drag the whole Intel down.

Several years ago during a Berkshire Hathaway annual shareholders meeting, Warren Buffett said he won't invest in a business if that business depends on government subsidies or tax credit to survive.

Intel is building 2 new fabs in Arizona and 2 new fabs in Germany. Intel is also expanding manufacturing capacity in New Mexico, Ireland, Israel, and Malaysia. Intel assumes when the fab construction is complete, customers will lineup to get Intel's latest and greatest.

Meanwhile Intel's revenue is shrinking and free cash flow is under enormous pressure due the heavy capital investment and negative growth. It's about time for Intel to put business and financial discipline in practice. Does Intel really need the Ohio fab?

 

[lilo777]

DoD and several federal agencies encouraged/invited TSMC to start a fab in Arizona. That's based on a very simple reason: TSMC is a critical part of the supply chain that provides essential semiconductors to US national security related applications.

There are plenty of DoD suppliers who are foreign companies. From DoD's point of view, as long as those vendors are from friendly countries, it is important to have them eventually to manufacture the products on US soil. DoD doesn't plan the weapon systems based on vendors' roadmaps or good intentions. DoD needs the real products to be delivered on time with the features promised. This realistic view of supply chains and weapons system development has served US very well since the end of World War II.

TSMC Arizona fab is a typical case of executing this tried-and-true policy.

I think the original TSMC subsidies was planned to be included the defense budget. But due to the nature of the wonderful democratic system and with 535 representatives and 100 senators' input, it grew to the so-called Chips Act and it became a legislation nightmare.

TSMC already said regardless of the status of Chips Act, TSMC Arizona fab will be moving forward.

If Intel believes the taxpayers money from Federal government (Ohio already committed $2.1 billion) is a make or break factor to build the Ohio fab, they should stop right there and cancel the whole project.

It's a clear indication that this Intel Ohio fab is not sustainable in the long run. It can drag the whole Intel down.

Several years ago during a Berkshire Hathaway annual shareholders meeting, Warren Buffett said he won't invest in a business if that business depends on government subsidies or tax credit to survive.

Intel is building 2 new fabs in Arizona and 2 new fabs in Germany. Intel is also expanding manufacturing capacity in New Mexico, Ireland, Israel, and Malaysia. Intel assumes when the fab construction is complete, customers will lineup to get Intel's latest and greatest.

Meanwhile Intel's revenue is shrinking and free cash flow is under enormous pressure due the heavy capital investment and negative growth. It's about time for Intel to put business and financial discipline in practice. Does Intel really need the Ohio fab?

I am sure you understand that there is more to semiconductor manufacturing than FAB. Unless you want the US to become the place for outsourcing manufacturing jobs (aka Vietnam, Malaysia etc.) and not to have the ability to develop advanced semiconductor processes/tech you can support TSMC instead of supporting US companies. Also, your interpretation of the genesis of the CHIPS act is really suspect. It definitely was not born out of desire to finance TSMC's Arizona FAB. It was a reaction to many things including the fact that since 2015, China has allocated at least $180 billion in support for semiconductor industry. It's a security and competitiveness issue first and foremost and TSMC FAB does very little to address this issue.

 

[hist78]

I am sure you understand that there is more to semiconductor manufacturing than FAB. Unless you want the US to become the place for outsourcing manufacturing jobs (aka Vietnam, Malaysia etc.) and not to have the ability to develop advanced semiconductor processes/tech you can support TSMC instead of supporting US companies. Also, your interpretation of the genesis of the CHIPS act is really suspect. It definitely was not born out of desire to finance TSMC's Arizona FAB. It was a reaction to many things including the fact that since 2015, China has allocated at least $180 billion in support for semiconductor industry. It's a security and competitiveness issue first and foremost and TSMC FAB does very little to address this issue.

"It's a security and competitiveness issue first and foremost and TSMC FAB does very little to address this issue."

I think my logic is very similar to DoD's: support companies who can build advanced semiconductor manufacturing capabilities for national security related applications on US soil. It can be Intel, TI, TSMC, or Samsung. Currently chips designed by US companies and built by TSMC are powering some important US weapon systems. It's a sensible and efficient approach to help TSMC to build an advanced fab in Arizona.

Additionally DoD, DOE and other agencies learned from real-world cases that relying on single company is not the best practice.

Take a look the important DOE Exascale Supercomputer project. While Intel processors/GPU based Aurora system has been delayed for several years, the late comer "Frontier" at Oak Ridge National Lab based on AMD solutions (built by TSMC) just claimed the world #1 fastest computer title.

 

[blueone]

Take a look the important DOE Exascale Supercomputer project. While Intel processors/GPU based Aurora system has been delayed for several years, the late comer "Frontier" at Oak Ridge National Lab based on AMD solutions (built by TSMC) just claimed the world #1 fastest computer title.

That #1 title is on the TOP500 supercomputer list (top500.org), and this is a list of publicly announced supercomputers. Two supercomputers in China are thought to have at least double the 1.1ExaFlop performance of Frontier.

https://www.datacenterdynamics.com/en/news/china-may-be-planning-10-exascale-supercomputers-by-2025/

It should also be noted that the TOP500 list is based on the LINPACK benchmark, which is focused on linear programming, and is well-liked by the popular press because it delivers a single figure of merit. Everyone loves a figure of merit. (Well, not me.)

I also suspect there is also at least one "dark" supercomputer system in the US that could have achieved EF status secretly years ago. As mentioned in the article, the US Dept of Energy (which oversees nuclear weapons development) is spending a lot on exascale software development, and I suspect they must have at least one test system to prove the designs.

 

[lilo777]

"It's a security and competitiveness issue first and foremost and TSMC FAB does very little to address this issue."

I think my logic is very similar to DoD's: support companies who can build advanced semiconductor manufacturing capabilities for national security related applications on US soil. It can be Intel, TI, TSMC, or Samsung. Currently chips designed by US companies and built by TSMC are powering some important US weapon systems. It's a sensible and efficient approach to help TSMC to build an advanced fab in Arizona.

Additionally DoD, DOE and other agencies learned from real-world cases that relying on single company is not the best practice.

Take a look the important DOE Exascale Supercomputer project. While Intel processors/GPU based Aurora system has been delayed for several years, the late comer "Frontier" at Oak Ridge National Lab based on AMD solutions (built by TSMC) just claimed the world #1 fastest computer title.

And what will TSMC FABs be able to build if/when China takes control of Taiwan? R&D is just as important. Is TSMC planning to move their research to US?

 

[hist78]

That #1 title is on the TOP500 supercomputer list (top500.org), and this is a list of publicly announced supercomputers. Two supercomputers in China are thought to have at least double the 1.1ExaFlop performance of Frontier.

https://www.datacenterdynamics.com/en/news/china-may-be-planning-10-exascale-supercomputers-by-2025/

It should also be noted that the TOP500 list is based on the LINPACK benchmark, which is focused on linear programming, and is well-liked by the popular press because it delivers a single figure of merit. Everyone loves a figure of merit. (Well, not me.)

I also suspect there is also at least one "dark" supercomputer system in the US that could have achieved EF status secretly years ago. As mentioned in the article, the US Dept of Energy (which oversees nuclear weapons development) is spending a lot on exascale software development, and I suspect they must have at least one test system to prove the designs.

Whether there is a secret super computer system is a separate topic which we have no ability to discuss, I guess :-(.

What we know is that Aurora super computer (based on Intel CPUs and GPUs) is at least 4 years late. At the end of 2021 Intel Federal division took $300 million write-off although Intel didn't confirm it's related to Aurora.

US is undergoing a multi-decade nuclear weapon system modernizing project that will cost several hundred billion dollar. Super computers are needed for reason in this process and I can imagine the agony DOE project managers have gone through over the Aurora delay.

Let's back to the the DoD and DOE strategy in managing supply chain. They have to be sensible and practical and they are doing just like that. For example, B21 Stealth Bomber will take the first test flight next year. If B21 is using FPGAs from AMD/Xilinx, then DoD has to bring TSMC to Arizona. People can talk other long term domestic semiconductor policy, but DoD and DOE have to deliver real weapon systems that they are assigned to do.

 

[blueone]

Let's back to the the DoD and DOE strategy in managing supply chain. They have to be sensible and practical and they are doing just like that. For example, B21 Stealth Bomber will take the first test flight next year. If B21 is using FPGAs from AMD/Xilinx, then DoD has to bring TSMC to Arizona. People can talk other long term domestic semiconductor policy, but DoD and DOE have to deliver real weapon systems that they are assigned to do.

I'm unfamiliar with US DoD procurement policies, but I wasn't aware that every part of every weapon system had to be made in the US. For the B21, for example, the productions volumes are so low (hundreds of planes, at most) that it would be easy to buy enough chips or even assemblies for the entire lifetime of the electronics. What are the actual procurement policies?