US Senate approves massive $52B spend on easing the chip shortage

zamardii12

2[H]4U
Joined
Jun 6, 2014
Messages
3,017
How is spending more money going to ease the chip shortage? After all, money is not the problem, capacity is.

" with more foundry capacity arising on home turf in the United States. It will be a slow process as getting new fabs up and running and old ones updated takes time, but it will eventually help ease a global reliance on manufacturing in Taipei, South Korea, and China."

It's basically an investment in foundries and other things to improve chip making in the US.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
32,061
How is spending more money going to ease the chip shortage? After all, money is not the problem, capacity is.

Agreed.

The current chip capacity problem is short term, a combination of pandemic related supply chain issues, higher than average demand due to crypto mining and Intel's production issues.

The pandemic related supply chain issues will go away on their own over time. The other issues are being worked on already by the likes of TSMC increasing capacity and Intel trying to fix their production lines, but these are slow long term projects.

Building new capacity using $52B is also a slow long term process.

In order to resolve the current shortage issue the $52B spending should have been started 5-6 years ago.

Once this actually turns into additional capacity, who knows if we'll still need it?

Normally I would say it is not the governments job to meddle in the free market like this. If additional capacity is needed, prices go up, and the higher prices convince the market to invest in more capacity. Adam Smith's invisible hand takes care of it.

These days though, China is manipulating the market so much that the free market isn't working the way it should, which may force us to take some action as well. I'm not sure this is what I would do if I were long for a day, but I also don't know what the correct answer is.
 
Last edited:

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,366
" with more foundry capacity arising on home turf in the United States. It will be a slow process as getting new fabs up and running and old ones updated takes time, but it will eventually help ease a global reliance on manufacturing in Taipei, South Korea, and China."

It's basically an investment in foundries and other things to improve chip making in the US.

That is only good if the fabs, new ones and updated ones, are capable of producing what is needed in the first place. The tech itself would need to be purchased by whomever is selling it, which is probably TSMC or something like that, I guess. By the time it changes anything, it may no longer even be needed.
 

doz

Supreme [H]ardness
Joined
Jun 16, 2009
Messages
5,623
Won't help the current situation but hopefully we bring more jobs back to American with new plants.

Intel has most of their chip plants stateside now. And more being built.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
4,067
How is spending more money going to ease the chip shortage? After all, money is not the problem, capacity is.
A single modern fab costs upwards of $30B to build and get operational in a 5-year timeframe, that is a massive investment and any company making that sort of investment wants certain assurances, this bill makes those assurances and makes money available to get those fabs built. This is a big move for the US to make especially given the current US/Taiwan/China tensions where if China were to move on Taiwan, the US estimates a complete take over with virtually no means of stopping or slowing the Chinese advance, which would force sanctions which would, in turn, cut off Taiwan manufacturing to the US which would kill huge swaths of the North American tech and manufacturing sectors. This is Step 1, in what is a needed 10 step, multi-trillion-dollar investment to bring US manufacturing to where it is needed to be. They are going to need to pass a good number of these bills and similar ones if they want to not only take back tech manufacturing from China but prevent it from moving to India or the Philippines, or any other "lower" developed nation.
 

1_rick

[H]ard|Gawd
Joined
Feb 7, 2017
Messages
1,831
How is spending more money going to ease the chip shortage? After all, money is not the problem, capacity is.
You ease the shortage by building more foundries, which this is supposed to help entice them to do, supposedly.
 

1_rick

[H]ard|Gawd
Joined
Feb 7, 2017
Messages
1,831
A single modern fab costs upwards of $30B to build and get operational in a 5-year timeframe
There are tons of things that dont' need to be on the latest nodes. Lots of microcontrollers--which are in everything--get by with 90 and 130nm nodes. Only the newest lines are migrating to 40nm. I was curious about automotive ECUs and found an article--no date, unfortunately--that says modern ECUs are typically 40MHz with 1MB flash. That's a $7 chip in 1,000-unit quantity (STM32F412RGY6TR, just as an example of a general-purpose Arm Cortex-M4, and admittedly not an ECU, although the automotive-grade version of that particular chip's only a few cents more.)
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
4,067
There are tons of things that dont' need to be on the latest nodes. Lots of microcontrollers--which are in everything--get by with 90 and 130nm nodes. Only the newest lines are migrating to 40nm. I was curious about automotive ECUs and found an article--no date, unfortunately--that says modern ECUs are typically 40MHz with 1MB flash. That's a $7 chip in 1,000-unit quantity (STM32F412RGY6TR, just as an example of a general-purpose Arm Cortex-M4, and admittedly not an ECU, although the automotive-grade version of that particular chip's only a few cents more.)
Oh yeah completely especially for automotive and most industrial purposes, but those still take years to build and still gonna be more than a few billion to build, It's just good to see the government actually moving on a problem that affects the people for once.
 

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
7,320
It won't help the current situation, but it would help any future problems in the long term (which will almost certainly happen at some point).

Since it's a long term investment, you have to start early, which means now, without absolute knowledge of what could happen in 5 or 10 years.
 

GoodBoy

2[H]4U
Joined
Nov 29, 2004
Messages
2,064
It will help, in maybe 5 years. Better late than never.

Why haven't they poked Global Foundries to get off it's butt and figure out these smaller node processes anyway.. If GF could start making 5nm, it would be a lot quicker than waiting on a new fab to be built.
 

workshop35

Gawd
Joined
Nov 24, 2013
Messages
668
Yeah lets see if the CEOs and other officers of these semiconductor all of a sudden get big raises because they managed to "secure loans" to build stuff.
I work for one of these big semiconductor companies here in the US, and all they want to do is move stuff to what they call "low cost geos". We're currently seeing our test equipment and headcount transfer slowly to Singapore, and I don't think any of this government money is going to change that. All that matters is the almighty quarterly earnings report...
 

jardows

[H]ard|Gawd
Joined
Jun 10, 2015
Messages
2,035
Agreed.

The current chip capacity problem is short term, a combination of pandemic related supply chain issues, higher than average demand due to crypto mining and Intel's production issues.

The pandemic related supply chain issues will go away on their own over time. The other issues are being worked on already by the likes of TSMC increasing capacity and Intel trying to fix their production lines, but these are slow long term projects.

Building new capacity using $52B is also a slow long term process.

In order to resolve the current shortage issue the $52B spending should have been started 5-6 years ago.

Once this actually turns into additional capacity, who knows if we'll still need it?

Normally I would say it is not the governments job to meddle in the free market like this. If additional capacity is needed, prices go up, and the higher prices convince the market to invest in more capacity. Adam Smith's invisible hand takes care of it.

These days though, China is manipulating the market so much that the free market isn't working the way it should, which may force us to take some action as well. I'm not sure this is what I would do if I were long for a day, but I also don't know what the correct answer is.
The larger picture here is the fragility of our current supply chain. The current shortage will work itself out, provided we don't have another interruption. Having more domestically produced electronics ensures that the supply chain is strengthened. Should have been done a long time ago, but sometimes it takes a crisis before people will attempt to come up with a solution.
 

Armenius

Fully [H]
Joined
Jan 28, 2014
Messages
25,808
That is only good if the fabs, new ones and updated ones, are capable of producing what is needed in the first place. The tech itself would need to be purchased by whomever is selling it, which is probably TSMC or something like that, I guess. By the time it changes anything, it may no longer even be needed.
TSMC already started building multiple fabs in Arizona last year. At least one of them is going to be making 5nm chips. Production is expected to start in 2024.
 
  • Like
Reactions: DPI
like this

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,366
TSMC already started building multiple fabs in Arizona last year. At least one of them is going to be making 5nm chips. Production is expected to start in 2024.

If GloFlo has changed as a company, I can see it helping. However, considering how poorly they did with AMD........ Maybe if they are doing non cutting edge stuff but otherwise, I am not seeing them doing well, in my opinion.
 

1_rick

[H]ard|Gawd
Joined
Feb 7, 2017
Messages
1,831
It will help, in maybe 5 years. Better late than never.

Why haven't they poked Global Foundries to get off it's butt and figure out these smaller node processes anyway.. If GF could start making 5nm, it would be a lot quicker than waiting on a new fab to be built.
GF gave up on the small nodes. That's why Zen is on TSMC now.
 

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,597
So how's that medicare for all working out? Too expensive you say? We can't afford it? I wonder why... Why is the public sector funding stuff that the private sector could easily afford? It won't create enough jobs to make it worth it since most of chip manufacturing is automated. Why are we funding this?
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
4,067
It will help, in maybe 5 years. Better late than never.

Why haven't they poked Global Foundries to get off it's butt and figure out these smaller node processes anyway.. If GF could start making 5nm, it would be a lot quicker than waiting on a new fab to be built.
GF gave up on chasing the small nodes, the R&D required to make it happen combined with the upfront capital costs put it well outside their budget.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
4,067
So how's that medicare for all working out? Too expensive you say? We can't afford it? I wonder why... Why is the public sector funding stuff that the private sector could easily afford? It won't create enough jobs to make it worth it since most of chip manufacturing is automated. Why are we funding this?
Because the conservative estimates in the government say China will annex Taiwan some time before 2030 (remember the US does not recognize Taiwan as a country). Their simulations on that war game scenario show China one sidedly bitch slapping the US out of the country, with little feasible means of holding them back or re taking it. Once that happens the US looses access to just about every major fab, so the decision to bring it back locally solves many of the technical, logistical, and economic implications of loosing that access. Hence the national security claims.
 

workshop35

Gawd
Joined
Nov 24, 2013
Messages
668
Because the conservative estimates in the government say China will annex Taiwan some time before 2030 (remember the US does not recognize Taiwan as a country). Their simulations on that war game scenario show China one sidedly bitch slapping the US out of the country, with little feasible means of holding them back or re taking it. Once that happens the US looses access to just about every major fab, so the decision to bring it back locally solves many of the technical, logistical, and economic implications of loosing that access. Hence the national security claims.
They won't bring things back to the US as long as there are cheaper alternatives. Corps will hem and haw about bringing things to the states but the meat of their operations will move to places like India who don't have the regulations we do here.
 

Endgame

Gawd
Joined
Jan 10, 2007
Messages
691
They won't bring things back to the US as long as there are cheaper alternatives. Corps will hem and haw about bringing things to the states but the meat of their operations will move to places like India who don't have the regulations we do here.
They will bring stuff back if they get free money doing it. What you have to watch out for, though, is stuff getting moved out of country that isn’t directly tied to the free money. Ex: Sure, the fabs are here now, but we outsourced all the design to India, or whatever.
 

GoodBoy

2[H]4U
Joined
Nov 29, 2004
Messages
2,064
GF gave up on chasing the small nodes, the R&D required to make it happen combined with the upfront capital costs put it well outside their budget.
Right, but if the Government is going to fund some of the cost anyway, they should take them up on it.
Because the conservative estimates in the government say China will annex Taiwan some time before 2030 (remember the US does not recognize Taiwan as a country). Their simulations on that war game scenario show China one sidedly bitch slapping the US out of the country, with little feasible means of holding them back or re taking it. Once that happens the US looses access to just about every major fab, so the decision to bring it back locally solves many of the technical, logistical, and economic implications of loosing that access. Hence the national security claims.
If China annexes Taiwan (I believe it is China that does not recognize it as a country, while the US does at least as far as many international organizations go), I'm sure they will still want to sell to the US.

But I agree, having China completely in control is fucking bonkers.. we would be stupid NOT to invest in this capability stateside, even if the corporations are too stupid to do it themselves.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
4,067
Right, but if the Government is going to fund some of the cost anyway, they should take them up on it.

If China annexes Taiwan (I believe it is China that does not recognize it as a country, while the US does at least as far as many international organizations go), I'm sure they will still want to sell to the US.

But I agree, having China completely in control is fucking bonkers.. we would be stupid NOT to invest in this capability stateside, even if the corporations are too stupid to do it themselves.
In keeping with its China policy, the U.S. does not support de jure Taiwanindependence, but it does support Taiwan's membership in appropriate international organizations, such as the World Trade Organization.

US is a proud supporter of the “One China” policy. It’s why it was such a big deal when Trump took a phone call from Taiwan’s president.
 

Armenius

Fully [H]
Joined
Jan 28, 2014
Messages
25,808
Don't fabs historically need absolutely huge supplies of clean water? Arizona is not exactly known for being an oasis of liquidity.
Intel has had fabs in Arizona for decades. I haven't really looked into how they do it, but don't fabs recycle their water anyway? So the biggest hurdle would just be getting the initial supply of water to fill their reservoirs.
 

HAL_404

[H]ard|Gawd
Joined
Dec 16, 2018
Messages
1,123
Intel has had fabs in Arizona for decades. I haven't really looked into how they do it, but don't fabs recycle their water anyway? So the biggest hurdle would just be getting the initial supply of water to fill their reservoirs.

excellent point. I'm in New Mexico and we're often in a state of drought (a State of drought :cool:). NM is High Desert and the pattern of change here has been towards Desertification and that will most likely continue on into the future and I would venture a guess that AZ is even more so.
 

NIZMOZ

[H]ard|Gawd
Joined
Oct 23, 2007
Messages
1,994
Agreed.

The current chip capacity problem is short term, a combination of pandemic related supply chain issues, higher than average demand due to crypto mining and Intel's production issues.

The pandemic related supply chain issues will go away on their own over time. The other issues are being worked on already by the likes of TSMC increasing capacity and Intel trying to fix their production lines, but these are slow long term projects.

Building new capacity using $52B is also a slow long term process.

In order to resolve the current shortage issue the $52B spending should have been started 5-6 years ago.

Once this actually turns into additional capacity, who knows if we'll still need it?

Normally I would say it is not the governments job to meddle in the free market like this. If additional capacity is needed, prices go up, and the higher prices convince the market to invest in more capacity. Adam Smith's invisible hand takes care of it.

These days though, China is manipulating the market so much that the free market isn't working the way it should, which may force us to take some action as well. I'm not sure this is what I would do if I were long for a day, but I also don't know what the correct answer is.
It's not short term. It's expected to last a couple of years.
 
Joined
May 27, 2017
Messages
610
I'm not advocating more government spending, but by the time this becomes fruitful it'll probably be just in time for another shortage crisis.
 

Endgame

Gawd
Joined
Jan 10, 2007
Messages
691
I knew I saw this linked somewhere, quick google search turned it up.

Glenn O'Donnell, vice president and research director at analyst firm Forrester, told CNBC that chip fabrication plants "recycle water religiously," adding that it's a bit like a swimming pool in an enclosed building.

"You need a lot to fill it, but you don't have to add much to keep it going," he said. "Also, being in an enclosed space, a lot of the water that evaporates can be captured with a dehumidifier and returned to the pool. The fabs will do similar things with their own water usage."

Intel notes on its website that it is striving to achieve "net positive water use" in Arizona and that it has funded 15 water restoration projects that aim to benefit the state. "Once fully implemented, these projects will restore an estimated 937 million gallons each year," the company says.

https://www.cnbc.com/2021/06/04/why...g-water-dependent-chip-plants-in-arizona.html
 

1_rick

[H]ard|Gawd
Joined
Feb 7, 2017
Messages
1,831
Intel has had fabs in Arizona for decades. I haven't really looked into how they do it, but don't fabs recycle their water anyway? So the biggest hurdle would just be getting the initial supply of water to fill their reservoirs.
There's some articles out there talking about this. Apparently they recycle massively, and, while they need lots of water, they already have most of what they need.
 
Top