News Intel and UMC team up on chip manufacturing — Intel will produce jointly developed new 12nm node in its US fabs

Status
Not open for further replies.
This is good for diversification of production. I wonder what the supply chain is like for US based manufacturing of advanced chips, not to mention these fabs are in AZ? Hope they can continue to rely on their water supply.
 
  • Like
Reactions: shady28
10+++++++2 nm
Well, that's how you make chips be still profitable while everybody else constantly raises their prices....

But also these will be for other things and not CPUs.
The new 12nm process, which will be used primarily for mobile communication infrastructure and networking applications, will feature industry-standard design tools (EDA) and process design kits (PDK), simplifying adoption for external customers.
 
Is arizona the best state for chip production? Don't you need loads of clean water? Maybe reclamation tech has improved a bunch. I really think upstate New York is one of the best places for chip production. Good rail, air, sea, and road access. Also tons of fresh water and other financial incentives.
 
Is arizona the best state for chip production? Don't you need loads of clean water? Maybe reclamation tech has improved a bunch. I really think upstate New York is one of the best places for chip production. Good rail, air, sea, and road access. Also tons of fresh water and other financial incentives.
I assume its because of the proximity of existing and other planned chip related sites. But I agree, there are other areas in the US that could be better like NY, Pacific Northwest, places in the Midwest.
 
  • Like
Reactions: gg83
Is arizona the best state for chip production? Don't you need loads of clean water? Maybe reclamation tech has improved a bunch. I really think upstate New York is one of the best places for chip production. Good rail, air, sea, and road access. Also tons of fresh water and other financial incentives.
It's a combination of regulatory benefits, space availability and already existing infrastructure. In the case of this partnership it's using existing fabs, and I'd assume the 2027 timeline is that they will likely be drawing down Intel 7 manufacture to make room for it.

In Oregon where their most advanced fabs are (fab R&D is largely there) there simply isn't enough land available for them to expand for adding large scale fabs. Ohio gave them extremely good terms and had a huge swath of land available which is why their next big US fab location is there.
 
It's a combination of regulatory benefits, space availability and already existing infrastructure. In the case of this partnership it's using existing fabs, and I'd assume the 2027 timeline is that they will likely be drawing down Intel 7 manufacture to make room for it.

In Oregon where their most advanced fabs are (fab R&D is largely there) there simply isn't enough land available for them to expand for adding large scale fabs. Ohio gave them extremely good terms and had a huge swath of land available which is why their next big US fab location is there.
Excellant point. It probably is a lot cheaper to build in AZ, no trees and less rocky earth to excavate.
 
Unexpected, but good. The major chip shortages we had were mostly not CPUs, but things like LCD drivers and automotive embedded chips. Repurposing these older nodes for those purposes makes a lot of sense.
 
  • Like
Reactions: parkerthon
These fabs are being set up in Arizona because they are already built up and full of expensive equipment. Instead of selling the equipment to China, Malaysia and others, they are reconfiguring it for a high-volume process that will likely stay in high volume production for at least 15 more years. There are many 28nm designs than can be ported to 12nm saving a lot of money for the customers.

Converting these old fabs to 2 nm and smaller is not cost effective. The high-NA EUV equipment will not fit in the buildings. A better approach is build new fabs in a location on the eastern side of the US where water and labor are plentiful, located within 1 day's driving distance to the band of automobile plants which strings from Georgia to Michigan. There are also at least 10 great engineering universities located within a day's drive from this location.

I would not be surprised to see another player add this common process, quite possibly Samsung and Global Foundries.
 
Is arizona the best state for chip production? Don't you need loads of clean water? Maybe reclamation tech has improved a bunch. I really think upstate New York is one of the best places for chip production. Good rail, air, sea, and road access. Also tons of fresh water and other financial incentives.
At this point the biggest hold up to building new fabs is finding enough highly trained workers. Water, electricity,transportation and regulatory environment are helpful as well. You generally see fab sites announced near existing ones (poach people) or near major universities and manufacturing hubs (train/retrain).
 
Intel chose Columbus OH, Micron choose New York, relatively close to Global Foundries. Both locations have a lot of engineers who would like to make good money, but don't want to move to or near the west coast because it is too far from family and friends.

The great news for those living in North America is that it is happening in the US and not Vietnam, Malaysia or India.
 
  • Like
Reactions: parkerthon
Is arizona the best state for chip production? Don't you need loads of clean water? Maybe reclamation tech has improved a bunch. I really think upstate New York is one of the best places for chip production. Good rail, air, sea, and road access. Also tons of fresh water and other financial incentives.
New York is one of the three worst places to do business in within the USA. Tells you a lot when a water hungry business will locate in a desert rather can go to NY.
 
I guess I'm not keeping up very well, but I expected new fabs for Intel to have evolved past 12 nm products. But I see here that this just answers the question: "What do we do with our old foundry equipment..."

Another article I read about this today kind of glossed over the fact that this isn't really cutting edge foundry stuff.
 
There is still a lot of money to be made by offering an improved version of an older node. The cost per transistor can be a much better for for many devices. A big problem affecting the industry since 22nm has been the cost per transistor rising instead of falling with new denser nodes, as had been the norm for decades. Unless your product really demands the cutting edge, there is a lot of incentive for making the most of lesser densities that are more cost effective for that instance.
 
I guess I'm not keeping up very well, but I expected new fabs for Intel to have evolved past 12 nm products. But I see here that this just answers the question: "What do we do with our old foundry equipment..."

Another article I read about this today kind of glossed over the fact that this isn't really cutting edge foundry stuff.

That's relative. Anything under 28nm is considered pretty cutting edge. The majority of chips made are on that, or higher, nodes.
There's a cost scaling issue that starts to show up around 10/7 nm, where going smaller gets more expensive due to exponentially more expensive wafers. That would be net sum zero if the density was also doubling each node, but it is not. It's more like 1.4X. So until or unless that changes, 12nm is going to be cutting edge for a lot of sectors.
 
Ny and California are the most heavily regulated states in the union for sure. Valid point!
Well Micron and Global Foundries are building more fabs in upstate NY so water and labor availability might outweigh regulations in some cases
 
Well Micron and Global Foundries are building more fabs in upstate NY so water and labor availability might outweigh regulations in some cases
Gf has been a little short on meeting their promises here in NY. When they abandoned 7nm and smaller it made a lot of people question what's going on. That's great to hear both are still planning on expanding here. IBM still has a huge presence here too. Arizona is closer to Taiwan and the rest of Asia. So maybe they save some shipping costs not going through the Panama canal, or something.
 
12nm is kind of outdated. Then you take into account that they are just building the plant and it will not come online for awhile and you got to wonder what is going on.
 
12nm is kind of outdated. Then you take into account that they are just building the plant and it will not come online for awhile and you got to wonder what is going on.
Seems like you may not have read the article carefully enough as this node is being implemented at existing fabs not a new one:
Intel Foundry Services (IFS) will produce the new node in its Fabs 12, 22, and 32, located at its Ocotillo Technology Fabrication site in Arizona, and the node will be made available to external customers.

...

These three fabs will slowly roll off work on the 10nm and 14nm nodes as Intel moves towards production of the new Intel/UMC 12nm node in 2027, thus making capacity available for the new node.
 
Seems like you may not have read the article carefully enough as this node is being implemented at existing fabs not a new one:
You are right. My mistake. But now it gets even scarier. Gearing up to produce 12nm at home means they are preparing for something else. Something bad.
 
You are right. My mistake. But now it gets even scarier. Gearing up to produce 12nm at home means they are preparing for something else. Something bad.
That's one way of looking at it, but I prefer the slightly more optimistic view: China stopped the Tower acquisition which would have given Intel this manufacturing capability in house. After that deal fell through Intel signed some deals with Tower so it makes sense they'll sign a deal with UMC to add new capability rather than try another acquisition.
 
Status
Not open for further replies.