Does anyone have any insight on how they make it economically viable?
US salaries are astronomically high compared to the rest of the world. In the tech sector that's doubly so. Everything is incredibly expensive there. Is this basically a small facility to keep some politicians happy?
Or is it used to provide some supply military gear at 50x the price?
Will it get shut down in a few years once everyone forgets about it?
Salaries aren't a huge part of manufacturing like people think. Once the factory is built, accounting for logistics of shipping and remote management, it's not a huge difference financially.
Getting the factory built, however, takes significantly more time and costs more in both actual dollars but, more importantly, the opportunity cost lost to not producing products in that extra build time.
Some hardware any superpower should be able to produce on home soil almost no matter the cost. That's why even Russia keeps domestic fabs making chips at some ancient process node - not to sell them for profit, but to maintain an ability to produce chips for the military and their own economy should things go down as they say.
US Workers while costing more do seem are still competitive at expensive high value tasks. There was some reports a year back where the TSMC labs in Arizona while employee costs were higher also had 4% higher yield [1].
I'd wager combining that with US defense contracts for US made chips would be lucrative for NVidia.
What I find weird is that a year ago there were reports in the NYT and elsewhere that TSMC was unable to make the AZ plant work because of lazy / dumb Americans. And then 6 months later, poof, 180, the reports were that they were right on track???
In terms of Labour, the number one cost of a node is R&D. And that is happening in Taiwan not in US.
And if US could somehow provide lower electricity cost to Fabs, which is the number one cost in production, it will offset a lot of expensive items on the list. The actual labour cost for running the Fab is comparatively small.
Modern high-end fabs have extremely expensive equipment and are highly automated... like they are so automated that people don't actually handle wafers... it is almost all robotic.
Thus, salaries and cost of services do not factor in as heavily as you might think to fab economics.
Data suggests that TSMC's per wafer costs in Arizona are 10-30% higher than Taiwan and that Arizona fab is relatively new. It's economics will probably improve over time, narrowing the margin to 5-15%.
Looking towards the future, power costs and other global supply chain factors could very easily make TSMC's Arizona fab less expensive and more reliable to operate over time. For one, the US is completely energy independent... Taiwan is not.
The EU was funding fab lines, several levels back up the chain IIRC? The idea being that the chips doing car comms, engine management, cruder FPGA, old ARM cores, can be done fast, and stop supply chain weaknesses for things Europe needs chips in, like cars (and tanks, and UAVs and ...)
I'm not saying tiny lines aren't cool. I'm just saying the idea you can't be successful if you make cruder, older resolution chips is probably wrong: Your printer and your car don't care if the Dice is 10mm not 5mm, and the track lines are 5x wider. MILSPEC stuff probably runs cruder for other reasons. Resiliency? Verilog proofs?
I also have no idea how many dice you get off a single ingot these days. 300mm wide, but how long?
>The EU was funding fab lines, several levels back up the chain IIRC?
Yeah but much larger(16-12nm) and much less profitable nodes than what Taiwan, the US or even Japan and China have now.
> I'm just saying the idea you can't be successful if you make cruder, older resolution chips is probably wrong
Define success. Smallest nodes are bringing in the most profits and every country prefers more profits versus less profits, especially Europe given it's budget deficits and welfare spending.
Larger nodes that aren't very profitable are good for national security but Russia and even North Korea are proof you don't need much domestic semiconductor industry to completely terrorize neighboring countries and level entire cities. WW1-style artillery shells and rifle rounds will do just fine.
24 comments
[ 2.9 ms ] story [ 53.8 ms ] threadI never thought this would happen, or that if it did, we'd be a few generations behind.
Now let's onshore or friendshore everything else we need! Rare earths, mid-tier processors, chemical precursors, pharmaceuticals, steel, robotics/mechatronics, solar, drones, ...
Why even stop there? Kill the Jones Act, get back to building naval drones and ships of all kinds, ...
US salaries are astronomically high compared to the rest of the world. In the tech sector that's doubly so. Everything is incredibly expensive there. Is this basically a small facility to keep some politicians happy?
Or is it used to provide some supply military gear at 50x the price?
Will it get shut down in a few years once everyone forgets about it?
Getting the factory built, however, takes significantly more time and costs more in both actual dollars but, more importantly, the opportunity cost lost to not producing products in that extra build time.
I'd wager combining that with US defense contracts for US made chips would be lucrative for NVidia.
1: https://news.ycombinator.com/item?id=41952534
And if US could somehow provide lower electricity cost to Fabs, which is the number one cost in production, it will offset a lot of expensive items on the list. The actual labour cost for running the Fab is comparatively small.
Thus, salaries and cost of services do not factor in as heavily as you might think to fab economics.
Data suggests that TSMC's per wafer costs in Arizona are 10-30% higher than Taiwan and that Arizona fab is relatively new. It's economics will probably improve over time, narrowing the margin to 5-15%.
Looking towards the future, power costs and other global supply chain factors could very easily make TSMC's Arizona fab less expensive and more reliable to operate over time. For one, the US is completely energy independent... Taiwan is not.
I'm not saying tiny lines aren't cool. I'm just saying the idea you can't be successful if you make cruder, older resolution chips is probably wrong: Your printer and your car don't care if the Dice is 10mm not 5mm, and the track lines are 5x wider. MILSPEC stuff probably runs cruder for other reasons. Resiliency? Verilog proofs?
I also have no idea how many dice you get off a single ingot these days. 300mm wide, but how long?
Yeah but much larger(16-12nm) and much less profitable nodes than what Taiwan, the US or even Japan and China have now.
> I'm just saying the idea you can't be successful if you make cruder, older resolution chips is probably wrong
Define success. Smallest nodes are bringing in the most profits and every country prefers more profits versus less profits, especially Europe given it's budget deficits and welfare spending.
Larger nodes that aren't very profitable are good for national security but Russia and even North Korea are proof you don't need much domestic semiconductor industry to completely terrorize neighboring countries and level entire cities. WW1-style artillery shells and rifle rounds will do just fine.