German TSMC fab will produce 16nm there, not 4nm though. Useful for the auto industry but much lower margin and less strategically important than 4nm fab in the US.
Strategic for that same German auto industry, though. I assume that the Covid disruption to the supply of boring but essential microcontrollers for cars was a wake-up call.
Speaking of the leading edge, though: while industrial policy, like other kinds of investment, is easier with the benefit of hindsight, there must be some regret at having let Global Foundries drop out of the peloton.
That's still nice, especially considering that it’s somewhere between Haswell and Broadwell from 2014.
Maybe not the kind of progress or initiative that gets headlines, but neither is it trying to push as far as what Intel has been trying to do for the past few years.
Sure, but coming dead last behind Taiwan, Korea, US, Japan and China in the race to cutting edge semiconductor manufacturing is nothing to brag about. That's like celebrating for coming last.
This means you're getting the lowest industry margins, meaning less profits, less money for R&D, less wages and also less geopolitical leverage. This is nothing to celebrate but should be an alarm clock for our elected leader to wake the f up.
A lot of semi research is done in the EU, like at IMEC in Belgium, but few of it ends up commercialized by EU companies, so EU taxpayer money gets spent but other nations get to reap the rewards.
What Europe wants is not necessarily profitability but rather resilience. You can't leave this kind of decision up to the irrationality of market forces. So—you're correct, germany (or the EU) should subsidize chips if they want to weather the future.
chip fabs are big and contain a lot of things like pumps (and even a few very exotic lasers). but they're not power-intensive the way a steel plant is - or even a datacenter.
German TSMC fab will produce 16nm there, not 4nm though. Useful for the auto industry but much lower margin and less strategically important than 4nm fab in the US.
Strategic for that same German auto industry, though. I assume that the Covid disruption to the supply of boring but essential microcontrollers for cars was a wake-up call.
Speaking of the leading edge, though: while industrial policy, like other kinds of investment, is easier with the benefit of hindsight, there must be some regret at having let Global Foundries drop out of the peloton.
That's still nice, especially considering that it’s somewhere between Haswell and Broadwell from 2014.
Maybe not the kind of progress or initiative that gets headlines, but neither is it trying to push as far as what Intel has been trying to do for the past few years.
Sure, but coming dead last behind Taiwan, Korea, US, Japan and China in the race to cutting edge semiconductor manufacturing is nothing to brag about. That's like celebrating for coming last.
This means you're getting the lowest industry margins, meaning less profits, less money for R&D, less wages and also less geopolitical leverage. This is nothing to celebrate but should be an alarm clock for our elected leader to wake the f up.
A lot of semi research is done in the EU, like at IMEC in Belgium, but few of it ends up commercialized by EU companies, so EU taxpayer money gets spent but other nations get to reap the rewards.
13 replies →
What Europe wants is not necessarily profitability but rather resilience. You can't leave this kind of decision up to the irrationality of market forces. So—you're correct, germany (or the EU) should subsidize chips if they want to weather the future.
If you mean the Intel factory, this is delayed by 2 years. If it ever will come.. And the other planed Wolfspeed factory is cancelled completely.
Both will never come. For obvious reasons.
What are the obvious reasons?
18 replies →
How much does Germany's very expensive electricity affect TSMC's costs?
At that size of node, semiconductor manufacturing costs are not material constrained.
electricity is not "material" it is energy input.
2 replies →
chip fabs are big and contain a lot of things like pumps (and even a few very exotic lasers). but they're not power-intensive the way a steel plant is - or even a datacenter.
[dead]