3 ms·

These headlines don't mean anything whatsoever until Biden and Congress show up with a briefcase full of money. This isn't an expensive problem to solve. TSMC'

These headlines don't mean anything whatsoever until Biden and Congress show up with a briefcase full of money. This isn't an expensive problem to solve. TSMC's 3nm fab (slated to open next year) was estimated to cost just under $20B [0] back in 2018 when the project began. To put that number in perspective, Apple has spent $528B [1] on share buybacks and dividends in the past decade. I surmise that a Taiwan-centric military conflict would cost the US far more than either of those two figures. In light of the recent storms in Texas, a federally run power plant that serves only semiconductor fabs could be built in Austin TX as a first step. [0] https://www.taiwannews.com.tw/en/news/3600724 https://www.taiwannews.com.tw/en/news/3600724 [1] https://s2.q4cdn.com/470004039/files/doc_financials/2021/q1/Q121-Return-of-Capital-Timeline.pdf https://s2.q4cdn.com/470004039/files/doc_financials/2021/q1/...

rsj_hn 6y ago
Unfortunately money can't be converted into chip fabs so easily. Building a chip fab requires brains, a functioning corporate culture that rewards and promotes said brains, and large scale cooperation to work towards a common goal as specified by said brains. If Intel was able to build a chip at a price that would yield profitable stream of chips, then they would have no trouble funding said chip fab. There is no shortage of investment dollars here. The US has more than enough investment capacity to do this. But Intel can't execute, and when you can't execute, the amount of money thrown at a project is irrelevant. What's more, there are very few organizations that have both the human capital and are able to out-execute TSMC when it comes to building chip fabs, and it's almost certain than none of them are in the US, even though we have our own printing press that can generate as many dollars as are needed, yet it still wont be enough paper to deal with the brain-organization gap. Real things are built with real resources, such as organizational capacity, organizational talent, intelligence, solidarity, focus, tenacity, and long term thinking. You can run the printing presses all day long, and you're not going to get a world-competitive chip fab if you lack these. But if you have them, the funding issues will take care of themselves, as investors will beat down the door to lend you money, which is what happened when Noyce first announced he was starting his own company.
- totalZero 6y ago
  You can pontificate all you want about money alone being insufficient, but mine isn't a comment about printing money, nor about Intel. Long term semiconductor capital expenditure won't give fruit for several years, and reliance on foreign production binds us into disadvantageous geopolitical power projection games. Companies are cautious in laying out the funds because they want to be certain of demand and efficient with their resources. They don't particularly care about the externalities. A government has power and scale that makes that job easier, and an incentive to establish better long term solutions that may cost more than the short-term greedy solution. In contrast, corporations have shorter time horizons, less risk tolerance, and a greater incentive to seek the greedy solution. You get brains when you hire them. You get foundry tooling when you buy it. For the specific problem of building a manufacturing ecosystem around semiconductor fabrication, both are widely available. Executive orders about China's influence upon the semiconductor supply chain are just lip service unless and until eleven or more zeroes are included somewhere in the package. It's not a PR problem. As you correctly point out, someone actually has to roll up their sleeves and build lots of shit. That doesn't happen for free, and it takes coordination and risk that almost no corporation can pull off unilaterally. We have an incredibly sophisticated defense industrial apparatus for the same reason. (BTW...Not all spending results in money creation. You can tax, issue bonds, or just recover the cost from the productivity of the project. The idea of "printing money" isn't necessarily relevant to capital projects that return revenue, although I suppose it is always a backstop.)
  - rsj_hn 6y ago
    Intel has been spending about 14B/year on long term capex and another 14 B/year on R&D. It doesn't seem to have gotten them a whole lot. https://www.macrotrends.net/stocks/charts/INTC/intel/research-development-expenses https://www.macrotrends.net/stocks/charts/INTC/intel/researc... You may be able to buy an old chip fab, but you can't just buy organizational capacity and the ability effectively run an R&D department. US spending on R&D has been increasing steadily, the private sector funds 600B/year https://fas.org/sgp/crs/misc/R44307.pdf https://fas.org/sgp/crs/misc/R44307.pdf There has been no slowdown in either public or private R&D spending. The world is literally filled with the carcasses of failed R&D projects when nations that tried to develop industries but failed to do so due to deficits in organizational skill, excess corruption, lack of human capital, and ineffective business practices. Meanwhile tiny Taiwan went from a nation that produced shoddy toys and bicycles to a world power. You can't buy solidarity, unity, or human capital. As for our defense industrial complex, I would say that real innovation there, too, has substantially slowed down compared to the 50s-60s. Look at how quickly we obtain supersonic flight or nuclearized the navy, and how much slower progress has been since. Similarly innovation per real dollar spent has dramatically fallen in the medical field, too. Software is the main area where the US remains a leader, but outside of that it's slim pickings, even as we spend more money on R&D than ever before.

rsj_hn 6y ago

Unfortunately money can't be converted into chip fabs so easily. Building a chip fab requires brains, a functioning corporate culture that rewards and promotes said brains, and large scale cooperation to work towards a common goal as specified by said brains. If Intel was able to build a chip at a price that would yield profitable stream of chips, then they would have no trouble funding said chip fab. There is no shortage of investment dollars here. The US has more than enough investment capacity to do this. But Intel can't execute, and when you can't execute, the amount of money thrown at a project is irrelevant. What's more, there are very few organizations that have both the human capital and are able to out-execute TSMC when it comes to building chip fabs, and it's almost certain than none of them are in the US, even though we have our own printing press that can generate as many dollars as are needed, yet it still wont be enough paper to deal with the brain-organization gap. Real things are built with real resources, such as organizational capacity, organizational talent, intelligence, solidarity, focus, tenacity, and long term thinking. You can run the printing presses all day long, and you're not going to get a world-competitive chip fab if you lack these. But if you have them, the funding issues will take care of themselves, as investors will beat down the door to lend you money, which is what happened when Noyce first announced he was starting his own company.

totalZero 6y ago
You can pontificate all you want about money alone being insufficient, but mine isn't a comment about printing money, nor about Intel. Long term semiconductor capital expenditure won't give fruit for several years, and reliance on foreign production binds us into disadvantageous geopolitical power projection games. Companies are cautious in laying out the funds because they want to be certain of demand and efficient with their resources. They don't particularly care about the externalities. A government has power and scale that makes that job easier, and an incentive to establish better long term solutions that may cost more than the short-term greedy solution. In contrast, corporations have shorter time horizons, less risk tolerance, and a greater incentive to seek the greedy solution. You get brains when you hire them. You get foundry tooling when you buy it. For the specific problem of building a manufacturing ecosystem around semiconductor fabrication, both are widely available. Executive orders about China's influence upon the semiconductor supply chain are just lip service unless and until eleven or more zeroes are included somewhere in the package. It's not a PR problem. As you correctly point out, someone actually has to roll up their sleeves and build lots of shit. That doesn't happen for free, and it takes coordination and risk that almost no corporation can pull off unilaterally. We have an incredibly sophisticated defense industrial apparatus for the same reason. (BTW...Not all spending results in money creation. You can tax, issue bonds, or just recover the cost from the productivity of the project. The idea of "printing money" isn't necessarily relevant to capital projects that return revenue, although I suppose it is always a backstop.)
- rsj_hn 6y ago
  Intel has been spending about 14B/year on long term capex and another 14 B/year on R&D. It doesn't seem to have gotten them a whole lot. https://www.macrotrends.net/stocks/charts/INTC/intel/research-development-expenses https://www.macrotrends.net/stocks/charts/INTC/intel/researc... You may be able to buy an old chip fab, but you can't just buy organizational capacity and the ability effectively run an R&D department. US spending on R&D has been increasing steadily, the private sector funds 600B/year https://fas.org/sgp/crs/misc/R44307.pdf https://fas.org/sgp/crs/misc/R44307.pdf There has been no slowdown in either public or private R&D spending. The world is literally filled with the carcasses of failed R&D projects when nations that tried to develop industries but failed to do so due to deficits in organizational skill, excess corruption, lack of human capital, and ineffective business practices. Meanwhile tiny Taiwan went from a nation that produced shoddy toys and bicycles to a world power. You can't buy solidarity, unity, or human capital. As for our defense industrial complex, I would say that real innovation there, too, has substantially slowed down compared to the 50s-60s. Look at how quickly we obtain supersonic flight or nuclearized the navy, and how much slower progress has been since. Similarly innovation per real dollar spent has dramatically fallen in the medical field, too. Software is the main area where the US remains a leader, but outside of that it's slim pickings, even as we spend more money on R&D than ever before.