Thank you for discussing these topics , Ive been trying to come up with a framework or mental model for how to think about the trade off between industrial resilience and peace time economics and it feels like this is the first thing that’s digging into this
«However, memory production operates very differently from customized ASIC chips or GPUs from a leading-edge foundry. Memory chips are a standardized good with an oligopoly (there are three DRAM producers) that are roughly substitutable. [...] That is not how it functions at a customized leading-edge foundry.»
Actually that is how things used to work for leading-edge chips too; since decades ago for example both Intel and AMD produce the same type of high-end CPU chips because Intel's customers (in particular the military) insisted on having a second fully independent source and forced Intel to license AMD their high-end CPU architecture.
But business schools teach vendors to differentiate and build monopolistic moats and teach buyers that what matters is only the lower price in the short term rather than having multiple sources; so while high-end CPU chips are still mostly software-compatible between AMD and Intel in particular Intel which had a dominant position tried very hard to reduce compatibility and become single-source and as part of that they introduced a new single-source high-end CPU design called "Itanium" (which fortunately flopped).
«Samsung, TSMC, and Intel are not saying, “Here’s a standard. Apple, come to us with a standard and we’ll all figure out how to make it, and then you can substitute among all three of us.” That’s not how it works.»
Actually that is exactly what they did for mid-range CPUs based on the "ARM" standard architecture. Then Apple took the "ARM" standard and worked hard to differentiate it.
«Practically, what does it mean to have dual supply for an AMD or Nvidia? The most sense is to have one chip fabricated at one foundry, another model at another foundry»
When Intel was dominant they built many foundries by *exactly* replicating a model foundry into several others. That provided more capacity (with which they hoped to destroy AMD as an independent source) but not much extra resilience.
«When I was at SK Hynix, I found it fascinating that the CEO didn’t want to exceed a certain market share for customers like Dell for average server DRAM. They wanted customers to have supply resilience through other suppliers, which is very counterintuitive. That’s how they wanted to do business – they preferred long-term contracts. This works because it’s a standardized good with standards set at JEDEC.»
It works for a very different reason: they are prepared to accept (at least initially) a lower ROI on capital in exchange for long-term survival. That is pretty much much the secret of the "Asian Miracle" so far.
«How had we effectively outsourced all of memory production, and was that a bad thing and to what degree?»
Many USA executives and investors have as their primary profit-taking model that of asset stripping: find money left on the table, take it. This incentivizes pursuing short-term cost cutting, in particular capital spending, over everything else ("I'll be gone, you'll be gone"); offshoring follows from that. This model has made them fantastic fortunes.
The taiwanese industries were based on investors accepting a lower ROI in exchange for longer-term sustainability. They achieved that, just as the USA executives achieved their goals by offshoring to Taiwan and Shenzhen.
Now it turns out that what is good for USA executives and investors is not necessarily good for USA national security. But to whom does USA national security matter?
Whether USA national security matters is the essential difference between the "globalist" and "MAGA" factions of the USA oligarchs:
* The wealth and power of the "globalist" oligarchs are portable and it does not matter much to them whether they live in splendid palaces in San Francisco or NYC or a global city like Singapore or or Geneve or Taipei or Tel Aviv or Rio or Shanghai or Dubai, and indeed they already have palaces in all or most of those places. For them USA citizenship is just a convenience. What matters to them is whether the local politicians can be easily and cheaply "sponsored" and 200 years of USA history show that the USA is excellent in that regard, but so are many other states
* For the "MAGA" faction much of their wealth and power is immobilized in the USA (car dealers, real estate owners, franchisees, healthcare investors, ...) and they cannot easily move to a global city somewhere else without becoming much less rich and powerful.
For people who have never worked in the US government, it may be hard to appreciate what is, to me, damn-near miraculous: going from a Vision for Success paper to leading-edge fab production in the US in the space of a few years. It's a stunning achievement, no matter what objections some may have with the CHIPS program. I place it on a par with building the nuclear Navy, if not even more impressive. Thanks to Dan and his incredible team of all-stars.
You're questions about "escalation management" and "digital age deterrence" in this new geopolitical and geoeconomics environment are extremely difficult to answer, yet might be the most pressing considerations of all over the next decade. If we spent a billion dollars and put the nation's best thinkers on it, and model & sim it to death, perhaps we come up with a 95% solution. Yet the remaining 5% of "unknown unknowns" could be so fundamentally consequential that they could change the course of history if either the US or China decide to embark on a path that ends in conflict.
Of course, all of this depends on having an industrial policy that is rational, logical, enduring, and resourced. Neo-royalism or kleptocracy might be policy, but not the kind that will ever lead to the desired outcomes for the country.
Thanks for this. It is refreshing to be reminded of the days when they were smart people who understood semi industry dynamics guiding US industrial policy. CHIPS accomplished a ton of heavy lifting in a very brief period of time. And I would def like to see the first draft of that report one day!
One thing that did jump out at me though was the trope that Xiaomi/BYD/the Chinese system is all about fast followers. It bleeds into the concept that China copycats while America internet. Understand this discussion is focused on capital-intensive hardware industries, but if you look at internet then its hard to argue that China has not genuinely innovated and become the leader (WeChat, AliPay &tc &tc). After all what is Reels if not a fast-follower of TikTok??
Thank you for discussing these topics , Ive been trying to come up with a framework or mental model for how to think about the trade off between industrial resilience and peace time economics and it feels like this is the first thing that’s digging into this
much more to come!
«However, memory production operates very differently from customized ASIC chips or GPUs from a leading-edge foundry. Memory chips are a standardized good with an oligopoly (there are three DRAM producers) that are roughly substitutable. [...] That is not how it functions at a customized leading-edge foundry.»
Actually that is how things used to work for leading-edge chips too; since decades ago for example both Intel and AMD produce the same type of high-end CPU chips because Intel's customers (in particular the military) insisted on having a second fully independent source and forced Intel to license AMD their high-end CPU architecture.
But business schools teach vendors to differentiate and build monopolistic moats and teach buyers that what matters is only the lower price in the short term rather than having multiple sources; so while high-end CPU chips are still mostly software-compatible between AMD and Intel in particular Intel which had a dominant position tried very hard to reduce compatibility and become single-source and as part of that they introduced a new single-source high-end CPU design called "Itanium" (which fortunately flopped).
«Samsung, TSMC, and Intel are not saying, “Here’s a standard. Apple, come to us with a standard and we’ll all figure out how to make it, and then you can substitute among all three of us.” That’s not how it works.»
Actually that is exactly what they did for mid-range CPUs based on the "ARM" standard architecture. Then Apple took the "ARM" standard and worked hard to differentiate it.
«Practically, what does it mean to have dual supply for an AMD or Nvidia? The most sense is to have one chip fabricated at one foundry, another model at another foundry»
When Intel was dominant they built many foundries by *exactly* replicating a model foundry into several others. That provided more capacity (with which they hoped to destroy AMD as an independent source) but not much extra resilience.
«When I was at SK Hynix, I found it fascinating that the CEO didn’t want to exceed a certain market share for customers like Dell for average server DRAM. They wanted customers to have supply resilience through other suppliers, which is very counterintuitive. That’s how they wanted to do business – they preferred long-term contracts. This works because it’s a standardized good with standards set at JEDEC.»
It works for a very different reason: they are prepared to accept (at least initially) a lower ROI on capital in exchange for long-term survival. That is pretty much much the secret of the "Asian Miracle" so far.
«How had we effectively outsourced all of memory production, and was that a bad thing and to what degree?»
Many USA executives and investors have as their primary profit-taking model that of asset stripping: find money left on the table, take it. This incentivizes pursuing short-term cost cutting, in particular capital spending, over everything else ("I'll be gone, you'll be gone"); offshoring follows from that. This model has made them fantastic fortunes.
The taiwanese industries were based on investors accepting a lower ROI in exchange for longer-term sustainability. They achieved that, just as the USA executives achieved their goals by offshoring to Taiwan and Shenzhen.
Now it turns out that what is good for USA executives and investors is not necessarily good for USA national security. But to whom does USA national security matter?
Whether USA national security matters is the essential difference between the "globalist" and "MAGA" factions of the USA oligarchs:
* The wealth and power of the "globalist" oligarchs are portable and it does not matter much to them whether they live in splendid palaces in San Francisco or NYC or a global city like Singapore or or Geneve or Taipei or Tel Aviv or Rio or Shanghai or Dubai, and indeed they already have palaces in all or most of those places. For them USA citizenship is just a convenience. What matters to them is whether the local politicians can be easily and cheaply "sponsored" and 200 years of USA history show that the USA is excellent in that regard, but so are many other states
* For the "MAGA" faction much of their wealth and power is immobilized in the USA (car dealers, real estate owners, franchisees, healthcare investors, ...) and they cannot easily move to a global city somewhere else without becoming much less rich and powerful.
For people who have never worked in the US government, it may be hard to appreciate what is, to me, damn-near miraculous: going from a Vision for Success paper to leading-edge fab production in the US in the space of a few years. It's a stunning achievement, no matter what objections some may have with the CHIPS program. I place it on a par with building the nuclear Navy, if not even more impressive. Thanks to Dan and his incredible team of all-stars.
You're questions about "escalation management" and "digital age deterrence" in this new geopolitical and geoeconomics environment are extremely difficult to answer, yet might be the most pressing considerations of all over the next decade. If we spent a billion dollars and put the nation's best thinkers on it, and model & sim it to death, perhaps we come up with a 95% solution. Yet the remaining 5% of "unknown unknowns" could be so fundamentally consequential that they could change the course of history if either the US or China decide to embark on a path that ends in conflict.
Of course, all of this depends on having an industrial policy that is rational, logical, enduring, and resourced. Neo-royalism or kleptocracy might be policy, but not the kind that will ever lead to the desired outcomes for the country.
Thanks for this. It is refreshing to be reminded of the days when they were smart people who understood semi industry dynamics guiding US industrial policy. CHIPS accomplished a ton of heavy lifting in a very brief period of time. And I would def like to see the first draft of that report one day!
One thing that did jump out at me though was the trope that Xiaomi/BYD/the Chinese system is all about fast followers. It bleeds into the concept that China copycats while America internet. Understand this discussion is focused on capital-intensive hardware industries, but if you look at internet then its hard to argue that China has not genuinely innovated and become the leader (WeChat, AliPay &tc &tc). After all what is Reels if not a fast-follower of TikTok??