CHIPS Act: Strategic Dilemmas in Manufacturing Incentives
Commerce has their work cut out for them!
The CHIPS and Science Act was signed into law last year, earmarking $52 billion of government investment into domestic semiconductor manufacturing. So what happens now?
Vishnu Kannan from the Carnegie Endowment and Jacob Feldgoise from Georgetown’s Center for Security and Emerging Technology guide us through the strategic objectives and complexities of the CHIPS Act. We discuss:
Tensions and trade-offs in allocating the $52 billion earmarked by the CHIPS Act;
The balancing act of increasing international competitiveness in the semiconductor industry while also creating domestic jobs in the US;
The potential need for — and political complexities following — additional funding in the future as the leading-edge technology evolves beyond current investments.
Reading Between the Lines of the CHIPS Act
Jordan Schneider: Say someone hands you $52 billion. Cool. But what if you have to spend that money to save an industry? Medium cool — but still pretty cool.
Still not entirely sure exactly how to save that industry? Well, that’s how Jacob Feldgoise and Vishnu Kannan’s paper comes in. Their recent report, “After the CHIPS Act: The Limits of Reshoring and Next Steps for US Semiconductor Policy,” explores the tensions and tradeoffs inherent in what’s going to happen with all the money that the Commerce Department is about to spend.
Vishnu Kannan: Once you get angry enough, I think you can write; if you read enough about a thing, things start to annoy you about that thing. And this came from that.
It was this weird puzzle: it’s the idea that economic dependence is vulnerability. And being the manufacturer of a manufacturer of a thing — that’s real power. That feels like the zeitgeist that has landed in Washington now. And that’s weird to me, because it feels as though the national-security logic of human activity is totally overtaken the economic logic. And somehow this symmetry in commercial transactions — that if I give you money and you give me a thing, we’ve both walked away level — that symmetry has broken down.
So the paper started by looking at that and asking, “If dependence is vulnerability, that’s a problem with no limiting principle.” The only answer is to reduce dependence and get to autarchy as national security — which seems concerning for many of the nuanced tradeoffs that we’ve talked about.
The CHIPS Act, then, is also kind of an experiment in this place of uncertainty: based on this puzzle that we haven’t really fleshed out and risks that we can’t really quantify (because we haven’t gathered the data or set targets), we’ve got to spend a whole bunch of money. That’s the origin of the paper.
I don’t know if that puzzle is going to be solved by anything other than the CHIPS Act playing out the way it does and people writing histories the way they do. But for folks thinking about this space, anyone who can clarify that question satisfactorily to me would be a hero.
Jordan Schneider: Your paper raises the interesting idea of the varied and potentially competing goals that will end up orienting the ultimate spending of this money. So let’s just walk through what all three of those goals are, and then we can go one by one into a little more detail about the nuances and intricacies of achieving those.
Vishnu Kannan: It’s funny: the policy is pitched as this really forward-looking game changer for the way the US thinks about technology — and that may be right, but it’s also fundamentally reactive, and we think it’s reactive to three things:
It’s reactive to the shortages that we saw during the covid pandemic;
It’s reactive to the kind of ubiquity of Chinese hardware, and how American national-security officials are concerned about what that might mean for the US and allies in other parts of the world;
And it’s reactive to the political concerns about what China’s rise as an economic power means for the United States and the relative decline of America’s manufacturing sector.
From those three concerns come three goals.
Protect the semiconductor supply chain against shocks originating abroad (and we saw those during the covid pandemic);
Boost America’s international economic competitiveness, and to create domestic jobs along with that (it’s kind of a two-part goal);
Reduce the risk that chips produced abroad are vulnerable to sabotage. In part this is by building more chips here in the US.
Jordan Schneider: What is there in the legislation that Congress put around for guardrails to have the executive branch see the point of spending this money their way?
Jacob Feldgoise: I’ll break this down into two parts. First, the manufacturing incentives — trying to reshore the semiconductor supply chain to the United States. There aren’t a ton of restrictions on the Commerce Department: there are some due-diligence requirements, but broadly speaking, they have a lot of flexibility in terms of how they can spend this money. It doesn’t all have to go into fabrication (big, expensive manufacturing plants like the one that TSMC is setting up in Arizona) — they can spend it on other segments of the supply chain, on materials, for example, or on packaging. There’s some amount of money set aside for mature nodes (older semiconductors), but most of it will likely go to support the leading edge.
Second, on the R&D side, there is some amount of money that is going to be spent which is required to be spent on advanced packaging. And the rest of it is up to the Commerce Department to decide how they want to allocate.
Vishnu Kannan: Within the manufacturing-incentive side, for example, there’s a requirement that, in the application process, companies make commitments to worker- and community-oriented training — say, by doing workforce training in partnership with local institutions. That’s an interesting case where the domestic employment piece ties in with the, “We need to protect this ecosystem from shocks abroad.”
Goal #1: Guard Against Supply-Chain Shocks
Jordan Schneider: Goal number one: reducing the supply-chain exposure to foreign shocks. If you were going to try to min-max that one, what would it entail, and what could putting all your eggs in that basket potentially cost you?
Vishnu Kannan: This goal was the response to covid goal, at least in our view. Actually, run it back — there’s a story behind this goal: rewind to 2020, 2021, when you have the covid shocks, when you were hearing the term “supply chain” and “chip shortage” basically everywhere. At that point, the chip manufacturers were using nearly all of their fab capacity and their manufacturing capacity; and the demand boom that we were seeing in chips made it extremely difficult for companies to get their hands on these things and manufacture other products.
In addition to that, concerns in the United States — both about what improvements in Chinese semiconductor technology might mean for long-term military planning and what the worsening US-China relationship would mean for a crisis if, for example, China cut off semiconductor exports to the US — all of that came together into goal number one: to say that we ought to have some proportion of this capacity — we’re not really sure how much, but some proportion of it in the United States, or at least somewhere where it can be accessed by American companies or where the end products can get into the hands of American consumers; in other words, access to all the things that you need to facilitate the smooth running of an economy.
To that problem, the CHIPS Act’s answer is, “We should build manufacturing capacity in the US. If not here, then where, really?” There were some conversations about friend-shoring, for example, and trying to think about the way you would optimize how you spend each dollar: does it really make sense to spend a whole bunch of money in the US if labor costs are higher or if land costs are higher? But at its core, I think the politics drove this piece, and so we ended up with the $52 billion that Jacob was describing at the top.
Vishnu Kannan: Supply-chain risk can include a pretty wide spectrum of things. You could try to achieve what one might call “absolute resilience.” And what would that be? Maybe everything is produced in the United States. That’s one extreme. The other extreme is where maybe nothing except the final product ends up in the US.
Both of those are insecure — for different reasons. The challenge with factoring supply-chain risk into this as a thing that we want to analyze is that people haven’t really figured out what the right tolerances are.
Jacob Feldgoise: There’s been a lot of discussion about this fabrication step, which is front-end manufacturing, the first manufacturing step. It’s what TSMC does in particular; it’s what people usually talk about Samsung doing.
But the issue is that this is just one of two big manufacturing steps. There’s also assembly, test and packaging, and backend manufacturing. If you’re worried about supply-chain resilience — you’re worried about maybe something happening in East Asia that knocks off a bunch of capacity — you also need to have resilience along the supply chain.
There’s also a heavy concentration in assembly, test and packaging in East Asia. And if you’re worried about some kind of natural disaster or geopolitical event or military conflict happening there, then you also need to think about how that situation might impact the other stages of the value chain, including assembly, test and packaging — also including the materials that go into fabrication: there are literally hundreds of chemicals, and also the silicon and the wafers. It’s a whole process that we need to think about, not just fabrication.
Jordan Schneider: I like this line you guys have in there: “The supply chain is only as secure as its least-secure link.” But it’s tricky because “least secure” can be defined in a lot of different ways. Yes, if something drops out, then you won’t be able to make chips anymore — but it may be the sort of thing that you can get up and running in a few weeks or months; or, it could be something that would take a two- to three-year build-out to fix.
So these are very tricky multivariate questions and the optimization functions are by no means obvious.
Vishnu Kannan: There’s also the question, to your point, “What are you optimizing for?” In 2020 and 2021, a lot of folks looked at the supply chain being distributed — there was a lot of talk about concentration in East Asia — and so people started optimizing, at least intuitively, for geography: “If stuff is distributed, we should bring it back, and that will make it more secure.”
Sure — but up to a point. Bringing it all back and reconcentrating it somewhere else isn’t necessarily secure, either. Concentrating it in such a way that even if parts of the supply chain are in different countries but are linked together by, for example, one of only a couple of major shipping providers (who themselves struggled during the pandemic to keep up with demand) — that’s also a type of vulnerability. So I think the multi-dimensional nature of supply-chain resilience and management really comes to bite folks here.
Goal #2: Boost International Competitiveness + Domestic Jobs
Jordan Schneider: This is a good way to talk about our next thing: concentrating it all in one place is also not the economical thing to do. This is a globally distributed industry for a reason — it’s cheaper to make these things in different places where there’s specialization. Even the Chinese government is potentially running into a limit of how much they’re going to be willing to spend to support their domestic semiconductor industry.
With that, let’s talk about goal number two: boosting American international economic competitiveness and creating domestic jobs. What’s going on here?
Vishnu Kannan: We started by thinking about competitiveness, in part because everyone was saying the word “competitiveness.” There’s this fear that the US has “become less competitive internationally,” in part because of China’s rise, in part because many countries are attuned to the risks of not having their own advanced technology champions, and in part because the US government itself hasn’t spent enough time or invested enough money in recent years in thinking about semiconductor R&D. That all connects pretty closely with a key priority of the Biden administration: thinking about the American middle class and its prosperity.
When you put those two things together, the administration has come out and said, “We don’t just want a foreign policy for the middle class” — though that is an important message — “we also want a manufacturing policy and a competitiveness policy that benefits the middle class.” The CHIPS Act is an answer to that.
It says we’re going to invest in semiconductor research and development through things like the Advanced Packaging Center, which is a component of the CHIPS Act — and that stuff is going to help companies continue to be competitive internationally. But we’re also going to build manufacturing capacity in the US, and we’re going to employ in the US — that stuff is supposed to help folks who have been hit by the combination of globalization and automation, essentially giving them jobs that the administration says are “good-paying union jobs,” all of the standard buzzwords.
But I think there’s a challenge here: those two things can be in tension with one another — because semiconductors are fundamentally a capital-intensive, high-skill industry. If you want to compete with China on semiconductors in order to prevent the Chinese government from, say, dominating advanced nodes that could be potentially useful in military or AI applications, that involves a very different group of people — and it directs money to a very different group of people — than if you want to employ in the United States and your goal is to make sure that you’re employing folks who might otherwise be out of a job because of a combination of automation and globalization. So that’s the challenge that we isolated.
Jacob Feldgoise: Not all of those jobs in a fab are going to be benefiting the same kinds of people who might have lost a manufacturing job over the past twenty years. Most of those jobs will be going to high-skilled workers, many of whom are going to need master’s degrees or PhDs. And only a small percentage — maybe something like eighteen percent of those jobs — will go to low-skilled technical workers.
You can go even further to say, “Well, how many of those jobs are actually going to be filled by US citizens or US residents?” Because at the moment, we don’t really have a sufficiently strong talent base to staff the fabs that the CHIPS Act will invest in — which is part of the reason why the Commerce Department is focused so much on workforce development in this space.
Vishnu Kannan: From a workforce perspective — or from an employment perspective — I think it makes a lot more sense to think about this as a medium- to long-term play in a high-skill sector. I don’t think that talking about this as a solution to America’s broader manufacturing employment problem — which we generally associate with lower-skilled industries or sectors — is the right way to think about it.
Then, when we think about what we could recommend that the government do on this front, admittedly we flail a little bit — but I think we flail because this is a really hard problem. We say that someone ought to be convening scholars that are thinking about this problem. I want to be candid about that: to my mind, that’s what a whole bunch of folks both here in DC and in universities around the country are trying to figure out.
Jordan Schneider: And I think the scary thing is that there aren’t many of those scholars out there. Erica Fuchs at Carnegie Mellon published a paper at The Brookings Institution saying, generously, “The intellectual underpinnings to inform technology strategy at the national level are limited,” and, “the extant research provides little guidance on how a nation should make strategic technical decisions across domains while looking simultaneously across multiple national objectives.”
Folks didn’t imagine that there would be a universe where of tens billions of dollars would be up for grabs. But still, it’s a real failing and oversight, which hopefully many future grad students and PhDs are falling into to try to orient their careers around.
Vishnu Kannan: I think there are a couple of initiatives to flag here — but they’re in early stages, which is part of the reason we don’t have a concrete recommendation on this front.
The Hewlett Foundation, the Omidyar Network, and other funders are thinking about at a pretty high level: “What is after neoliberalism, or what is an alternative to that model?” And whatever your feelings about the word “neoliberalism,” I think they’re funding scholars who are thinking about precisely these problems — for example, “How ought the manufacturing sector of the country connect to employment, and what’s the economic framework that should facilitate that type of activity?”
Goal #3: Reduce Vulnerability to Sabotage
Jordan Schneider: Okay — number three: the risk of sabotage. Kind of niche a one.
Jacob Feldgoise: This is very much a tertiary goal of the CHIPS Act, and one that was not spoken about as frequently as the other two — but it stems from, first of all, general concerns over the past ten years or so (or frankly, even longer) about vulnerabilities in our hardware.
For example, there were the Meltdown/Spectre vulnerabilities a few years ago — and those weren’t necessarily a result of sabotage, but they demonstrated how hardware vulnerabilities can be absolutely devastating for millions of people. And it also kind of brings in concerns about hardware that originates in China — questions about the provenance of that hardware: “Is it safe? Was it tested properly? Are there any backdoors or vulnerabilities?”
A lot of this is speculation: we have few public examples in the open source to demonstrate that this has happened. But it remains a concern, and anecdotally, it seems that there are non-public examples of this happening.
So the CHIPS Act essentially says — in particular, the implementation strategy that was put out by the Commerce Department a few months ago, basically says — that the Commerce Department is going to work with DoD and with the Director of National Intelligence to define requirements for what constitutes secure and assured microelectronics. This is something that the government has been working on for literally decades, so it isn’t new. But they’re trying to figure out how to incorporate that into the new money through the CHIPS Act, and make sure that that money is going to build fabs and ecosystems that are safe and secure.
A lot of this stems from the concept that, by reshoring production and by taking these additional measures, we can reduce the vulnerabilities of the chips that are provided to US companies and the chips that eventually make their way into the hands of US consumers.
Our basic thoughts here are that the CHIPS Act is on the right track, and that in changing the location of semiconductor manufacturing, you do address a subset of risks related to chip sabotage; and in restoring manufacturing you can build a trusted supply chain.
But you don’t necessarily need to reshore to accomplish that. There are allies around the world who we trust — friend-shoring may reduce the risks to a tolerable level. If we’re talking about shifting some amount of production or some segment of the value chain out of China to other countries, it doesn’t necessarily need to come back to the US.
Vishnu Kannan: It’s worth remembering — for folks who think about hardware security or have thought about that problem for a long time — this is something they don’t really deal with by moving production. That’s not to say it’s a solved problem, but they think about how changes in their manufacturing process can help make the thing more secure.
There are some proposals, mostly out of parts of academia (a really good lab at NYU has thought pretty hard about this): you can try to fabricate a chip in two different places, or at least break up certain stages into kind of multi-location manufacturing processes — which makes each stage when it finally comes together a bit more secure, because it’s harder to either alter the recipe or other components that are used to produce a finished chip. I think those types of proposals are interesting, and we recommend in the paper a broader approach to security standards for hardware that folks at NIST have already been thinking about for a while.
Jordan Schneider: You can even go one bigger level of extraction — think about your marginal dollar in securing hardware versus securing software. Because what you’re optimizing for is not secured hardware, but a secured system — whether that’s military or civilian critical infrastructure.
This is a paper that I have not read because it’s probably impossible, because doing secured manufacturing implies increasing costs. So the extent to which you are getting more security or resilience out of spending that on hiring an extra person (say, to test your network or localize your data or whatever have you) is a really important question — because the goal isn’t secure hardware: the goal is security broadly defined in a conflict situation. Just having great chips won’t do you any good if you get completely pwned and you can’t even press the start button.
Vishnu Kannan: And I think that’s also a reason a big reason that folks in the hardware-security community — at least at DoD — were thinking about counterfeits. When you treat the counterfeit problem as similar to the hardware-security problem, then it makes the type of cost analysis you’re doing a lot easier.
Three Strategic Dilemmas
Jordan Schneider: Okay — let’s recap with our three strategic dilemmas.
Vishnu Kannan: We’ve got three dilemmas; we think these are really difficult problems — potentially intractable — but hopefully one could make progress on them with some time and some political breakthroughs. I’ll list them out, and then we can chat about them.
The first is aligning domestic and foreign semiconductor policy. If you’ve got a goal that involves domestic employment, and you’re thinking about this in a zero-sum fashion as a policymaker, you say, “Well, every fab I am able to stand up in Arizona or Ohio is good for domestic employment.” But that comes into conflict with some of the friend-shoring ideas and alliance-coordination ideas, because countries around the world are vying for standing up fabs in their own jurisdictions.
Jacob Feldgoise: This issue around syncing up domestic and foreign semiconductor policy comes down to the problem that the US wants to invest in its own domestic semiconductor sector — but so do all of our allies. And the worry is that if the US passes its $50 billion CHIPS Act, the EU will do the same, Taiwan and South Korea and Japan all pass their own versions, so does India — so we might be in a race to the bottom. We might be kicking off a subsidy race with our own allies, which doesn’t really benefit anyone.
Jordan Schneider: Number two: ensuring opportunities for the domestic labor force. It’s nice creating American jobs — but they’re supposed to spend this money within five years. There are not enough electrical engineers and plant managers that have US citizenship for all those jobs to go to Americans.
There are clearly going to be investments, and it would be nice to create some jobs. But if you want to achieve the other goals, I think it’s going to be challenging to do this from an ultra “made in America” perspective.
Vishnu Kannan: The third strategic dilemma — maybe it’s not as much a dilemma as the other two, but it’s on the politics of this whole thing — the CHIPS Act was pitched as a kind of one-time investment that’s going to attract a whole bunch of private-sector money. And certainly we will have, if not solved, materially improved the security and reliability of America’s semiconductor supply chains and everything it relies on.
But the combination of these medium-term changes in the technology paradigm that we use to manufacture semiconductors — the kind of generation-on-generation improvements in the technology that mean you have to modernize equipment or sometimes build whole new fabs — all of that together suggests that we might actually need more money.
And the news over the first couple of weeks of 2023 hasn’t been particularly spiriting: the stock prices for these companies are taking pretty big hits; there seems to be a glut (even though the semiconductor glut is now turning into something of a meme); there is potentially some overcapacity in key nodes. That all raises the question, “If we are going to need more money down the line, will Congress be willing to actually make those appropriations?”
Jordan Schneider: It’s interesting thinking about what the scorecard is on which this will be graded. We talked about these three goals, but ultimately, when you’re spending at this order of magnitude, CHIPS won’t be something that flies under the radar. It will inevitably turn into some sort of political football.
Vishnu Kannan: I am pretty struck by how this involves giving a lot of money to a lot of very profitable companies. In the run-up to the CHIPS Act, we saw folks primarily on the left — Senator Bernie Sanders, for example, made a statement to this effect — that these are the most profitable companies in the world, so why do they need subsidies? I think that’s actually a hard question to answer if you don’t have the intellectual scaffolding for industrial policy that we’re in the process of building as we go. So I think that that challenge is going to come back, especially if you need another round of chip money.
Jacob Feldgoise: The goal here, as it has been articulated, is to build leading-edge capacity in the United States — and the leading edge is constantly moving. What is currently the leading edge today is not going to be even in a year or two, let alone in five or ten.
And sure, maybe the process improvements for semiconductor manufacturing are going to slow down as we approach the limits of Moore’s Law — you might get more of your performance improvements out of really smart design and new transistor architectures — but there’s still going to be new kinds of fabs that are able to manufacture the most leading-edge advanced chips.
The investments we’re making now will be great for a couple of years, but at some point the leading edge is going to move on — and if the goal is to maintain leading-edge capacity in the United States, then that almost certainly is going to necessitate new investments down the road. So this is going to be a political problem that we’re going to have to face at some point.
Jordan Schneider: You alluded to it at the very beginning — just how much this comes down to market forces and corporate actors versus something the government really has agency in it. To an extent it reminds me of macroeconomics and the president: if there’s a good jobs day, the president is like, “Awesome — jobs, great! Inflation, lower!” Yet month to month, or even year to year, there are cycles that are larger than what any president can do to turn the dials. How this turns out and gets perceived three to five years from now is definitely not all in the government’s hands.
A How-To Guide
Let’s come to some of your recommendations on exactly what you think the knowledge base and types of thinking that the Commerce Department should be building on as they’re thinking about how to spend all this money.
Jacob Feldgoise: I can walk us through our high-level recommendations for what we call “the foundations of US semiconductor policy” that we think should be created.
The first step is gathering data, which is essentially supply-chain mapping. Maybe it would be helpful to frame this as a series of questions: “What choke points exist along the semiconductor value chain right now?” Right now, one of those choke points is that essentially all leading-edge manufacturing is concentrated in Taiwan and South Korea. But we also want to know what other choke points might exist along the value chain — maybe, for example, in chemicals. We also want to know what kinds of choke points might develop over time. To do all of that, we need supply-chain mapping. That’s the first step.
The second step builds on top of that: crisis planning. The questions we want to answer with this piece of the foundation are: “What kinds of crises would trigger each of these choke points?” “Which of the choke points should we be most worried about?” And also, “How can we alleviate the most concerning choke points?” To use our example of leading-edge manufacturing concentrated in South Korea and Taiwan: providing TSMC and Samsung with subsidies to build new fabs at the leading edge in the United States is a pretty good way to alleviate that choke point.
Finally we get to the third piece of this foundation: setting targets. We’ve done our supply-chain mapping, we’ve gathered the data, we’ve conducted crisis planning to figure out which choke points we should be most worried about — and now we need to set some kind of target. That involves figuring out what we’re trying to preserve in the event of a crisis.
For example, if we’re worried about a military conflict over Taiwan, what would our objective be? Would we want to make sure that the US military has access to all the semiconductors it needs to continue building weapons and planes? Would we want to make sure that US consumers will be able to get access to all of the consumer electronics they need — maybe 50% of the consumer electronics they need? Or do we want to focus on US companies that could be selling to US consumers or abroad? There are many different angles here.
And in the context of the CHIPS Act specifically, we might want to think about, “What amount of domestic production along the value chain do we need to reduce the risk to a satisfactory degree?” And that’s a really broad statement, because you have to do all the other stuff first before you can get to setting targets.
Vishnu Kannan: And these three things — the data, the crisis planning or scenario planning, and the targets — are all mutually reinforcing. You can even take them out of just the supply-chain context and start to apply that framework to some of the other goals that we’ve talked about as well.
The CHIPS Act has a broad domestic-employment objective. It’s not clear to me that there’s a real target to pin that down — I don’t really know what success on that front would look like. But if you said that we wanted growth in a given region to hit x percent, or if we wanted to employ y number of people, or in the context Jacob was talking about, if we want domestic fabrication capacity in the United States to equal z percent of total fab capacity — those can then inform the type of data you gather, the type of crisis planning you do, and all of the questions that you’re asking along the way.
The administration is due to come out with a Continuity of the Economy (COTE) plan at some point, which ties pretty closely into this type of activity. I’m really curious to see what they come out with, because to me this is an infinite loop — figuring out which parts of the economy are important, how you crisis-plan for them, and what types of data you need to be tracking in near-real-time are really hard questions. But if they can answer them in the context of a Continuity of the Economy plan, you can then take some of those lessons and apply them to the semiconductor sector specifically.
Jacob Feldgoise: I think it’s an open question, “Who exactly should do each of these three things — the supply-chain mapping, the crisis planning, the setting targets?” It will need to be an interagency process.
There are a couple of proposals out there for where the supply-chain mapping in particular could take place. It could be a new bureau in the Commerce Department, for example. Maybe on a temporary basis it could be built into the CHIPS Program Office. Maybe it could be part of a larger critical-technology analytics program down the road. But that’s something we’re going to have to think about: “Who exactly is going to do this work, and how will they be resourced?”
Jordan Schneider: There have been a handful of papers that have come out in the past year. We talked about Erica Fuchs’. The Special Competitive Studies Project had another one about what bureaucratic structure would make most sense to do the analysis that we’ve been talking about.
Do you guys have any hot takes on your favorite organizational structure?
Jacob Feldgoise: Professor Fuchs lays out a proposal that is pretty interesting: it basically would be a network of scholars that the US can draw upon to research questions about critical-technology analytics. And it’s a pretty nimble infrastructure that she’s proposing: the office that would do this analytics work would not necessarily have a large staff, but rather would have a series of program managers who are working with a network of scholars.
A common theme between this proposal and the one that SCSP lays out in one of their papers is this two-step approach (which I want to give Vishnu credit for identifying): analyzing and then implementing. In the case of Professor Fuch’s proposal, that organization would primarily just analyze — leaving the actual implementation of a critical-technology strategy to other government agencies. The SCSP proposals are primarily doing both of those two things to varying degrees — both analyzing and implementing a critical-technology strategy.
I’ll briefly outline the proposals that SCSP lays out. The first one is reforming existing institutions in the White House. The second would be these two proposals that were presented in Congress last session for a Technology Competitiveness Council (which would sit in the White House) and an Office of Global Competition Analysis (which would work as an external nonpartisan analytics hub). SCSP also suggested a non-governmental entity that could do this work, and also a fully private-sector entity.
The conclusion that SCSP comes to — and that I generally agree with — is that you can’t just pick one of these; you probably need multiple organizations that are overlapping with each other to achieve the best outcome. Critical-technology analysis and implementation probably shouldn’t be assigned to just a single organization — we should have a whole ecosystem of organizations that is working on this.
Vishnu Kannan: I was a little underwhelmed by the four options — not because I don’t think the nitty-gritty of bureaucratic organization is interesting (it’s clearly deeply consequential), but ultimately any of the four proposals will be an improvement over what we’re doing.
Now, the more complicated question to me — and what the body of the SCSP paper kind of touches on but I think glazes over a bit — is, “How do you define a critical technology?” Does this “small-yard, high-fence” thing that the Biden administration talks about hold up? Is that possible to do, especially when the technologies that we’re thinking about are foundational to the economy of the twenty-first century? I’m not sure that they have satisfactory satisfactorily inserted that — and for me, without satisfactorily answering that, it’s hard to then create an institution and do all of that analysis.
And one more for fun: the prompt was “trying to map out the global semiconductor supply chain”
Book(s) recommendation - I don’t think there is a great newish equivalent of “Men, Machines and Modern Times” but if you smash together Manuel DeLanda’s “War in the Age of Intelligent Machines” and Timothy Morton’s “Hyperobjects” you almost get there.