The Chip Shortage: Brought to You By American Neoliberalism

Truth be told, I’m surprised it’s taken this long for America (and even some foreign) manufacturers to have trouble with their chip supplies. I had taken note of America’s dependence on chips made in China and Taiwan before I started blogging, in the mid 2000s. Even though even back then, it was over half, the justification was that those were low-end, commodity chips. Not to worry.

Once in a great while, I would look for data on the percentage of chips used in the US and would come up with nothing more current than shortly after 2010. Now admittedly you have definitional issues: for manufacturers? Or for all uses, meaning OEM and consumer end products? Nevertheless, it was clear that even more chip manufacture was moving abroad and and the Chinese were producing more sophisticated chips.

If you’ve missed it, here’s a short recap of the chip mess. US automakers have had to cut production due to limited supplies. Maybe their project to shorten product lives and restrict consumer repair options by using lots of electronics is proving to have some downsides. Short chip supplies are also hurting some other consumer products companies. From CNBC:

Automakers across the globe are expected to lose billions of dollars in earnings this year due to a shortage of semiconductor chips, a situation that’s expected to worsen as companies battle for supplies of the critical parts.

Consulting firm AlixPartners expects the shortage will cut $60.6 billion in revenue from the global automotive industry this year. That conservative estimate includes the entire supply chain — from dealers and automakers to large tier-1 suppliers and their smaller counterparts, according to Dan Hearsch, a managing director in the New York-based firm’s automotive and industrial practice….

Automakers are scrambling to get supplies of the chips, which have extremely long lead times due to their complexity. The shortage is far down the supply chain, causing a ripple effect through the entire network….

Although major semiconductor suppliers such as Taiwan-based Taiwan Semiconductor Manufacturing and United Microelectronics have announced investment plans to increase production capacities, IHS says such plans will do little to nothing to relieve the short-term shortage.

“Because the cause of these constraints is the result of increasing demand from OEMs and limited supply of semiconductors, it will not be resolved until both forces are aligned,” said Phil Amsrud, IHS Markit’s senior principal analyst for advanced driver-assistance systems, semiconductors and components….

A 26-week lead time is needed to build the chips before they are installed in a vehicle, according to Hau Thai-Tang, Ford’s chief product platform and operations officer.

The origin of the shortage dates to early last year when Covid caused rolling shutdowns of vehicle assembly plants. As the facilities closed, the wafer and chip suppliers diverted the parts to other sectors such as consumer electronics, which weren’t expected to be as hurt by stay-at-home orders.

“Those chip manufacturers as well as wafer manufacturers started redeploying their capacity to like consumer electronics, which was growing because of people working from home and virtual working patterns,” Thai-Tang said during an investor conference last year. “Fast forward, if you add 26 weeks to when they made those decisions, the drop-off or the trough in the supply started to hit automotive the latter half of last year, going into Q1.”

Amusingly, Toyota has no chip problems. Per the Wall Street Journal:

A sharp rebound in car sales has paradoxically threatened the auto sector’s recovery as car makers contend with severe shortages of automotive chips. Except, apparently, for Japan’s automotive colossus: Toyota says it’s doing just fine…

The strong rebound in car sales has brought an unexpected problem, however, for most car makers: a severe chip shortage. A quick recovery after the sharp declines in auto sales early last year means chip makers are now unable to meet sudden new demand from car manufacturers. Adding to the problem, chip manufacturers are already running at capacity to produce components for other gadgets like personal computers and data centers, all of which got a boost from stay-at-home demand. The problem could last for months as it takes time for chip makers to ramp up capacity and readjust their product mix.

Mind you, there’s no way to add capacity remotely quickly enough to solve this problem unless there are mothballed fabs that can be brought on line quickly, and I haven’t seen this idea mentioned as an optioned. It takes two years to build a new fab and ramp up production.

But the Biden Administration has nevertheless decided that handwaving would be a useful contribution. From Bloomberg:

The Biden administration is working to address the global semiconductor shortage that has caused production halts in U.S. industries including autos, White House Press Secretary Jen Psaki said.

The administration is identifying choke points in supply chains and discussing an immediate path forward with businesses and trading partners, Psaki told reporters at the White House on Thursday. In the longer term, policy makers are looking for a comprehensive strategy to avoid bottlenecks and other issues the semiconductor industry has been facing for years…

The order will compel a 100-day review led by the National Economic Council and National Security Council focused on semiconductor manufacturing and advanced packaging, critical minerals, medical supplies and high-capacity batteries, such as those used in electric vehicles, two people familiar with the draft said.

Additional supply-chain assessments are expected within a year, focused on critical products — materials, technology and infrastructure — and other materials tied to defense, public health, telecommunications, energy and transportation…

On Thursday, chief executive officers of chip companies including Intel Corp., Qualcomm Inc. and Advanced Micro Devices Inc. wrote the president, urging him to support domestic production and stop the country from losing its edge in innovation…

U.S. companies mostly outsource production to Taiwan Semiconductor Manufacturing Co. and South Korea’s Samsung Electronics Co. That’s becoming a national security issue as tensions rise between the U.S. and China, which is investing heavily to expand its own chip industry.

In case you don’t recognize the playbook, the Biden Administration is launching a bunch of studies. Studies are the way to pretend you are taking a problem seriously but don’t intend to do much. Because for the US to takes these supply vulnerabilities seriously, it would have to do engage in industrial policy, the formal type, as opposed to our current versions, which is by default, determined by the effectiveness of duels among lobbyists. So our favored sectors – real estate, health care, higher education, finance, oil & gas – are big on extracting rents yet do little to advance being able to defend ourselves in a pinch. And that is not so much about the military as to having access to key supplies as well as adequate domestic manufacturing capability.

Will the Biden assessments show how little in the way of the chips we use are made here? How about our drugs? From a 2018 post:

A recent book, China RX: Exposing the Risks of America’s Dependence on China for Medicine by Rosemary Gibson and Janardan Prasad Singh, appears not to have gotten the attention it warrants. I learned about it thanks to reader Patrick F, who recommended a C-SPAN panel on the book with author Gibson plus other experts, such as former Clinton Administration official Patrick Malloy. If you go to C-SPAN, you can read a transcript auto-generated from the closed captioning. You can also listen Gibson describe some of the key points from her book in the interview below.

The big message of Gibon’s and Singh’s book is that the US relies on China for the production of active ingredients in drugs and in many cases, of the medications themselves, to the degree that we would have a public health crisis if supplies were interrupted. As Gibson said on C-SPAN:

Many people that we spoke to, both former government officials and some in industry said that if China shut the door on exports, within months, pharmacy shelves in the United States to be empty, and hospitals would cease to function.

And don’t assume generics king India would step into the breach. India gets many of the active ingredients for its pharmaceuticals from China. Gibson forecasts that China will overtake India in generics manufacture within a decade.

As Gibson explains, the US no longer makes its own penicillin, in part because China dumped penicillin in 2004, driving the last US plant out of business.

The medications where the US relies on China include heparin, a blood thinner that among other things is used for IV drips. No heparin, no IV treatments. Due to the difficulty in tracing the source of drug company ingredients, the authors could make only case by case investigations, but they China production to be critical for treatments for Alzheimer’s HIV, depression, schizophrenia, cancer, epilepsy, and high blood pressure.

Dependency is not the only risk. US drug companies shifted production to China not just to save cost but to escape regulation. The FDA has only limited access to Chinese factories, with the Chinese having well over 700, yet the FDA able to inspect only 15 a year on average. As Gibson said on C-SPAN:

The FDA is trying to get inspection on site in China. The Chinese have severely restricted the number of inspections that they will allow and the whole program has become completely ineffective.

And the Chinese are often less than cooperative. Gibson describes even then how the agency has been directed to a Potemkin facility, as in the goods were made somewhere else…and the FDA was not able to figure out where. Similarly, reports presented by the health authorities to the FDA is understood to be as reliable as Chinese economic data.

Back to the current post. America buys its military uniforms and boots from China. How smart is that?

And when the US has tried to provide incentives for manufacturing for new high tech industries to stay in the US, our brilliant business elite has turned up its nose. Mariana Mazzucato, in her classic The Entrepreneurial State, describes how the US funded the development of LCD panels and every much wanted to have them manufactured here. Silicon Valley had no interest.

That’s not saying it’s would be impossible to pursue industrial policy in the US, but the odds are vanishingly small. You’d need a President with solid Congressional support to pass legislation that gave our national champions tax breaks and subsidies, just like Big Finance and the medical-industrial complex get now. And it would also need to be explicit that all those government gimmies imposed obligations, like curbs on executive pay, stock buybacks, and offshoring. But implementing a program like that takes a decade-plus time, while our governments can barely hold the course through an election cycle.

Print Friendly, PDF & Email

97 comments

  1. vlade

    Every time I hear someone needs a super-specialised customised chip, I suspect they were sold something they didn’t understand.

    On similar but different – I don’t understand how the US can consider China a strategic threat and still be happy to move production of many strategic items there. It’s like the USSR sending strategic resources to Germany all the way to the night of Jun 21.

    Reply
    1. rusti

      super-specialised customised chip

      I agree with you in a broader sense, but there are some complications with swapping out automotive components even for simpler components like discrete capacitors, inductors and resistors. For one thing, AEC-Q components are tested to much higher standards on temperature range and cycling, mechanical durability, lifetime and other things. But even using other AEC-Q components there can be traceability issues if there is another “equivalent” part number that meets the data sheet specifications but maybe a circuit design was pushing a component beyond spec and that won’t be captured on the data sheet.

      Typically if an ECU supplier is going to swap out even minor discrete ICs to a different manufacturer they need to re-do function and reliability testing to convince their customer that it isn’t going to lead to a performance delta. Especially for a safety-critical system.

      Reply
      1. MartyH

        All true. Designing to the state of the art of “The Possible” consigns you to the processes and Fabs that put you into the most competitive part of the suppliers’ market. Small volumes (auto) have trouble competing with other gizmos generating vastly more volume. Volume tends to increase yield and reduce overhead costs of tooling and line-changeovers.

        Just saying.

        Reply
        1. Mike Smitka

          Good details, I added others in a comment below. Doing a “running change” invites all sorts of problems! Anyone who visits a supplier development center will see a warehouse-like room filled with shake tables and boxes that cycle temperature up and down, and setups that turn the electronics on and off. Or all 3 at the same time. Anything that moves sits in a jig with an arm moving it nonstop, 24/7, potentially for weeks. Those who’ve been around a while know lots of horror tales, the big Toyota sticky accelerator pedal recall was partly a failure to design in clearance for Americans who cover up the factory floor mat with another floor mat to keep it clean. But there was also a substitution of a resin from a different source causing mechanical failures.

          Reply
    2. Louis Fyne

      actually that is what literally happened, USSR was shipping (and Germany was paying for) commodities right until Jun 22 as Stalin continually ignored advice re. the blatant build-up pre-6/22.

      Moral of the story: the Establishment can act more stupidly than one can imagine

      Reply
      1. David

        There’s a little more to it than that. Stalin was well aware of his country’s weakness, and wanted to avoid giving the Germans any pretext at all for creating a crisis. More generally, he knew that Germany had practically no raw materials and could not feed itself, and that overrunning much of Western Europe had made these problems worse. Germany was highly dependent on the Soviet Union to be able to continue the war – in that sense, the Soviet Union was China compared to Germany’s USA. He hoped to exploit this, and that as a consequence, perhaps Germany and Britain would fight each other to a standstill. What he never really believed was that Hitler would try the unbelievable gamble of trying to capture all those resources and that food by military means. But once the Germans failed to take the territory they needed by October 1941, defeat was mostly a question of time. (The weakness of the German war economy, not often realised, is well described in Adam Tooze’s book on the subject.)

        Reply
    3. TimH

      I work for a semi company. Auto companies cancelled their orders last year, and suddenly it’s not their fault that no-one can supply in an upside? Any custom parts will have a 16 week lead time from scratch, so the business works on forecast-forecast-forecast. Auto companies may not want to hold parts inventory, well, neither do the suppliers.

      Reply
      1. Mike Smitka

        Yes! Supply chains can’t suddenly spit out parts above their existing level of production, because companies don’t leave pallets of work-in-progress lying around. The goal is just-in-time up the value chain, both because it lowers costs directly, and because it improves quality (and hence lowers costs indirectly, fewer production interruptions – a supplier that stops an assembly line may gets charged over $1,000 per minute).

        It’s even more of a challenge for semiconductors, because the “fabs” had plenty of other customers wanting more chips than were in the pipeline. When car companies (or their suppliers, most of the chips would not have been directly ordered by the Fords and Toyotas of the world), the fabs devoted that capacity to different chips for electronic items for which sales boomed under the pandemic.

        Making chips has a physical lag of 3-6 months, because each layer takes time, and there are many layers – coat, etch, lay down material, wash off the coat, repeat. Add in new contractual obligations, and you get further delay in the ability to restart production aimed at automotive. There is absolutely zero substitutability – these are microcontrollers (and perhaps specialized sensors, but as an auto industry person I’ve seen no references to shortages of those). All are embedded in printed circuit boards specific to that chip and function, all with specific packaging needs to fit into the appropriate place in a vehicle.

        As one earlier comment noted, they have to meet automotive standards. The shaking a chip takes in a car going over a gravel road, day after day (not hypothetical, as I live in a rural area) would be equivalent to dropping your cell phone from 1 meter, nonstop, for weeks on end. Plus they face thermal expansion, in my area -20ºC to over 70ºC under the hood. Under such stresses your cell phone will die, sooner rather than later. So the electrical leads on an automotive chip soldered more robustly, and may be in a different ceramic package. And so on. Vehicles need to design for 15 years for safety-critical items; cell phone and computer producers don’t aim for that! [Someone made a snide comment about designed obsolescence, but the empirics are clear: today’s cars last longer, despite increased complexity.]

        Reply
    4. Kevin

      The Soviets were dragging their feet on delivering raw materials to Germany in exchange for technology and machinery as per the Hitler-Stalin pact. It caused so much consternation for the Germans that a German-Soviet conference was held in November 1940. The conference was inconclusive and certainly moved up the timetable for the invasion of the USSR.

      Reply
  2. Taurus

    People in the US – in my observations- do not worry about war. War is something that happens in other places. History is not part of the curriculum except for American history which is taught several times over at different school levels. Most people in the US would have no idea what June 21st you are referring to. The Second World War started at Pearl Harbor and was won on the beaches of Normandy. All the decision makers went through those schools. So the industrial policy of the United States can be summed up as: “The free market and Jesus will provide. Pass the burgers.”

    Reply
    1. Kurtismayfield

      The US has two non belligerents on either side of its borders, and two very large oceans separating it from the rest of the world. There is a reason why the US population has the luxury of being isolationists; they know no one is invading this country anytime soon.

      We choose when we go to war. That is why no one cares. There has not been a war we have been forced into in over a century. We chose to go into WWI. We set ourselves up to be forced to join WWII eventually.

      Reply
      1. Synoia

        The price of entering WW1 and WW2 was paid by the assumption of mantle of the the British Empire.

        If the British had actually known the cost, and armistice with Germany in WW 1, throwing the French under the bus would hare been sensible (That is ignoring the European Triple Alliance).

        I read that entering the triple alliance was controversial at the time.

        Reply
      2. John Wright

        I view the US elites as having European elite envy, as in the European elite get to contend with interesting border disputes and conflicts at home, while the USA is separated from these conflicts and issues.

        This may be personified by former European, current USA citizen and HRC advisor Madeleine Albright.

        From a 2001 account of an incident in 1992.

        From https://www.theguardian.com/world/2001/sep/30/usa.afghanistan

        ‘What’s the point of having this superb military you’re always talking about if we can’t use it?’ Madeleine Albright screamed at Colin Powell. Her stinging rebuke could not have been better designed to scrape a raw American nerve, challenging the nation’s machismo and role as leader of the free word. Powell reacted furiously. ‘I thought I would have an aneurysm,’ he recalled. ‘American GIs are not toy soldiers to be moved around on some global game board’

        Powell has been proven incorrect, American GIs ARE to be moved around on a global game board.

        Reply
        1. JBird4049

          Like Smedly Butler, the good American officers have viewed the military, especially its people, to be treated with respect, to only use them when necessary; the American elites have been using, abusing, and disposing the people in the military in the same way as the average citizen or foreigner for centuries; something to be consumed be it their possessions, beliefs, lives, hopes, dreams, whatever they have as a form of rapacious, predatory, even parasitical neoliberalism (libertarianism really) since before the War of Independence.

          Madame Albright merely is an example of such a person.

          Reply
      3. scott s.

        True, if you consider Alaska not to be part of the US, then there’s also Guam, CNMI, and American Samoa, but I suppose you can discount those as not legitimately part of the US.

        Reply
        1. Kurtismaufield

          All good points. How many of those are actual targets for foreign aggression? I don’t see China fighting the Battle of Ankorage for the oil reserves anytime soon.

          Reply
  3. rusti

    Maybe their project to shorten product lives and restrict consumer repair options by using lots of electronics is proving to have some downsides.

    This ascribes a level of competence and cunning that I find farcical after more than a decade in the automotive and semiconductor industries.

    The number of electronic control units (ECUs) has grown rapidly with the introduction of new features that are increasingly expected to be standard by customers (NC commenters are not at all representative of new car buyers). Infotainment system? New ECU. Lane Keep assist? New ECU. Wireless phone charger? New ECU. They can’t operate as totally isolated standalone units because there’s a lot of information exchange among them for things like setting power states. Of course the engine, transmission, airbag system, climate control, chassis, etc have dedicated units.

    This trend is bad for the OEMs and primarily driven by Tier 1 suppliers (Bosch, Continental, Delphi, etc) who sell slight variations of the same ECU to all OEMs. For at least 5 years now everyone has talked about centralizing processing power and having dumber end nodes which might entail a different set of winners in Tier2/3 suppliers like Infineon or Nvidia. I don’t think anyone is really steering the ship, cars are so unbelievably complicated that no single person has a good understanding of all the subsystems. This picture is dense but informative.

    Reply
      1. cnchal

        My cars, decades old have that. What’s new is 99% of the digital crapola stuck in new cars is not needed for the purpose of driving and you can’t opt out.

        Now we have a mad dash of production, so moar failures embedded right there, never mind that having the electronic guts of the car be made of Chinesium is what clownsumers are getting for their 50 large. Wonder what the scrap rate will be? To boot, the recent and future cars are not going to worth saving either.

        Reply
        1. vegeholic

          And don’t forget that all of the “digital crapola” gets transmuted into Hedonic Quality Adjustments (see Wolf Richter, 10 Feb.) so we have to pay for all of it, but are not allowed to complain about the price increases. Also this is just another indication, in case you need it, that perpetual growth and finite resources eventually collide in ways that are not allowed to be considered by the neoliberal consensus.

          Reply
    1. vlade

      If the picture is correct, it means it’s a disaster waiting to happen, as any software of that complexity is not even remotely manageable. Unless of course they count everything and the dog and its writen in assembly.

      I shudder to think when the VW or so then try to sell “always online” cars. IoT vendors, and I’d put car manufacturers there, have exactly zero track record on making things secure, and I’ve seen remote breaks that could for example play with the brakes.

      Reply
      1. Synoia

        I pine for my 1962 Short Wheel Base Land Rover. The only “Electronic” component was the Motorola Alternator.

        We did 4,000 miles around Southern Africa in the Month of My 1973. Had to replace the Prince of Darkness’ Ignition Coil and replace a Core Plug.

        It was slow, noisy and uncomfortable, but easy to fix.

        Today? I cannot name much of what I can see under the Bonnet.

        Reply
      2. Louis Fyne

        It is a feature , not a bug.

        The vast, vast majority of new car buyers lease or change cars at least every five years.

        for this group (which is the group car makers really care about), more electronics is a net positive…more fuel efficiency, better diagnostics, more user assist safety aids. And ultra long-term reliability is not a concern.

        For the second, third, fourth-hand buyers, more electronics is a net negative as the supplier/OEM have the upper-hand in pricing spare parts.

        arguably the environment will be a net loser too as otherwise driveable cars will be junked because the owner can’t afford the repairs should the electronics get damaged or break down.

        Reply
      3. lordkoos

        “…which have extremely long lead times due to their complexity.”

        I’m a bit surprised at this as I assumed that the kind of IC chips that are used in cars are much simpler than say what is used in your average desktop or laptop.

        Reply
      4. MartyH

        Not waiting to happen. It sounds like it happened. The earlier comment about canceling orders is not a casual, “Oh by the way, we’re back and we want priority”.

        Reply
    2. Jeremy Grimm

      I was surprised to see how large the Debian software was — but I am not sure what that number aggregates, or indeed many of the other numbers in the diagram.

      Reply
      1. Alena Shahadat

        Me too, maybe if they count all the packages, it is huge. Not the OS. Anybody knows more? I knew that Firefox was pretty bad.

        Reply
  4. Thomas P

    Does it even have to be a war? What would happen if China one day without warning stopped all deliveries to the West. Would it be possible to rebuild or rerout production fast enough to avoid a disaster?

    Reply
    1. Louis Fyne

      No. Building even one chip foundry takes years.

      Just ramping up US N95 mask production took all of 2020.

      Even little things like wire harnesses or capacitors are nearly all imports, mostly made in China.

      But hey, the US exports its pollution and CO2, so it’s a win, right!?

      Reply
      1. Massinissa

        I know you’re joking, but China creates more pollution and CO2.

        Of course, that’s because they ACTUALLY MAKE THINGS and we no longer do. If industrial pollution doesn’t take place in once country, it takes place in another, if the global industrial output stays relatively the same. To be fair, China has taken steps to lower their emission rates, probably more than the US has recently, though they still have more smog in many areas than we do.

        Reply
      2. Mike Smitka

        Wire harnesses are imported, but the biggest suppliers (for automotive) are Mexico and the Philippines, not China. Note that with the explosion of circuits in a modern car, wire harnesses have exploded in weight, cost and complexity. Generally a seat set is the most expensive single component a car company purchases, but that’s now followed by wiring. Total weight is up there too. It’s the #1 or #2 quality problem and managerial headache in an assembly operation, lots of connectors to plug in, and in places that can’t be seen so the connectors have to be designed to provide tactile feedback when they’re properly plugged together. Miss a clip, and it sags, making it harder to connect – and potentially leading to an expensive warranty claim when the insulation abrades because sans clip the harness rubs against something.

        Digression: The other headache is the paint shop, which is often the bottleneck that limits how much a plant can turn out. Companies try to control temperature and humidity and airflow in the paint booth, but a big storm can throw things out of kilter, and suddenly more paint than usual is ending up somewhere and causing a drip, or failing to get applied with sufficient thickness (say, 17 microns instead of 22 microns).

        Reply
    2. Alena Shahadat

      Already, in Europe, we had no masks even for hospitals during the first four months of the Covid shutdown. In Geneva it was a private wealthy guy who paid for a cargo airplane to go to China and bring back chirurgical masks. Strange.

      Reply
  5. Eustache de Saint Pierre

    Rocks, hard places & short & curlys come to mind in relation to the West & it appears to me that there is a potential noose at the end of that there supply chain.

    Perhaps PPE was the singing canary in the coalmine.

    Reply
  6. Fazal Majid

    The US has tried industrial policy in semiconductors before, in the form of the Sematech consortium, and it was a complete failure.

    A cutting-edge “fab” that makes chips with transistors 5 nanometers in size like the ones used in the iPhone 12 and M1 Macs costs $17B. Only two companies have the ability to make these, TSMC and Samsung. Intel used to have the top fab technology, underpinned by its highly lucrative processor quasi-monopoly, but they missed the boat on mobile, are 4 years late on their 10nm fabs (due to differences in how transistors are measured, Intel’s 10nm is equivalent to TSMC and Samsung’s 7nm). Because of the capital-intensive nature of the fab business, it takes huge volumes to amortize the facilities and the volume is now in smartphones, not PCs or datacenters, and Intel completely missed the boat on mobile, in no small part because their CEO at the time, Paul Otellini, was a bean counter instead of an engineer, unusually for Intel. Other US fabs like IBM’s or Global Foundries (ex AMD) dropped out of the race earlier because they didn’t have volume either.

    The only company that has the wherewithal to fund the massive investments needed to rebuild fab capacity in the US is Apple. They use their supply-chain management to build competitive advantage, by financing factories for key vendors in exchange for exclusivity, and never share their tech with others. For instance, they bought up all of TSMC’s 5nm fab capacity, and others have to go to Samsung (that Apple detests because they compete in smartphones) or use older 7nm tech like AMD does (and is using that advantage over Intel to grab market share, thus further weakening Intel’s ability to finance a rebound in fabs).

    China’s fabs are generations behind, but Trump’s trade war was a rude awakening out of their complacency, and they are now prioritizing removing dependency on the US for any key technology. It’s always easier to catch up than to invent new tech, so it may take 10 years, but they will get there, possibly with industrial espionage involved against targets like ASML of the Netherlands which provides the lithography machines under embargo.

    Samsung is building a fab in Austin, provided Texas caves and gives them the tax cuts they are demanding, but TSMC has only announced desultory plans to build a minor and not leading-edge fab in the US, despite pressure from the Trump administration. Clearly TSMC management is deliberately putting all its eggs in its Taiwanese home, out of patriotism, something completely quaint and unfathomable to US managerial elites. The aim is to force the US to defend Taiwan against Chinese aggression, as it has become the single most important strategic point on the planet, far, far more important than Saudi Arabia. The problem is, Taiwan is also a highly active seismic zone and all it would take to severely disrupt the world economy would be a big earthquake there.

    I think the important thing here is that making things as opposed to vaporous intellectual property is out of fashion in the West, or at least in the finance-centric Anglo-Saxon part of it (Japan, Germany, Korea, France and Italy still like to make stuff). The Chinese are very aware of the trap of deindustrialization, specially when fashionable web companies exert a kind of Dutch disease diverting capital and more importantly talent from productive industry. I highly advise reading this view from the ground by an analyst in China, which challenges conventional wisdom:

    Just to whet your appetite, my favorite quote from the letter:

    With its emphasis on manufacturing, it cannot be like the UK, which is so successful in the sounding-clever industries—television, journalism, finance, and universities—while seeing a falling share of R&D intensity and a global loss of standing among its largest firms.

    Reply
    1. The Rev Kev

      Yeah, a major earthquake could blow up a serious part of the world’s economy if it occurred in the wrong place. I was just reading up earlier on the 1976 Tangshan earthquake which was massive though it as in China itself. One could hit Taiwan and it could be followed up by a tsunami to make it even worse and there goes a major part of the world’s chip production-

      https://en.wikipedia.org/wiki/1976_Tangshan_earthquake

      Reply
      1. John Wright

        There is also the hyped USA fear of North Korean ballistic missiles being able to strike Hawaii.

        But if NK wanted to inflame China/USA and the industrial world, it could missile target semiconductor producing Taiwan (1948 km away) rather than tourist destination Hawaii (7501 km away).

        Reply
    2. CoryP

      Thanks for the Dan Wang link! I have no idea who he is, but I learned a few things from that letter and I’ll enjoy poking around the rest of his site.

      Reply
    3. hunkerdown

      Those closer to the electronics engineering profession please correct me if I’m wrong, but from the tens of thousands of pages of technical data I’ve read, I get the impression that there are no such things as “automotive chips”, only chips that are graded and tested according to automotive specifications and process documentation standards.

      The Chinese line, according to a China watcher with an alleged inside source, supposes that the Taiwanese government, privately considering an independence referendum later in 2021, is allocating existing leading-edge manufacturing capacity based on Western geopolitical concerns and suppressing TSMC’s capacity expansion through the usual bureaucratic pretexts, under the conceit of an artificial shortage, with one possible aim of forcing China to sit out the 5G rollout and therefore becoming dependent on Western IP. China’s big fab SMIC, currently under US sanctions, is nonetheless expanding their non-leading-edge capacity later in the year, which suits the tens of thousands of non-leading-edge chip types that the industry (and the leading-edge chips themselves) need to keep the world running.

      Aside, a few months ago, the Ford employee parking lot down the road had been filled with inventory of 200-300 new pickups starting in November or so, which cleared out over the course of a few weeks in late December and early January. The streets are still not nearly as busy as they were pre-COVID. I don’t know who’s buying cars to not drive them, but I’m out of that market.

      Reply
      1. MartyH

        @hunkerdown: yes and no. “Automative Grade” chips have stringent requirements that place requirements on the supplier. That increases cost, of course. It also increases the spin-up time for testing of chips in the order. Automotive chips are unique designs. It is likely that different automotiive OEMs have bespoke designs. The chips for their orders can not be aggregated into the orders of others unless they are insustry standard designs. So the designed chips have setup and breakdown costs and testing costs and they’re probably “modest” orders compared to the more standardized chips (Memory, modems, cpus).

        It would be useful to review the history of the funding and founding of TSMC My memory is weak on the details but this well may have been a problem created by those who now feel the sting of their short-sighted actions of the past. There was a lot of pressure on US semiconductor manufacturers to move semiconductor fabrication off-shor. It became difficult to impossible to overcome financing, environmental, and political pressures and build significant new capacity. IMHO, the fabrication of semiconductors was largely ceded to Asia from the 1970s onward. To a multinational business community, the impact on national producers and markets are externalities.

        Reply
      2. Janie

        Someone I know just bought a new 50,000 vehicle, all the bells and whistles, with part of an inheritance from a parent who died of covid. Don’t know how common that might be, but fifty -somethings are inheriting houses sooner than they expected.

        Reply
    4. Krombopulous Michael

      Wow, I got more from your comment than I did from the article. I dipped my toe in the semiconductor industry briefly working at Lam Research in Fremont. They make the machines that make chips. It is very expensive tech, and the Bay Area used to have a lot of sq footage dedicated to fabs. Since Clinton, more and more of this was being outsourced. Though I can say that groundwater quality improved after those big fabs stopped polluting the local area. Of course, there are ways to handle that chemistry, but it was doubtless more rewarding to dump the chemicals and enjoy more profit. Until the jobs left. Whoopsy. I’m not sure how much (if any) fab space is available here, but it would be logical to put it back here if possible.
      As far as everything needing a chip (IoT), it is the same kind of “advanced gimmickry” that has plagued the car industry for some time. Just for laughs, go see how much a car in Mexico costs. They know their customers won’t care about a lot of that crap and pay a lot less for more or less the same workability. Having recently shopped for a new car, it was eye-watering how high the prices were. I make good money but I was thinking I didn’t want to be on the hook for 700-800/mo for 5-7 years.
      It’s nice that someone noticed a “strategic issue”, dare I say a “chip gap”, with this but it’s been going on for years. Will the new administration do anything about it? My bet is no, but a fat commission for a series of “studies” will doubtless be done.

      Reply
      1. JBird4049

        The lower cost of Mexican car comes from the lack of safety features. Those features can be expensive and aren’t needed until you hit that tree. One can easily save thousands of dollars, but personally I really don’t want to die from an accident.

        Reply
  7. Bill Smith

    “Back to the current post. America buys its military uniforms and boots from China. How smart is that?”

    When I saw the above I remembered this quote which is attributed to Marx, Lenin and others:

    ‘capitalists willing to sell the rope used the hang them’

    The answer is it depends. If things work out in the US’s favor then the Chinese are the bigger capitalists than the US? If things work out in the Chinese favor then it was stupid?

    But is that statement true? I was amused by the statement given it would seem to fit, but a quick google search seems to indicate that at least the uniforms are made in Alabama? By a company called ‘American Apparel’. Of course, through cutouts, they could be a wholly owned subsidiary of the PLA…?

    Reply
    1. Thomas P

      I think a good deal of US military equipment are made in USA by prisoners. Only way to beat Chinese wages is to use forced labor.

      Reply
      1. scott s.

        All I ever saw from Federal Prison Industries was furniture. Sourced from GSA, not in DOD supply channels. But DOD had its own problems. We used to use Sacramento Army Depot and Naval Weapons Station Seal Beach, both in CA, as repair depots for circuit cards. Thanks to the “peace dividend” and base closure, we got rid of those facilities.

        Reply
      2. Jeremy Grimm

        Strange you would think the US must beat Chinese wages for DoD to purchase US-made [really, truly US made — not “US-made”] equipment. Go easy on the KoolAide.

        Reply
      3. Yves Smith Post author

        Huh? No:

        https://www.reuters.com/article/us-usa-military-china/u-s-military-comes-to-grips-with-over-reliance-on-chinese-imports-idUSKCN1MC275

        https://nationalinterest.org/blog/the-buzz/the-us-militarys-greatest-weakness-china-builds-huge-chunk-25966

        All I could find was a 2913 article saying US prisoners were to make $100 million in uniforms, which is tantamount to zero. As a frame of reference, the military spent $3.2 million just on camo pattern redesign.

        Generally speaking, US prison labor is for low skill tasks or dangerous ones like firefighting.

        Reply
  8. Sound of the Suburbs

    How did we get here?
    It seems like a new generation has picked up neoclassical economics and built up an ideology from it that they totally believe in.
    They haven’t worked out that neoclassical economics is a pseudo economics that came into being to hide the discoveries of the classical economists.

    The Classical Economists had a quick look around and noticed the aristocracy were maintained in luxury and leisure by the hard work of everyone else.
    They haven’t done anything economically productive for centuries, they couldn’t miss it.
    The Classical economist, Adam Smith:
    “The labour and time of the poor is in civilised countries sacrificed to the maintaining of the rich in ease and luxury. The Landlord is maintained in idleness and luxury by the labour of his tenants. The moneyed man is supported by his extractions from the industrious merchant and the needy who are obliged to support him in ease by a return for the use of his money. But every savage has the full fruits of his own labours; there are no landlords, no usurers and no tax gatherers.”
    There was no benefits system in those days, and if those at the bottom didn’t work they died.
    They had to earn money to live.
    The classical economists could never imagine those at the bottom rising out of a bare subsistence existence as that was the way it had always been.

    Economics was always far too dangerous to be allowed to reveal the truth about the economy.
    How can we protect those powerful vested interests at the top of society?
    The early neoclassical economists hid the problems of rentier activity in the economy by removing the difference between “earned” and “unearned” income and they conflated “land” with “capital”.
    They took the focus off the cost of living that had been so important to the Classical Economists as this is where rentier activity in the economy shows up.
    The landowners, landlords and usurers were now just productive members of society again.

    Now everyone trips up over the cost of living, even the Chinese.
    It is well hidden.

    What’s the problem?
    Someone from the CBI (Confederation of British Industry) has just seen the equation.
    Disposable income = wages – (taxes + the cost of living)
    Two seconds later …..
    They realise the UK’s high housing costs push up wages, and are actually paid by the UK’s employers reducing profit.
    UK’s high housing costs make UK labour very expensive compared to elsewhere in the world, and it makes it very expensive to do anything in the UK.
    Employees get their money from wages.
    Employers pay the UK’s high housing costs in wages reducing profit.

    You can pay wages elsewhere that people couldn’t live on in the West.
    To maximise profit you will need to off-shore.

    Why was China always going to be the winner in an open, globalised world?
    Maximising profit is all about reducing costs.
    Western companies couldn’t wait to off-shore to low cost China, where they could make higher profits.
    China had coal fired power stations to provide cheap energy.
    China had lax regulations reducing environmental and health and safety costs.
    China had a low cost of living so employers could pay low wages.
    China had low taxes and a minimal welfare state.
    China had all the advantages in an open globalised world.
    It did have, but now China has become more expensive and developed Eastern economies are off-shoring to places like Vietnam, Bangladesh and the Philippines.

    Reply
    1. Sound of the Suburbs

      China trips up over the cost of living.

      Davos 2019 – The Chinese have now realised high housing costs eat into consumer spending and they wanted to increase internal consumption.
      https://www.youtube.com/watch?v=MNBcIFu-_V0
      They let real estate rip and have now realised why that wasn’t a good idea.

      The equation makes it so easy.
      Disposable income = wages – (taxes + the cost of living)
      The cost of living term goes up with increased housing costs.
      The disposable income term goes down.
      They didn’t have the equation, they used neoclassical economics.
      The Chinese had to learn the hard way and it took years.

      They have let the cost of living rise and they want to increase internal consumption.
      Disposable income = wages – (taxes + the cost of living)
      It’s a double whammy on wages.
      China isn’t as competitive as it used to be.
      China has become more expensive and developed Eastern economies are off-shoring to places like Vietnam, Bangladesh and the Philippines.

      Reply
      1. John

        Does no one understand the danger of putting all their eggs in one basket? But when the essence of doing business is manipulating money with no connection to the world of things, what might one expect. The chip squeeze/crunch/splatter is a textbook case.

        Reply
        1. Sound of the Suburbs

          When you can see what neoclassical economics does, and who benefits from it, you start to realise why they wanted this to be the global economics, and why it had to be rolled out really fast.
          You are disadvantaged if you use it.
          If everyone uses it, that won’t show up.

          Economists do identify where real wealth creation in the economy occurs, but this is a most inconvenient truth as it reveals many at the top don’t actually create any wealth.
          Much of their money comes from wealth extraction rather than wealth creation, and they need to get everyone thoroughly confused so we don’t realise what they are really up to.
          They need to confuse making money with creating wealth, so all rich people look good.
          Neoclassical economics is the pseudo economics that performs this task.

          They wrapped some fluffy stuff around it to hide what it really was.

          Reply
          1. Sound of the Suburbs

            I never got invited to the bright international future coming from a state school background in the UK.
            I never actually heard what the fluffy stuff was.
            I do know it must have been very convincing.

            Reply
            1. NotTimothyGeithner

              The fluff was everywhere. “The World is Flat”, free trade ends wars with some mumbling about France and Germany ignoring how that occurred, even just fighting dinosaur unions. Learn to code. Jobs of the future.

              Reply
          2. NotTimothyGeithner

            The neoliberals cooked up all kinds of nutty exuses, but short term profiteering is all that matters. Who cares if a town is destroyed when the CEO will retire to the Hamptons? My favorite was the US would export legal expertise.

            https://m.youtube.com/watch?v=8dGkiJcEK78 It’s from The West Wing, so it’s not technically NSFW. I wouldn’t let anyone see me watching this. Having the Democrat attack a crowd and appeal to a cop is just Sorkin magic.

            Reply
            1. Massinissa

              The US produces more lawyers per capita than almost any other country despite having a similar level of lawyering jobs.

              If it were possible to export lawyers on a large scale, we would already be doing it. Problem: FOREIGN COUNTRIES HAVE FOREIGN LAWS. Its not like we use Common Law or the Napoleonic Code or whatever is used in other places.

              Reply
              1. Krombopulous Michael

                I know that transcriptions and other paperwork were long since outsourced. Perhaps lawyers (especially younger ones) feel that they’re immune to the shrinking pie, but it’s been slowly eroding away at their incomes too. And yes, far too many of them made when there isn’t a need. Worse, those lawyers will end up “making a need” and add to the already painfully slow and expensive legal process.

                As to the argument for no foreign lawyers, well, “law” is just another set of rules to learn. I think passing the Bar exam would be the biggest issue, though I don’t know if there are any requirements to be a US citizen or to have residency in the US. If not, then I don’t see why this won’t happen at some point.
                Similar thing happened with architects, used to be a good paying profession. Now swamped with lots of foreign architects coming in on a visa, and the salaries have been forced downward.

                Medicine has kept supplies fairly tight with limited medical schools and with 3-4 yr residencies (even for experienced Drs coming from other countries). The residency ends up being a $5/hr job, which the hospitals love. Other Drs who’ve already been thru the process don’t want the hazing to stop since they had to endure it too. I know the Philippines is seeing the other side of this equation, where the top med school grads end up leaving PI and coming to the US to do nursing jobs since even being a nurse pays better than a Dr in PI.

                I’m kind of rambling, but you get the idea. Neoliberal policy and practice leaves a wake of misallocation and financial disaster for most participate. I’m sure those at the top want to turn the planet into some Victorian era society, where 75-80% are the bottom (and stay there).

                Reply
  9. Howard Beale IV

    What is missing in all of this is the fact that decades ago, the automotive companies had their own captive semiconductor plants that would make their own chips for their products – but, things being what they are being, they decided to sell the plants (or in the case of IBM, they PAID Global Foundries $1bn to take over IBM’s facilities in NY state that make the chips for their mainframes.)

    Reply
      1. Walter

        I won’t be long, Bill, until electronics is the primary business of ‘car companies’. At least, those who are smart enough to read the future… the others will be gone.
        It’s a changing world and if ‘chips’ (as is now the fashion to generically label them) are the essence of that business then a carmaker ought to damned well either be in that business or on top of it with a great logistical tie-in to those who are in it.
        US automakers were shown up as dinosaurs long ago. It started with the Japanese and the long tradition is carried on now by Elon Musk, the Chinese and Koreans… all of whom have generally surpassed the US makers.
        Next, we’ll be trying to hang Intel’s failure on ‘politics’ instead of ineptness.

        Reply
          1. Walter

            That is apparently happening now… finally. At least a CEO change has been made and the new CEO is planning to bring in some design and engineering talent.
            But now they are playing catch-up in at least 3 areas: design of chips relevant to current markets; cutting-edge chip fabrication; stable customer relationships.

            Reply
        1. Jeremy Grimm

          At one time, making personal vehicles — cars — was the primary business of ‘car companies’. When I go looking for a vehicle I do not look for the electronics, other than as a very great negative. I like simple and functional, and I believe I am not alone. Which is to say I am appalled to think that “electronics is the primary business of car companies” or should be. That makes as much sense to me as the wisdom that finance should be the primary business of car companies — as it was no long ago.

          The Japanese beat out the US car makers using a very simple strategy. The Japanese management understood the difference between total net profits and highest markup. When Japanese cars offered compact, reliable cars into a Market primed by the gasoline lines of the mid 1970s, GM doubled down on Buicks and Cadillacs, and big cars and trucks, which offered the highest markups. The Japanese established a foothood in the US market with their lower-price, lower-markup cars eventually making a place for sales of their high-end Lexus and other Japanese luxury and performance cars. General Motors[GM] worked hard to shave 10 and 15 mills of costs off the protection circuits on the input lines for their Cadillac computer controllers [based on a hearsay of a complaint heard from a computer hardware engineer in the GM Delco division where the Cadillac computer was designed and programmed], and learned tricks like putting Buick transmissions into GM trucks [the transmission in my GM Blazer — the last GM car I bought or ever will buy — died at 110K miles which I was told was more miles than most people got]. That little trick added to GM profit margins on the showroom floor and later in selling just out-of-warranty repairs — like light bulbs that last past their use ‘expiration date’. The Japanese and Koreans did not wrest market share from US carmakers, they filled a market void left by US auto management, and moved into the high-end market as US makers squeezed quality and ramped-up prices to extract short-term profits. You greatly undermine your comment in even suggesting Elon Musk surpassed the other US carmakers. Elon Musk belongs on Mars as soon as possible … or better sooner. Were US car makers dinosaurs or did US management cash-out their industry for their own benefit? Politics? really — how about greed, greed beyond human greed. You need to watch a little bit of Michael Moore’s “Roger & Me”.

          If electronics is the primary business of car makers I guess Intel or should buy-out GM. I suppose that follows the logic of the all-knowing Market.

          Reply
  10. BillS

    Thanks for an interesting article. There is one area where the USA still seems to lead in chip-making: microwave/RF (Gallium-arsenide, Gallium-nitride, Silicon-germanium, etc.) and high power switching devices (SiC and GaN). Companies like Cree/Wolfspeed, Qorvo and Analog Devices rule the roost in this area. Interesting to note that Analog Devices is based in Massachusetts whereas Qorvo and Cree are both based in North Carolina – far from the dead hand of Silicon Valley.

    Reply
  11. Walter

    Automakers are to blame for their own lack of planning. Maintaining the logistics to meet their market is their number one job and silly claims like “chips got diverted to consumer goods” are trivial excuses. Obviously Toyota got it right… so it looks like the blame goes on automakers who screwed up.
    Putting up a headline trying to blame a political attitude of ‘neoliberalism’ is childish… especially after four years of butthead conservatism in government. Another factor that seems to be comfortably overlooked is stockpiling – hoarding, if you will – by some manufaturers.
    Trying to twist stuff like this around to justify a pre-formed political opinion is pointless and you ought to be ashamed for being so stuck in that rut.

    Reply
    1. Howard Beale IV

      Toyota got it right because they still own Denso, a semiconductor company that also supplies other manufacturers with semiconductors.

      Reply
      1. lincoln

        Agreed. Toyota is part of a Keiretsu, which means a vast network of supply chain partners that give priority treatment to supplying companies within this network because of cross-shareholding associations or centuries old relationships. This means Denso, as a Keiretsu member, will give priority to supplying Toyota with chips above all other considerations.

        Reply
  12. Peerke

    It was naive of the car companies to cut orders as they did in early 2020. They panicked and one easy way to help their bottom lines was to cancel orders or scale them back. What were the Waferfabs to do? Instead of laying off workers they had a temporary low utilization period and back filled with orders from the gadgetry industry as demand shifted during the pandemic. If the car companies had held their nerve or understood what they were doing (like Toyota) then we would be talking about severe gadgetry shortages and price inflation now. There isn’t really that big of a wafer fab capacity shortfall. Demand for school from home and work from home gadgetry will fall quickly most likely as the pandemic eases. There would always be a long lag to get the auto products out from fab and assembly even if there was infinite capacity. The scale of the contractors that the car industry use is enormous especially in Taiwan. A single contract OSAT ( off shore assembly and test) building could employ 10000 workers in a 10 storey factory in Taiwan. There would be thousands of products qualified for hundreds of end customers. Part of the problem for the car industry could be that they do not understand the complexity now since they are shielded from it by using OEMs who then sub contract etc etc. So a car company “thinks” it is just shifting the burden onto the OEM but they have to protect themselves too.

    Reply
  13. ptb

    two areas where it would be nice to have more context in this discussion, IMO.

    (1) Automobile demand (outside of China) should be going down from Covid. Auto makers aren’t dumb, supply chain management is their core competence. They saw this coming from months away. It sure seems like a convenient excuse to dial back production in response to reduced car use throughout the west.. IMO it would be the correct move on their part to pretend to be surprised by it, and deflect blame to a third party.

    Speaking of China, if GM and VW have to cut back, and Japanese and Korean auto makers take over some of the western market share in the massive Chinese auto market (larger than either US or EU markets by number of vehicles sold), would be awfully interesting, wouldn’t it? It would be similarly significant if GM or VW shift a larger fraction of their operations to China, and cut back operations only in their home region.

    (2) Where’s all that high performance fab output going anyway? Bitcoin is today’s big driver of high performance computing, it would be interesting to know how much is used for *that*.

    Reply
  14. Synoia

    What puzzles me about the Chip shortage is the lack of Noise from the Military.

    For an example, look no further than the F35.

    Reply
    1. Louis Fyne

      don’t remember exactly, but at most, the entire production run of the F-35 will be a few thousand over 15 years.

      compare with the tens of millions of cars made every year

      Reply
      1. Jeremy Grimm

        You beat me! There have been DoD studies and complaints about the lack of second sources and ‘friendly’ sources for mil-spec hardware for a long time, going back to the early aughts by my own studies and probably well before that. There is long series of testimony and documents flowing from the Senate Armed Services Committee.

        Counterfeit parts are another long and often studied, testified upon, and documented problem for DoD.

        Reply
  15. Mike von Plato

    We buy chips from China. We buy Russian rocket engines for our Atlas I missile. Israel makes our “smart” artillery munitions. And during the Iraq Operation Enduring Freedom, the US Army had to buy rifle ammo from Canada.
    Make it profitable enough for a capitalist, and he will buy the rope from his hangman. (To be safely politically correct: make that he/she, his/her, and LGBTQ).
    This enhances current and temporary profitability, and is understandable from a purely capitalist POV.
    Let me suggest what amounts to a First Principle in manufacturing:
    “Initial inventions are often made in laboratories; but subsequent improvements originate on the factory floor”.
    This brings up a deeper problem: We CAN’T make many of these things here anymore, at least not without breaking international patent laws. For example, microchips were invented and, at first, manufactured in the US. But the machinery to make chips has undergone continuous improvement, and who do you think owns those patents? Ditto for flat-screen TVs and monitors, ad Infinium. Hint: English is not their native tongue.
    The US may be the world’s innovator, but our manufacturers have always resisted the cost of implementing innovations. We would still have 5,000-mile bias-ply car tires if imported radial-ply belted tires hadn’t forced their hand. Ditto disk brakes, ABS etc. etc.
    And CRT TVs and monitors.

    Reply
    1. Walter

      Perfect example: Apple elects to move its chip fab contracts to TSM because Intel just plain failed to keep up. 100% of that is on Intel’s internal management and zero of it is on government policy or any other catchphrase politics.
      The failure is not in our stars, but in us.

      Reply
  16. Greg Taylor

    The chip shortage along with excessive crypto mining demand is hindering my efforts to upgrade an ancient machine’s video. The older inefficient video cards can be had and might be the best I can do.

    Reply
  17. Jermy Grimm

    Biden’s concern about US automakers and horrors of the US supply chains reminded me of the mayor’s speech from the movie “The City of Ember”:
    “Our problems are grave. We need answers. But beyond answers — more important than answers — we need solutions. And in order to find those solutions…I propose we launch a thorough investigation. I hereby declare the creation of a task force…to investigate the blackouts.”
    “Do I have volunteers?”
    “Today we, the people…stand united against the darkness.”

    Reply
  18. run75441

    It has been slightly more than 10 years since I was purchasing semiconductors for a manufacturer of Fobs for cars which were supplied to Mercedes, BMW, Audi, Skoda, Chrysler, Marquardt used semiconductors in the fobs. The year was late 2009 and we were fresh off of the Wall Street crash and the economy bottomed out which included automotive. When demand disappears, the semiconductor companies start shutting plants down which they did in 2008/9. In 2010, we were holding daily meetings with Infineon, Panasonic, etc. trying to get additional supplies of semiconductors to meet US customer demand. Our own German owner company was competing with us for the same supply. As the PM for North America, there was heated discussions with my German equivalents and above.

    My point to this is, once those plants shut down, it is six months to start them back up again. Fab has moved out of the US pretty much. OnSemi used to have a crystal growing plant in Phoenix. I could not find it now. Packaging was pretty much done off shore all over the place.

    To give you an idea of how they work. We needed 20,000 semiconductors a month from Panasonic which came out of the Philippines. Panasonic had 6 plants there packaging semiconductors. We could not get this small amount. So Chrysler (our customer) scheduled a meeting with Panasonic. It included the VP of Purchasing for Chrysler. We talked back and forth (I was silent for once as I was the little guy)and finally the VP asked the Lead Japanese if they could give Marquardt the 20,000. Since Chrysler was also using the same semiconductor, The Lead Japanese told the VP of Chrysler, Chrysler should give us siome of the semiconductors from their allocation. I was chuckling internally as they could say something to Chrysler I could not without major repercussions.

    Here is the issue; just because the economy goes belly up, it does not mean demand goes away. We could have built up inventor for three-four months to cover ourselves. WE didn’t because everyone is just-in-time planning and Chrysler will not buy your inventory if they did not place an order for the finished product. The product did not go away and Tens of thousand of dollars was wasted blowing hot air.

    Strategically, the US has given up manufacturing to avoid the OVERHEAD as Labor is a pittance. I consulted in Manufacturing throughput. I am a LSS Blackbelt. 50 years I have been in Manufacturing and repeatedly (with a few exceptions when I had control), we box ourselves in trying to control cost and spend more.in the end.

    Few people know the process of manufacture which means they are at the mercy of the seller.

    Sorry Yves, I am on a crusade here

    Reply
  19. Mike Smitka

    Let me try to pull together 3 key points, drawing upon comments.

    0. The current issue has nothing to do with geography or international trade.
    It’s rather a function of leads and lags in the automotive supply chain, which has been whipsawed by the pandemic, from near-total shutdown to trying to run full-out to meet unprecedented demand. I don’t know why Toyota has so far indicated fewer problems, but suspect it’s because Japan and China went through shutdown early and recovery early. (I get a Japanese-langauge auto industry newsfeed but have looked at it only once the past month.)

    Now to the real issues.

    1. Without the hyperspecialization that international trade enables, the world will take 5 steps back.

    Going it alone is not feasible technically or economically. The modern world is simply too complicated to confine things with the borders of any single nation, even ones as large physically and economically as the US, or the EU or China (all three comparable in GDP and all three continental in geography).

    See Fazal Majid’s and other comments on the nature of the semiconductor industry. I add automotive examples:
    • There are auto parts that cross borders within the USMCA (former NAFTA) 8 times, highly specialized plants carrying out different manufacturing steps. They can’t be relocated, each has its group of managers and engineers and skilled workers appropriate to that one plant. Some locations were due to the chance accumulation of production capabilities in a location when new needs arose, one to the availability of inexpensive hydropower. I was told the most skill-intensive plant, and hence the hardest to move, happened to be in Mexico.
    • One plant in the small city of Zeeland, Michigan handles the majority of global production of autodimming review mirrors. The #2 firm, not by chance, is a couple miles down the road. Why Zeeland? – it’s another of those idiosyncratic location stories. But those mirrors get assembled into housings in plants all around the world, they deliberately did not get into the plastic molding and color matching skills needed to supply car companies directly.
    • I’ve visited a highly specialized printed circuit board manufacturer in western Michigan that exports a key product to China, where it is put into a sub-assembly that is then shipped back to the US and goes through one additional supplier before going to a final assembly plant to be put into a car. It’s a competitive industry, lots of players around the globe, and a product that’s easy to ship, even to airfreight when speed matters. So once you as a PCB user find a supplier for a specialized product, and your engineers develop working relationships with their engineers, you stick with them, the costs of switching outweigh all other cost considerations. In this case, the nexus happens to be in the US, but policy can’t create such micro-level networks combining skills, production capacity, and personal relationships.

    The same is true of R&D – the sheer range of technologies under development at any given time outstrips the capability of any single country, including the science end of the spectrum.

    • I asked a physics post-doc why in the world he’d ended up in Japan. His response was that while the Harvard-MIT nexus was great, those labs could only cover so many bases. So as his cohort moved on to post-docs, they scattered around the globe to be in the one lab that was the player. In his case, it was that of a prof at a Japanese university.
    • Brazil, for example, has 2,000 different sugarcane cultivars and is where you must be to work on bioethanol fermentation (US corn-based ethanol is a dead-end, a creation of subsidies pushed by our farm lobby).
    • I’ve been a judge for a global auto supplier innovation award for 25+ years, the R&D facilities of the big Tier I’s are scattered globally, they can’t do everything in one place, and different locations/countries have different sorts of technical expertise. Shanghai is one such center, central Japan another, metro Detroit the third, the Paris – Munich corridor the fourth. Those in Japan and Europe aren’t “tight” so you can’t easily have a meeting at a supplier or customer in the morning, and be back in your home facility in the afternoon. You can do that in Detroit or Shanghai. But there are key facilities on the technical side ranging from Ciudad Juarez in Mexico to southwest Poland, the Tyrols of Italy or Shenzhen in China.

    Within economics, received trade theory in the 1980s was still implicitly based on the country-source model. Paul Krugman helped bring economies of scale models to trade theory, and also reinvigorated – along with a Japanese colleague – the field of economic geography, all stemming back to where the grad students of Walter Isard ended up.

    2. Industrial policy is determined by incumbents and politics. Policy is more likely to reflect the interests of incumbents facing declining markets as to focus on potential future needs. I happened to be at a National Academy of Sciences meeting with, among others, reps from Semitech. What the industry wanted was money to let them compete against Japan in making RAM chips. Had they succeeded, it would have killed the US industry, as it did a couple of the big electronics groups in Japan that diversified into making chips. Why? – it ended up a competitive industry, cutthroat competitive. The first to market in each generation made money, the third firm on down were bloodbaths. No one at that meeting foresaw that microprocessors would be huge and immensely profitable.

    I (co)translated the seminal Japanese-language study of industrial policy, the big money was rice farming, coal and shipbuilding, all of course the future drivers of Japan’s economy. The protection that the auto industry got was designed to prevent used car imports, and to keep GM and Ford – who dominated the industry into 1936 – from reestablishing operations. But in Japan in the 1950s, when policies were put in place, no one foresaw the importance of automobiles and consumer electronics. One Mr. Toyoda, for example, was to meet with Henry Ford II in June 1950, hoping to sell the company. But it was scheduled for the day the Korean War broke out, and the meeting never took place.

    3. Back to semiconductors. So what should we want? The chip design capability? the manufacturing capability? the capability to design chips into products? the ability to make those products? the marketing and branding of those products? And if so, which chips? Will those be the ones in short supply the next time around? The industry is becoming more specialized, chips aimed at datacenters, chips for AI, chips for cell phones, if systems on a chip, which systems? Then there are all those microcontrollers, they don’t need state-of-the-art 7nm circuitry, old would be fine but keeping old plants open quickly becomes uneconomic, they use out-of-date equipment for which spare parts and maintenance engineers can’t be had.

    4. So we reshore … while labor markets aren’t tight at the moment, there aren’t any places in the US where you could quickly hire thousands of workers with the computer literacy, physical stamina and picayune personality that manufacturing needs. One-third of the local population are too obese to spend the 10- to 12-hours on their feet required to survive in the plant where my son works. And these plants have to compete with careers in healthcare, which employs 30% more than all of manufacturing, is growing strongly, and is largely immune to recessions. So should we close hospitals to make computer chips?

    Obviously I’ve been thinking about industrial policy for some time, I will work on developing these sorts of points, this is a first pass.

    Reply
    1. Synoia

      And these plants have to compete with careers in healthcare, which employs 30% more than all of manufacturing, is growing strongly, and is largely immune to recessions

      That’s a good thing? Or a symptom of a society encumbered with much disease?

      Reply

Leave a Reply

Your email address will not be published. Required fields are marked *