Yves here. “Never let a crisis go to waste” comes to Silicon Valley, in the form of seeking a handout for a chips crisis that they had a large hand in creating. Recall that your humble blogger for years has put chips high if not on the top of the “WTF were they thinking?” list in terms of the US making itself dependent on foreign sources for critical supplies. As the post by Bill Lazonick and Matt Hopkins explains, tech titans like Apple that both spent kazillions in buybacks and whose orders made particular Asian chip makers into industry leaders are now in line for subsidies, when they have more cash than they can deploy in their operations. As one investor said, “Why should I invest in a business that won’t invest in the business of its business?”
By William Lazonick Professor of Economics, University of Massachusetts Lowell,
President, The Academic-Industry Research Network; and Matt Hopkins, Senior Researcher, The Academic-Industry Research Network. Originally published at the Institute for New Economic Thinking website
The Semiconductor Industry Association (SIA) is promoting the Creating Helpful Incentives to Produce Semiconductors (CHIPS) for America Act, introduced in Congress in June 2020. An SIA press release describes the bill as “bipartisan legislation that would invest tens of billions of dollars in semiconductor manufacturing incentives and research initiatives over the next 5-10 years to strengthen and sustain American leadership in chip technology, which is essential to our country’s economy and national security.”
On June 8, 2021, the Senate approved $52 billion for the CHIPS for America Act, dedicated to supporting the U.S. semiconductor industry over the next decade. As of this writing, the Act awaits approval in the House of Representatives. Our INET working paper “When the CHIPS are Down,” documents in some detail a curious paradox: Most of the SIA corporate members now lobbying for the CHIPS for America Act have squandered past support that the U.S. semiconductor industry has received from the U.S. government for decades by using their corporate cash to do buybacks to boost their own companies’ stock prices. Among the SIA corporate signatories of a letter to President Biden in February 2021, the five largest stock repurchasers—Intel, IBM, Qualcomm, Texas Instruments, and Broadcom—did a combined $249 billion in buybacks over the decade 2011-2020, equal to 71 percent of their profits and almost five times the subsidies over the next decade for which the SIA is lobbying.
In addition, among the members of the Semiconductors in America Coalition (SIAC), formed specifically in May 2021 to lobby Congress for the passage of the CHIPS for America Act, are Apple, Microsoft, Cisco, and Google. These four firms spent a combined $633 billion on buybacks during 2011-2020. That is about 12 times the government subsidies provided under the CHIPS for America Act to support semiconductor fabrication in the United States in the upcoming decade. Do these companies need $52 billion in subsidies from U.S. taxpayers to give them incentives to invest in semiconductor fabs?
The main purpose of these hundreds of billions of dollars in buybacks has been to give boosts to the stock prices of the repurchasing companies. As a result of buybacks, a company’s stock price increases a) if, as is typically the case, stock traders bid up the price when a company announces a repurchase program giving the CEO and CFO the authority (but not the obligation) to do a certain value of open-market repurchases (say, $10 billion) over a certain time period (say, three years); b) when, at the direction of the CFO, the actual execution of buybacks by the company’s broker on any particular day or series of days increases the market demand for the company’s shares; and c) if, with the release of the company’s quarterly financial report, the increase in the company’s earnings per share (EPS) because of the reduction in shares outstanding prompts stock traders to bid up the price of the company’s stock even more.
From October 2012 through June 2021, Apple alone spent an astounding $444 billion on buybacks, equal to 87 percent of its net income. That is in addition to another $114 billion paid out as dividends, representing an additional 22 percent of net income, over these eight and three-quarter years. With just a fraction of those funds wasted on buybacks, Apple could have invested directly in a state-of-the-art fab to produce its own chips on U.S. soil.
Back in August 2010, electronics journalist Mark LaPedus published a column, addressed to Apple CEO Steve Jobs, entitled “Apple should build a fab.” At the time, Apple was reliant for chip fabrication on its emerging smartphone competitor, Samsung Electronics. LaPedus recognized that “in an age when real men go fabless, I concede it’s an unconventional idea. You might think it’s absurd. But an Apple A4 fab today could keep the iProduct franchise in hay—and Samsung at bay.”
In August 2011, two months before he passed away, Jobs had handed over the CEO position to Tim Cook, the Apple executive who had added value to the company by outsourcing manufacturing of devices to Hon Hai, enabling the Taiwanese company’s Foxconn subsidiary in China to emerge as the world’s leading electronics manufacturing services provider. Under Cook, Apple shifted its chip-fabrication contract from Samsung to Taiwan Semiconductor Manufacturing Company (TSMC), the pioneer in the “pure play” foundry model, which, largely because of Apple’s contract, is now the world leader.
This past April, TSMC announced plans to invest $100 billion in fabs over the next three years, including a $12-billion 5nm facilityin Arizona to fabricate Apple’s M-series processors. In potentially providing $52 billion of taxpayers’ money to support the U.S. semiconductor industry under the CHIPS for America Act, federal lawmakers might contemplate the fact that TSMC’s multiyear investment of $12 billion in a state-of-the-art fab in Arizona pales in comparison to the vast sums that Apple has been routinely spending on buybacks: $73 billion in 2018, $69 billion in 2019, $72 billion in 2020, and $66 billion in the first nine months of 2021. A country, never mind one enormously rich company, could have built a lot of state-of-the-art fabs with all that cash.
As a key part of its trade war with China, one year ago the Trump Administration coerced TSMC to cease manufacturing chips for Huawei Technologies, which had emerged the previous year as the world’s leading smartphone company, ahead of Samsung and Apple. In 2019, Apple had accounted for 24 percent of TSMC’s revenues and Hi-Silicon, Huawei’s wholly-owned chip-design subsidiary, 15 percent. In the third quarter of 2020, TSMC began shipping smartphone chips to the Chinese company produced on the fabricator’s 5nm technology. In that same quarter, 59 percent of TSMC’s revenues came from North America, followed by 22 percent from China. In the fourth quarter of 2020, revenues from North America soared to 73 percent of TSMC’s total, while those from China plummeted to six percent, as TSMC complied with U.S. government directives to cut off Huawei’s chip supply.
At the same time, 5nm wafer revenue, predominantly from the fabrication of the most advanced smartphone processors, which had been zero percent of TSMC’s total in 2Q20 and eight percent in 3Q20, jumped to 20 percent in 4Q20. In supporting TSMC’s revenues and profits by increasing its purchase of 5nm chips, Apple in effect partnered with the U.S. government to demolish the smartphone business of Huawei, its prime global competitor.
The key weapon that the U.S. government used to force TSMC to cut off Huawei was the credible threat that leading U.S. semiconductor equipment companies—most notably Applied Materials, Lam Research, and KLA—would otherwise withhold sales of their critical products to the Taiwanese foundry. These three companies, which rank #1, #3, and #5 respectively in the global chip-equipment industry, grew immensely in the 1990s, with support from Sematech, a non-profit consortium of 14 leading U.S. semiconductor companies that received $500 million from the U.S. government between 1989 and 1996 in response to Japanese competition in the machinery segment of the industry.
During the 1990s and into the 2000s, Applied Materials, Lam Research, and KLA became innovative world leaders by retaining all their profits and reinvesting in productive capabilities. From the mid-2000s, however, they all caught the American financialization disease. Over the decade 2011-2020, the three companies combined did $30 billion in buybacks, equal to 76 percent of their profits, while also distributing 35 percent of profits as dividends. Our research suggests that if these companies remain focused on using their profits to prop up their stock prices, innovative equipment suppliers, most likely based outside the United States, will take over the segment’s top spots in the next decade or two.
In providing the U.S. semiconductor industry with $52 billion in subsidies under the CHIPS for America Act, Congress could require the SIA and SIAC to extract pledges from its member corporations that they will cease doing stock buybacks as open-market repurchases over the next ten years. President Biden should be open to this idea. Exactly five years ago, as vice president in the Obama administration, he published an op-ed in the Wall Street Journal that was highly critical of buybacks, stating that “government should…take a look at regulations that promote share buybacks, tax laws that discourage long-term investment and corporate reporting standards that fail to account for long-run growth. The future of the economy depends on it.” A moratorium on buybacks by U.S. tech companies could also be a first step in rescinding Securities and Exchange Commission Rule 10b-18, which has since 1982 been a major cause of extreme income inequality and loss of global industrial competitiveness in the United States.
14 comments
Comments are closed.
Lots of working parts here to consider. I doubt that this movement will be a success however as the current financialization of American business will work against it. Silicon valley will probably take the free money, slow-walk any work on setting up these foundries, and then wait for a future change of administration so that they will be allowed to sell to Taiwan what they have developed. What they probably want is like what big pharma has – the government provides money and solid research while big pharma, does a bit of fine tuning, slaps a patent on it and then goes off to the bank. So Silicon Valley is probably hoping that the government pays for all the construction and the like so that they can keep on doing their stock buy-backs. And it is not like building these places are easy. Not according to TSMC’s founder-
https://asiatimes.com/2021/04/tsmc-founder-doubts-us-competence-in-chip-making/
It’s not that American workers are dumb or anything but that the expertise of the work force has to be built up and nurtured over a number of years. And that is something that you cannot buy off the shelf. As an example of what I mean, back during Cold War One an American naval officer was talking about the expertise required to run and operate a super-carrier. He said that the US Navy could hand over one of their super-carriers to the Russian navy from the nuclear plant, the aircraft, ancillary equipment and right down to the chalk on the boards for keeping track of the aircraft. But that it would take the Russians at least ten years to learn how to operate it. Same here. So it is not a matter of building a foundry and they will come. It will be more a matter of will there be able to be enough people to come if one is built.
One of the big by-products of SIA, DARPA, and other directed government research grants back in ‘the old days’ was a base level of funding into graduate programs, that ultimately led to training those necessary bodies. So a lot of US universities had graduate research assistantships available that were filled by students eager to learn. In the 80s and 90s, the electrical engineering graduate program track in solid state physics was dominated by students from South Korea and Taiwan, who at the time almost 100% returned to their countries.
So there will be loss whatever you do. But one thing is certain, is that interest in this area needs to be seeded early. My kids are in middle-school and high-school. When the schools talk tech, they only go as far as Java programming. Maybe robotics (unfortunately with Lego kits). Yeah, they learn something, but jury is out on what real innovation comes from it. Not sure the future innovators are getting a good exposure of what tech is beyond dog-walking apps. 40 years ago Silicon Valley was filled with all sorts of startups that were Applied Materials wanna-bes.
Funding like this is a blunt instrument. Stock buybacks should have remained off limits. Offshoring manufacturing should have consequences (costs) to the companies to mitigate the social costs that they leave the rest of us by employing those strategies (not unlike pollution).
From it it seems as if the business of offshoring routinely ignores the loss of manufacturing know how as a cost.
Once gone it takes a real investment of time and money to get any industrial / manufacturing operation up and running profitably.
I’m actually familiar a bit with fabs. Austin has a bunch. But there are a few things I don’t understand.
How has TSM been able to produce 5 nm plants while the litho equiptment is produced by the US and Netherlands and apparently no one else has production lines below 11 nm? What is the real situation with Samsung? Why in a business producing huge profits which uses relatively small labor inputs are we trying to pay line workers in bunny suits $20/hour when even random plumbers and electricians make twice that much?
A FAB which costs roughly $11 billion dollar to build has approximately 1,000 employees. Even if they were paid $100,000/year the FAB would have a labor cost of $ 100 million dollars/year. I believe Tiawan and Korea (and to some degree the mainland) have “the Japanese advantage” of the 1970s-80s; a very organized, very competent labor force which works for less than an American labor force of the same quality. But labor costs are a small input in the chip business.
The same goes for the PHDs. I personally know one very high level researcher of east Asian descent who was offered a seven figure salary to relocate back to Asia. He didn’t take it. He, his wife and his kids like it here. American Universities have come to depend on foreign grad students in the hard sciences because they are cheaper- the foreign governments pay the freight. The day that American kids with a first-rate BA in Physics or Computer Science get $50,000/year to persue a PHD in their subject instead of the $15,000-and-tuition they are offered now because of the glut of foreign students, is the day that we regain our technical edge. $15,000/year versus $100,000/year from Indeed or Google- or Wall Street is a no-brainer for many kids I meet who would love to get a PHD in the hard sciences. And the total cost to the US of subsidizing these students is small.
During the 1960s science grad students even at 2nd level universities like the one I went to were earning enough from their assistantships that they were able to get married and start a family, something which is no longer possible because the universities say “Why should I hire you when I can get a Chinese or Turk to work here for 1/4 the price?”. Providing real fellowships to US grad students would be cheap, although I’m sure the usual Congressional grease to provide grad student for mediocre state universities and other “favored” mediocre schools would be required. And if you think there is a shortage of good, US, non-degreed techs in the chip fab business, begin to pay them as well as the oil companies pay techs who work on the offshore rigs and out-of-the-way places like the Permian Basin and South Dakota. Money talks.
@ Dave, I think there is a bit of marketing fudge with TSMCs ,”5nm” . Transistor density is reported as 170million per mm2, against 100 million for Intel’s 10nm, which intel will now rename to “intel7”. Intel can’t quite get themselves to call it 7nm
In a straight comparison, a 5nm process should have twice the density of a 7nm process, so Intel’s rename is still a bit on the conservative side.
The TSMC process is clearly better than Intel’s, but the difference is not quite as big as the numbers suggest.
There so much over-simplification over fabs and so forth, often along the lines of “why are chips being produced on old processes”.
Two notes on the auto industry:
1. Auto industry cancelled orders (or didn’t re-order) last year for product that is largely auto-specific and longish lead time from wafer start. Now they are blaming their suppliers. Really? Not your fault?
2. The Intel etc small geometries are for massive digital circuits, i.e. CPUs, and have a core voltage under 1V. They aren’t suitable for moderate voltage (5V and higher) and mixed signal (analog plus digital) circuits.
Lastly on Intel… one of the ways to save power is to turn off sections that aren’t in use, and another is to dynamically drop the voltage for sections to be just enough to run at the speed. The CPUs have a lot more analog sneaky stuff that might be expected, and are difficult (slow) to port to a smaller process node.
It’s misleading to look at this money as an investment in the activities of these companies, as the article frame the issue. The companies are wildly profitable, they do not need the US government to invest, they do not need anyone to invest in them.
It makes more sense to look at it at as deal, a bribe if you want. The US government wants companies to do something unpopular, namely build production facilities in the US.
TSMC is based in a US ally, so they could be extorted to do so (with some spoonfuls of sugar to make the medicine go down ). The US is less willing to extort companies that are already mostly based in the US. That would be counterproductive if you want to encourage them to stay there. So, those companies have to be paid. Apparently, the price ticket was 52 billion in various subsidies and goodies.
Now, the people in this article want those companies to do something else that they don’t want to do, namely increase investment beyond their original plans. That’s perhaps a laudable goal, but it’s also a renegotiation of the deal. If you want more, you have to pay more.
I suppose the authors would argue that the original deal was bad, that the US is overpaying for what it gets in return, and should demand more for the money. But such overpayment is somewhat inherent in “national champion” style industrial policy. If you aim to have powerful, world-domineering companies in your borders, then you might end up with powerful companies in your borders, who are not easily bossed around.
So if the carrot doesn’t work, why not the stick; i.e., pass legislation that explicitly limits federal subsidies for or contracts with companies that buy back their stock, pay out large cash dividends, or are the subject of leveraged buyouts? Otherwise, such legislation is just directly funneling public money to these companies for socially and economically damaging behavior that benefits a relatively small percentage of the population financially while public needs are going unmet. Alternatively just make stock buybacks illegal, as they were until the early 1980s.
I don’t think that’s a stick, it’s just making the subsidy-carrot less attractive by adding extra conditions. Which means that you’ll have to make the subsidy larger, to get the same response.
Would you be willing to do so? The industry says, we’re willing to limit buybacks and dividends for the next 5 years, but then you have to pay us an extra 40 billion (or some other number) on top of this 52. Of course, they will formulate is in more palatable language.
That’s the dirty side of industrial policy, especially the kind that creates powerful companies. Part of the money will end up in the grubby hands of insiders, through one channel or another. If not through direct subsidies, then indirectly from their strong market position.
This is very interesting. Because it puts a whole new light on offshoring our technology in the 80s and 90s. I always assumed it was done to arbitrage American labor and make a bigger profit from foreign labor – until, of course, those companies could not make enough from the American market because it had fallen into such a deep recession. Now this makes me think we off-shored as fast as we could to set up a global consumer market. Because it was going to happen anyway. So it was twice the crime – first to save on labor costs and second to create a new/global market. If we had had proper government we’d never have suffered this race to the bottom at all. And now that the gloves are off with China, we are not messing around with setting up factories in Asia – we are now putting embargoes on semiconductors and other stuff to literally put them out of business. Things are just as insane as they have always been. That never changes.
i suspect that the idea for the bill, wasnt the top line chips, but those that are causing train wrecks all of the US economy today (see auto OEMs who cant produce vehicles because of the chip shortage, and many others), and i think thats what they are addressing. today, we cant get the chips from Asia very quickly (see the ships off the west coast that have been anchored there because they cant off load their containers). but when looking at US companies in general, they really cant seem to invest in their businesses any way. and with the …..US view of corporations as being for investors, with no other purpose, and that will end up with the best for all (you might notice that those who say this…never explain how that actually works…but i suspect they dont really care either). and the buy back plans….are just another way to increase stock prices …that also feed into executive pay, what could ever go wrong with that? while in the short term plan sounds good (though GOP support for it. taints more than a little. maybe
make it so that companies taking support cant do buy backs too? i suspect that we might not have many takers
just a thought, what if we restricted the companies who got funding so that they cant see the company to sny one else for 10 years, unless they paid back their funding?
and with the global transport problem, US companies that are so dependent on long supply chains, may look to go else where. no guarantees of that.
more on that chip shortage
https://www.youtube.com/watch?v=cEdKaEVNg3M
may take a few years to fi it self. if it does
I hope the CHIPS debate will help us reflect on U.S. corporate use of contract manufacturers, and how these entities have helped drive the industrial policies of other countries. Contract manufacturers should be at the core of any discussion about industrial policy or supply chains or outsourcing.
When U.S. manufacturers move their production overseas they usually don’t build a new factory in a low labor cost country, but rather they outsource the entire production to an independent contract manufacturer that will build a product on their behalf in a factory overseas. Contract manufacturers build factories with massive subsidies that they arrange from overseas countries, which is important because a factories contribution to the total manufacturing cost can greatly exceed the cost of labor, as well as arranging cheap local labor and low cost materials. An outsourcing manufacturer prefers to simply pay a fee for the production of each unit of product, because the contract manufacturer will bear the risks and the costs of plant, materials, and labor. And the contract manufacturer probably hires a construction company like Fluor to actually build these facilities.
These contract manufacturers create products overseas for many large U.S. companies. Foxconn manufactures devices for Apple (iPhones, iPods, iPads, and the Apple Watch), Hewlett Packard (servers), Dell (desktops & servers), Amazon (Kindle), and Sony (televisions). Quanta Computer manufactures Apple notebooks, Flextronics manufactures Microsofts Xbox, and Compal Electronics manufactures Dell laptops. Nike’s footwear is produced by hundreds of independent contract manufacturers in 14 countries, and Gap clothing (Gap, Banana Republic, Old Navy) is produced by hundreds of independent contract manufacturers in 29 countries. Taiwan Semiconductor manufactures chips for Apple, Qualcomm, AMD, and Nvidia.
Many companies no longer want to pay for upgrading or building new factories because for the last decade contract manufacturers have been willing to do this for them. Hopefully we can find a way to discourage this without handing out massive subsidies like those given to the natural gas industry or Tesla.