Transcription
I'm Bill Lazonick. I am a Professor Emeritus of Economics at University of Massachusetts and President of the Academic Industry Research Network.
So Intel was founded in Silicon Valley in 1968 by Gordon Moore, who's famous for Moore's Law, that the power of chips would've increased, double every 18 months, and over time. And Robert Noyce, who was one of the co-inventors of the integrated circuit in the late fifties. And they had been at a company called Fairchild Semiconductor, which was a pioneer in really producing integrated circuits in the 1960s, mainly for the military at first. And they learned there how to really, no one knew how to manufacture these things, how to manufacture integrated circuits, what became known as chips with, you know, low defects, high yields. And they founded Intel in 1968. And actually when they founded Intel, they didn't have a separate R and D department. 'Cause the whole thing was really R and D, you know, figuring out how to do these things. They also focused on, well they were focusing then on memory chips, just to keep the memory in devices and computers. And they also said that they did not want to do military work. They wanted to do commercial applications. So you had people who knew what they were doing and in a really new industry, had figured out how to produce chips, semiconductors. And in 1971 already, they developed a computer on a chip. So that is one, which we now know as a microprocessor. So now you get logic on the chip. And so they were in the forefront of that. But the main business, right until the early 1980s at least, was these memory chips that they were producing.
But when they, ultimately, the Japanese out competed in the memory chips in the mid eighties, and they then moved into logic chips, particularly microprocessors. They started producing those in Oregon in 1974. That was their plant. And actually since then, or at least since the eighties, it's been the biggest location for Intel employees, not Silicon Valley. When they shifted, well really the big shift to having a market from memory chips was when they got the franchise for the IBM PC, which was developed from 1980-81, came on the market and then dominated the PC market. And later, in the nineties, they called it Intel Inside, you know, so it was kind of branded, the chip inside there. And by the early 1990s, they were the leading chip producer in the world, and also were fabricating their chips.
Now in the 1980s, there was a number of companies that saw, okay, we can get other companies to fabricate our chips, which is the really capital intensive part. And often, and its part because of Moore's law, the equipment was always changing. And so it also became very, very expensive in terms of capital expenditures and kind of, you're destroying your previous investments 'cause there's new, new, new technology you have to invest in constantly. And they became the world's leader really in semiconductor fabrication. But in the late 1980s, particularly in Taiwan, this company, TSMC, Taiwan Semiconductor Manufacturing Corporation, started as what they call the Pure Play Foundry. So now there was not enough chip companies that were designing chips that would want to get them fabricated, and they became a specialist in that. And later, Samsung Electronics became a leader in fabrication too. But they also produced chips for their own, actually for their own devices. Not only their own chips design, but their own devices later, much later smartphones.
Okay, so skip ahead. Intel is really a dominant company in the 1990s and into the 2000s. It's a leader technologically. And around the late 1990s, their stock price is booming. And also what they're doing is what a book called Sustainable Prosperity in the New Economy? that I published in 2009, talked about, they were a new economy company in the sense that what was really important was that they could not offer their employees full-time employment for a career. You know, they couldn't say we'll be around in 10 or 20 years, 'cause they didn't know, like other companies were doing, like in Silicon Valley, Hewlett Packard, you'd have a career for life and a fine pension. So they gave people stock options. And in the 1990s, those stock options started becoming quite, they gave up quite a lot of them to a broad base of employees, and so they started buying back their stock ostensibly to offset the dilution from the stock options. But then they found that buying back the stock was a good way of just getting their stock price up. So that became very widespread in the late 1990s.
Now that didn't mean that was taking away from R and D. And actually, it's never been an issue of has Intel spent enough on R and D. It became a world leader by actually reinvesting its profits, being able to track the best workers, despite the fact that you had, you couldn't offer workers, which most certainly college educated workers had, white collar workers had what I call a career with one company, at older companies like IBM and Hewlett Packard. They didn't need to have these workers stay there, these high level workers, working as teams to engage in learning and how to do these things constantly. So stability of the labor force was important. And that actually became a big issue, not just at Intel, but in the Silicon Valley model generally, which would include a company like Microsoft in Seattle, is that there's a lot of mobility of labor, you see, because 'cause workers would go to new startups or they go to other companies, they'd offer 'em more stock options, et cetera. That became a big deal in the late 1990s. And they had to manage that mobility of labor. Really, that was the essence of getting productivity out of their R and D spending was keeping the people there, the teams there, learning how to implement, particularly, all the new technology. There was a design of the chips, but lots of companies were doing that. What was really critical was how you can manufacture these chips, more and more logic chips, at very high yields, low defects and they became a leader in that.
A couple words about shareholder value. Right next to here, where we're doing this interview is Harvard Business School. And I was there in the mid eighties, that was about 40 years ago. And I came there after being in the Harvard Economics Department, which is across the river here. We're in Boston, it's in Cambridge. But at Harvard Business School in 1984, when I came there after a few years doing research, no one was talking about shareholder value. By 1986, everybody was talking about it. That's 'cause in between, they hired a guy named Michael Jensen who was the guru of maximizing shareholder value. And that's the only purpose of a company. And that goes back to arguments that Milton Friedman had made and others at Chicago school. But basically the East coast liberals at Harvard and MIT, they've never critiqued that perspective.
Now for Intel, it really wasn't about shareholder value, it was about innovation in the beginning. And to do that, they engaged in what I call retain and reinvest. They retained their profits, reinvest it. But unlike the old economy companies, the stock market played a lot more role at Intel, particularly in the compensation of labor, not just top executives, a broad base of employees who were getting stock options. So they became very aware of the stock price. And then also, let's say in 1997, Andrew Grove, who was like the third Intel employee and now was the CEO, all of a sudden he made $97 million from his stock options, which, you know, has been repeated many times since then. But that was a lot and many people said, "Whoa, what's going on here?" You can make a lot of money from this stuff if you get to the top particularly. But even actually, if you look across all their employees, in the late nineties, I have it in my book, I think it was something like $150,000 from employee from stock options in addition to their salary. Microsoft, in 2000, it was about 450,000 per employee in addition to whatever they were getting, maybe 125,000, 150,000 as salary. So this became, everybody got an interest in this.
And then of course, when the stock market went down after the internet boom turned to bust, and that happened to all companies, whether they were good or bad, they started saying, "Okay, let's boost up our stock price doing stock buybacks." And Intel had done this in the late nineties. They saw this as a good way of getting their stock price up. The stock price would've gone up anyway because they were dominating the market. So the stock price can go up through innovation, speculation, manipulation, which is what buybacks do. But Intel had, particularly in the early nineties, early 2000s, they saw, okay, let's just do, and it turned out to be 5 billion, 10 billion, then even 15 billion in buybacks per year just to get our stock price up. And so that kind of set in in this period, and basically between, as I've shown and documented some papers that have been published by INET on this, I think 2021, with Matt Hopkins. From 1998 to 2020, under four different CEOs, they did $142 billion worth of stock buybacks. And this is at a time that it wasn't that they weren't spending on R and D, they were spending maybe even $10 billion, at some point, a year on R and D, maybe 15 billion on capital expenditures. And they were very profitable and they were using a lot of their money to just buy back their stock. And that is the context in which they became financialized.
Then the question is, okay, if they had enough money to invest in R and D and capital expenditures, you know, they were super rich like a company like Apple and Microsoft and a few others, why does it matter? Well, it mattered because there were other things they could have done with that money that they didn't do. And that I attribute, not to a lack of money, but who is really running the company, what they called strategic control. And basically, when they started doing a lot of the buybacks, it was under a CEO named Craig Barrett, who was the successor to Andrew Grove from 1998 to I think about 2005. And he was an engineer, but he then got into, oversaw this trying to prop up the stock price both in the boom and in the bust. Then in 2005, they put a guy named Paul Otellini in as a CEO and he was more a finance guy and he was there just to really, to get the stock price up. And under Otellini, they did a few things that, looking back, they regretted that they could have easily done with the money they were spending on buybacks.
So people now have heard of Nvidia, which has had as high as a $3 trillion capitalization. Whereas Intel is, you know, down, you know, 150 billion or something, I'm not sure the exact number now, but a fraction of that because of AI chips. Well, they had the opportunity to buy Nvidia in 2005 for $20 billion. They spent more than that in 2004-2006, just buying back their stock, just wasting money. 2007 they had the chance to get the Apple kind of contract for Apple microprocessors, which is for smartphones. And, you know, and Apple of course being in the lead, even a few years later, that would be hugely valuable 'cause those are the highest end chips that are being produced and still is the case. And they passed on that and Apple gave that instead to Samsung Electronics, which then became a leader in fabrication because they had this in particularly the high end, which is called like, it's kind of low nanometer, like at that point it would've been maybe 20, 15, 14 nanometer, something like that in the 2010s. But then Apple realized that Samsung, well of course, became a competitor of theirs in devices and around 2011, they switched to a company I mentioned before, TSMC, which Apple still uses. And TSMC is a leader in basically the high-end chips and in chips altogether, I mean, in fabrication in large part because they got things like particularly, the Apple contract, which Intel gave up.
So in retrospect, while they were just doing what they were doing, they seemed to be doing all right, making lots of profits. It was all Intel inside. You know, their biggest customers are still HP, Dell and Lenovo, which are all, you know, computers. It's still computer chips. That is a big part of their revenues. They seem to be doing okay, but in fact they were falling behind because they weren't really producing at the very highest end. And at least until, even into the pandemic, it wasn't that they lacked money, it was they lacked strategic focus. And that was under Otellini, then there was another guy named Krzanich and then another guy named Swan who were the CEOs. And they were there really just to oversee Intel's dominance and distribute cash to shareholders. So unlike retain and reinvest, I call that dominate-distribute. And then some point they come to downsize and distribute. They distributed so much cash, they actually has to downsize the labor force. In a few minutes, I'll introduce a fourth category, retain and reinvest, dominate-distribute, downsize-distribute. I'll introduce a fourth category that applies to Intel in actually 2024, but I'll come to that. But basically by 2020 they realized that they had fallen behind TSMC and Samsung and there aren't that many companies that are big players in fabrication. There's Global Foundries, Texas Instruments, American company does it, but it doesn't do the high end. It does stuff, chips for automobiles and other devices that aren't high end. There is SMIC in China, which has always been trying to catch up to TSMC and Samsung, and Intel. So there's not a lot of big players out there and only a few really Samsung, TSMC and Intel that can really produce at the highest end using the latest technology, which is now called EUV technology, produced by a Dutch company, ASML, which machines cost 300 million and very hard to implement. You don't just buy them and then they work. You have to have all kinds of engineers working on them for years to get them to work, to produce the various high end chips.
Okay, Intel was falling behind by the beginning of the pandemic. So then basically, for reasons which are not all together clear, but which Matt Hopkins and I wrote about in our papers on this in 2021. In January, 2021, there was a strong demand, you're into the pandemic, there was actually a very, you know, if you're in the semiconductor industry or you know Apple and you know, any computer industry, you're making lots of money in the pandemic. And so the market was there. Intel actually was getting record revenues and profit, but they were falling behind in the high end chip fabrication. So the board decided to get rid of this guy, Swan, who was purely a financial guy and bring in a guy named Pat Gelsinger, who had actually started at Intel at I think the age of 18 or something, back in the 1970s. In his late twenties, he had become head of the division of Intel that was producing the microprocessors, it's called x86 architecture for, you know, Intel Inside. He had been their first chief technology officer in the early 2000s. And they could have made him CEO around 2005 instead of Otellini, but they wanted a finance guy. So actually, Gelsinger role was to tutor Otellini in technology before he became CEO. He then went elsewhere to, I think it was EMC, which a storage company bought by Dell and then VMware, which was another company bought by Dell. And then, so being a CEO there, but he was brought back, you know, and he started in mid-February 2021.
And what caught our attention, and one of the reasons we wrote articles on Intel, I've been following Intel for a long time 'cause he's the first CEO who I've ever seen take the job and say to the board, "I'm taking this job on the condition he do no more buybacks." And he said that in a number of interviews. Now, I don't think Pat Gelsinger would be kind of opposed to buybacks per se, the way I am, that they're just a manipulation in the market. He would just see them as a waste of money and they needed to invest. And so he said, not only that we're getting rid of buybacks, but we're gonna invest in our learning processes, in our capabilities. So he understood it wasn't just that they were wasting the money, that there was all these people at Intel who were engineers primarily, who had all these capabilities, but their skills and efforts weren't being pulled together to implement the latest technologies. And he said, okay, by having someone in charge of the company who said we're gonna invest in the future, that he could mobilize their efforts. And I think he started to do that. And Matt and I have been actually talking to some intel engineers through LinkedIn. And then there's actually someone who is an Intel engineer for 24 years who got in touch with us and we're having a long conversation. It would probably be a long conversation this afternoon about what went on there, because that's the hard, it's really hard to know what's going on inside these companies in terms of these learning processes and turnover of labor and the motivation of labor. But basically Intel, that's what Gelsinger was there for. And that, and also to in invest.
Now as I said, a company like Intel would often have to do 10 to 15, maybe even $20 billion in capital expenditure per year just to stay in the lead or to have a chance of being in the lead. And because basically, the technologies, even the technologies for fabrication that they've invested in relatively recently, they may become obsolete just by technological change. And so they have to kind of, you know, Schumpeter called, creative destruction, they have to do it within their own company. And there was a method that they called the tick-tock method at Intel for doing this when by the 2010s, the power of technological change had become greater, particularly through what's called deep ultraviolet, DUV machines, extreme ultraviolet photo lithographic machines that came again, from a Dutch company, ASML, that only a few companies in the world, it's only worth it for a few companies in the world to buy them. And so he became the CEO in February, 2021, no more buybacks, and now they're investing. And also at that time, when he became CEO, there was something out there called the CHIPS Act, which had been actually pushed by the industry and also companies like Apple that used chips in 2020, in Congress. And this was when Trump was still president and now under Biden, was now going to be, became into 2022, the CHIPS and Science Act, okay? And it had $52 billion of subsidies for companies, particularly investing in semiconductor fabrication. So Gelsinger had his eye on that money, but it was only gonna be, in the end they got, as of now, 7.9 billion. But they were planning on spending a hundred billion by 2030 on catching up with TSMC, which is Samsung's other main competitor. TSMC is the one that's really at the front. And that's what Gelsinger was trying to do, at least from 2021 on.
December 1st, 2024, however, the board told him, "We can fire you or you can resign." And that's actually, Matt and I are writing a paper now, and I can give a kind of preview of this, we're trying to finish it fairly soon, but trying to figure out why did they fire Pat Gelsinger? 'Cause I still think he was the right person for the job if what they want to do is have a chance of being competitive at this high end of semiconductor fabrication. And if Intel doesn't do it, then at least from a US point of view, then there's no company that's gonna be able to do it that's US based. Now, TSMC and Samsung are now building plants in the United States, TSMC in Arizona, Samsung, in Texas, to produce higher end chips, particularly TSMC. But you know, the strategic decisions about what to produce and for whom is really being made in Taiwan. And then the other dimension of this, which I'm sure most people are aware of, is that it's actually TSMC that has become a big flashpoint in China's political attitudes toward Taiwan because this is really, that technological capabilities and spinoffs from that in Taiwan have given Taiwan capabilities that China actually doesn't have right now. And in the summer of 2020, TSMC, under pressure from the United States, cut off Huawei, which was the biggest competitor along with Samsung and smartphones, and actually had surpassed Samsung and Apple and smartphones and devices. They cut off Huawei from chips, destroyed the high end of their smartphone market. They've since come found a work around this, but they still can't get a hold of these EUV machines from ASML. But they've managed to produce chips with the older technology that will do. And Apple was the big beneficiary of that because they now had all the capacity of TSMC to their self.
The other thing I'll point out, which has not to do with Intel, but with Apple. Apple outsourcing its semiconductor fabrication for Samsung, TSMC, that's where the leadership went abroad. And Apple is an even bigger repurchaser of their stock than Intel. Apple only started repurchasing their stock on a large scale in 2013. But in 12 years since then, they've done, I think, it's 726 billion, like about 60 billion a year. Now to have a state-of-the-art fab, they could have been spending 20 billion a year at most. You know, so that could have been created in the United States with a bit of industrial policy, et cetera. And also we're now writing another, it's on a current grant we have from INET, Matt and I are writing about why the United States doesn't have a major EV battery producer. And partly, companies like Apple that use a lot of rechargeable batteries, they've outsourced all that stuff to China. And so, you know, that's a bigger story of US loss of competitors in critical technologies. And I'm not in DC I'm not a policy wonk, but there's people there always whining about China in doing this and that stealing intellectual property, all these companies steal intellectual property, they get workers, you know, employees at high levels from other places. You know, that's not the problem. You know, China is in a retain and reinvest mode in terms of their technology companies and it includes, you know, in electric vehicles, and United States is not. And so I attribute the loss of this competitiveness, ultimately, to the shareholder value ideology. Not because there's not enough money, because the most profitable companies are doing most of these buybacks, but it's because of who they put to direct the company and how they then lose connection with the labor force, in this case, a high end labor force that has to really come to work every day and really be highly motivated to get these technologies to work. And by the way, other work I've done with INET on a book coming on African American employment, we look at the lack of African Americans in tech companies. But what we found looking at Equal Employment Opportunity Commission data is that company like Intel will have, and Apple, about 45% of their employees being Asian Americans, many of them having been educated in Asia and now actually many of them under the relations that United States has with China, particularly Chinese, maybe going back to China, finding better opportunities there.
Okay, so there's all kinds of issues here with anybody who's talking about national security that they have to deal with. Nevermind all kinds of issues that have to do with income distribution, have to do with lack of education to upgrade the American labor force and a whole bunch of things that are much more complicated, you know, you have to look beyond one company. But if you're looking at a company that through financialization has just shot itself in the foot and then left us holding the bag, basically, if you're in the United States, Intel is is one of them. And one point we made in the articles we wrote for INET in 2021 was that nine companies, which included Intel and Apple and Broadcom and IBM and others, that were lobbying, so we took nine companies, big companies that were lobbying for the CHIPS Act, which is $52 billion of subsidies. The previous decade they had done something like $887 billion in buybacks. So I was saying, you know, I'm not against the government supporting the industry, but why support the industry when they've done, whatever it is, 17 times that amount of the CHIPS Act subsidies in buybacks. And so there was a kind of a weak provision in, ultimately, I'm not sure it was even in the act itself, but in its implementation by the Department of Commerce, that companies that got subsidies are not supposed to be doing buybacks. And in fact, under Gelsinger, they stopped the buybacks.
Now, so that's the problem. Now, I guess one thing I should talk about is we're writing about now, is why did they get rid of Gelsinger? Now it turned out that there was this big internet boom in tech, Intel went up to about 132,000 employees worldwide. They're now at about a hundred, they're 2024 report just came out, they're about 108-109,000, so they're way down from that. They expanded because there was a huge increase in demand for chips. They would've expanded anyway, but last summer, they announced they were cutting 15,000 employees. So they already cut some employees from that peak the year before, and then 15,000. In the third quarter of 2024, they showed a $19 billion loss, which is huge, in one quarter, 19 billion. And so, Gelsinger, if you're looking at this, he looks terrible, okay? Now in that annual report, they used the term, which fits right in this kind of alliterations that I used, resize, retain and reinvest, downsize, dominate-distribute, downsize-distribute. I needed one more to go with the R and R. They actually use it, it's called resize and refocus. So they said, we're going to cut our labor force and refocus on, you know, being the leader in chip fabrication. This is gonna cost a lot of money. Okay, now they didn't just lay off, they laid off a lot of people, 15,000 people. But that $19 billion loss was because of that, okay? So we calculated, there's restructuring charges that a company takes when it lays off people. And often that is the charges that have to do a lot of it, most of it is what you pay people for severance. So the engineers, they just don't say, "Hey, you're fired," because no one will come work for, there's lots of other companies they could work for. So basically, that comes out to about 500,000 per employee laid off. The other thing, a part of that loss is there's about 3 billion, which is a write off of equipment. Now it turns out it's not write off of real old equipment, it's a write off of relatively new equipment because as I said, the technology's changing very rapidly. So this was what we call seven nanometer equipment, and now they're going to something called 18A, which A is for angstrom and is much kind of smaller, you know, in terms of micro technology. And so they had to write off that equipment. Then, so they got close to $12 billion loss before taxes from this need to restructure the labor force and then to write off equipment. Now they replacing with the latest equipment, but relatively new equipment. Then we looked at their tax liability, and you would think that with the $12 billion loss, they wouldn't have to pay any taxes. But in fact they have a $7 billion tax liability, more than $7 billion. Why is that? Because they got tax credits for that new machinery that were predicated on them actually using the machinery. So the point is that almost all that 19 billion, maybe less a billion or 2 billion, is the cost of actually this problem of what they call right-size and refocus.
Now, I wouldn't give them the credit of right-size and refocus if they hadn't done one other thing. And that is they still were doing no buybacks, but when they did that and they said we're laying off 15,000 employees, they cut the dividend. And so in the fourth quarter of 2024, they paid no dividends. So there is no money going out to shareholders now.
Okay, now maybe the last thing I'll say, 'cause this is a little speculative, but why did they fire Gelsinger? It was a surprise on December 1st, as far as I can see, 2024, that he was being fired. A week before, you wouldn't think that he was being fired. I think the reason is, or at least part of the reason, and maybe a big part is the week before, you know, Joe Biden was still president. The CHIPS Act was a big thing for Biden. They were trying to give out all the money before Trump came into power, the 52 billion. They've been slow giving it out. Gelsinger had been a big cheerleader of the CHIPS Act and he had actually, Biden's first State of the Union address in 2022, Gelsinger was there in the audience and Biden said, "Here's Pat Gelsinger, here's what he's doing, we're trying to help him do this." I don't know what his political affiliations were or had been, I would say just from his, you know, whatever it is, he's probably Republican. But he became very, very much kind of a spokesperson and a cheerleader for the Biden industrial policy. So my suspicion is, and maybe someone who knows about this can give us some evidence, I have a little bit of evidence, is that December 1st, the board decided, hey, you know, you know, Trump's now coming into power. He's not gonna like Gelsinger, so let's get rid of him. And you know, we've seen this in other ways. Like Tim Cook, you know, disgustingly sitting behind Trump at the inauguration. Alphabet and Amazon giving a million dollars to the inauguration, et cetera, et cetera. So this may have been the way that Intel was doing it, get rid of the CEO who is now seen as being part of the Biden legacy.