The rise and fall of Intel

In 2007, Steve Jobs called Intel CEO with a proposal that would change the future of computing, well, almost forever. Apple needed a new chip for their upcoming device. It was going to be called the iPhone. The CEO of Intel at the time, a man named Paul Otellini, read the numbers. He said he didn't like the future of the device, uh, or the chips they would have to make for it. And because of that, he said no. And in retrospect, he regrets doing that for obvious reasons. And he said at the time, actually, his gut told him to make the deal and say yes. But this time we followed the numbers and it was a mistake. But it turns out Paul's forecast about the numbers for the chips needed for the iPhone were absolutely totally wrong. Just wrong in the wrong way. Uh, they were 100 times greater than anybody could possibly have imagined. Apple went with a type of chip called an ARM, which is different than the way Intel makes chips. Uh, but the numbers tell the story. In the end, because of Apple and all the other smartphone makers going with ARM, nearly 2 billion of those chips have been sent since that fateful call in 2007. But in that same period of time, only about a quarter billion, 250 million PCs have shipped during the same time, dwarfing the market of PCs. But here's the thing, this one mistake wasn't the thing that's caused Intel to basically be a shell of its former self today. And in fact, it's not even close to the biggest mistake they made. Back 25 years ago in the year 2000, Intel was worth over $500 billion. It had 90% market share of the global PC market. Fast forward 25 years later, in November 2024, they were kicked off the S&P 500 by their former laughingstock, Nvidia. Who was once king is now a peasant. So what could take down a company that was this dominant for this long, that had so much market share and produced so much money? Well, was it leadership problems? Was it market changes? Was it technology shifting? Or was it something else entirely? This video will talk about it. This video is the rise and fall of Intel. By the way, I made this video in partnership with my company, Near. More on them later. As ubiquitous as they are today, computer chips weren't always everywhere. In fact, they had been invented at once back in the 1960s. In 1968, the man who invented them, a guy named Robert Noyce, left with a guy named, uh, Gordon Moore, uh, and they left a groundbreaking company in Silicon Valley called Fairchild Semiconductors to go start their own business. And what was kind of a crazy thing back then, they raised $2.5 million without a business plan, just an idea, from an early venture capitalist, a guy named Arthur Rock. And basically he was betting on them being a thing someday, but, uh, they didn't have an idea yet that would be that great. Employee number 3 was a PhD named Andy Grove, who would go on to be critical for this entire story. He was a Hungarian refugee. He was partially deaf from a bout with scarlet fever when he was a kid, slight in stature, paranoid, restless, and full of energy and ambition. Intel actually started making memory chips called DRAM. By 1972, their 1103 model was selling really well worldwide, dominating the market. In 1971, one of their engineers took the things they had learned from building semiconductors during that period of time, making RAM, and built the very first integrated CPU, central processing unit, the brain that would go on to run every device and every piece of electronics that we have. And they invented it at the time. They didn't even really know what they had. 90% of their business was spent making and selling and profiting from memory. This early model was the 4004, and it kind of sat on the shelf for a bit. Nobody really cared. They were making money from the memory business. And then Japan happened. In the early 1980s, the Japanese government had decided that they wanted to be in the electronics business. That was how they were gonna transform and globalize basically their economy. So they put a lot of money, backing and effort into making Japan electronics powerhouse. The competition was fierce. In 1974, Intel had 83% of the DRAM, the memory market. Just 10 years later, that number would be 1.3%. That's how fierce the competition got between the US and Japan. In 1985, Andy Grove and Moore are sitting in Moore's office. They're having a discussion. Basically, Grove asks Moore, Well, if the board brought in a new CEO into our dire situation, what would the new CEO do? Moore responded that the new CEO would get out of the memory business. It was a commodity. It was one where every vendor was producing something replaceable by something else, by somebody else. So it was just a race to the bottom. Everybody was competing on price. That's where Moore and Grove agreed, like it was time to move on and get out of the memories business. At the time, one manager compared this to like Ford getting out of the vehicle business. That's how big getting out of the memory business was for Intel. Grove was curious courageous. And, uh, he was the leader at the time. He killed the whole thing, closed down plants, laid people off, made the courageous, difficult decision in face of what the market was saying. But stuck on the shelf inside of Intel was something that would change the company forever. That microprocessor I talked about before, well, it turns out the PC was coming because of Apple and other folks who were revolutionizing the entire marketplace. IBM was getting into the market, and at the time, people forget how dominant IBM was in terms of selling computers and mainframes to businesses across the United States and the world. Like, you didn't go wrong buying IBM. And suddenly here was IBM getting into the PC market. And IBM had chosen Intel as their processor supplier back in 1981. Basically, IBM thought they were buying something that was like a commodity. In reality, they were setting Intel up to be one of the great economic success stories. Well, basically a boltar. Basically, the guys who had founded the company were starting to get old. More noise. So Andy Grove became CEO around that time in the late 1980s. He wrote a great book, you should check it out, called Only the Paranoid Survive, basically about when you're in fast-moving industries like electronics or technology, you have to expect that you're going to be disrupted. And here Intel had faced that moment and made the right call. In the '80s, when IBM had chosen Intel to provide the processor for their PC, which by the way, they didn't take that seriously. IBM was a mainframe company. And here they were building a PC and they were rushing to market. So they went to Intel for the CPU. Along the way, they went to a little company at the time called Microsoft because they needed software that would run on top of these PCs. And so what happened was eventually the entire industry standardized on what would become called the Wintel monopoly. That was Intel's basically chips and Windows running on top of it. And the reason it got so valuable, well, when everybody used it, that meant that everybody knew how to use it. And so companies standardized on it. And that meant that it got even worse because everybody was using it. That meant that all the apps will be built for the platform and all the developers would know how to use it. And suddenly you look at it and if you're a buyer of PCs at that point, you have one option and one option only. You buy Windows with Intel on, you know, the hardware side. And both Microsoft and Intel at this point, they have the same ambition. How do we keep making better chips and better software so people keep upgrading and keep paying us more money? And so that's what happened. It worked so well that by 1992, Intel became the world's largest semiconductor company by revenue. And in 1993, '93, they launched something that seemed pretty miraculous at the time. This is the first year that I was in college. They launched a thing called the Pentium model of processors. It was such a big launch and such a big business that they spent $500 million promoting it. This duopoly got so dominant, by the late 1990s, Intel had over 90% of the processor market. It was a license to print money. By the year 2000, at the peak of the dot-com boom, Intel's market capitalization, the value of it by the market, was over $500 billion, half a trillion dollars, real money back then. And to put that in perspective, the profits and revenue made by Intel and Microsoft together, if you add those two together, it was more than everybody else made in the PC industry combined. That's all the software makers, the hardware makers, the monitors, the everything else that was going in there had been commoditized and was a bloodbath except for the businesses of Microsoft and Intel. They had a license to print money. Intel at the time was what was called totally vertically integrated. They did design, production, distribution, sales didn't rely upon partners. They owned it entirely in-house, which meant they could control every aspect of production. CEO Andy Grove, who'd made the fateful decision with the memory, he retired, uh, in 1998 after being named Time's Man of the Year back in 1997. And to show how different the world was back then, this is when people still read Time magazine. They didn't go on the internet for it. Got it in paper with pictures. It's pretty amazing. Oh, and then you know what most people do with those magazines, especially People magazine? You know where they ended up mostly? Yeah, next to your toilet in the bathroom. I remember those days, crazy. And at that moment was the time when Intel seemed like it was unstoppable, when nothing could end for them. And that's when it all started to fall apart. Like many empires, the rot of Intel, like the Roman Empire, didn't come from the outside. It wasn't the Mongol hordes that took it down. It was coming from inside the house. But before we get into that, let me give you a short pitch about a company that I co-founded called Near. It helps American businesses hire great talent in Latin America. If you're interested in hiring from places like Argentina, Colombia, Brazil, it's a great service for you. Besides the obvious savings where you can do 60 or 70% off US salaries, Latin America is the best experience of overseas hiring I've ever found. It's a great culture fit. There's no language barrier and they're in the same time zone. Which matters way more than you think. They're just part of the team in the day to day and not second-class talent. I've hired dozens of people through them and the talent is just awesome. Huge businesses have been outsourcing entire departments to Latin America for decades, and Near lets you do the same no matter your size. They just placed an 18-person team of sales reps at a big company here in Texas, and that group of people added $20 million in annual recurring revenue quickly. Or they're happy to just fill a single seat if that's all you need. I've hired dozens of people through the company, accountants, sales reps, everybody in between, engineers, and I encourage you to check them out. There's a link in the description. Now back to our story. To understand what happened next, you gotta understand how chips work. And there are typically two types of chips out there. One is called RISC, which is a reduced instruction set type of chip. And the next one is CISC, or a complex instruction set. Intel at the time was huge about CISC. Which meant that basically the chips were more complicated, they were bigger, and they were more power hungry, which meant it was great for people who were developing applications for them. They got to market faster 'cause more of the smarts were in the chip. But the downside was they used a lot of power. That's perfect for desktops, but if you're gonna be out mobile, when battery technology back then was super like not effective and efficient, and you look at Gen X, my generation, you see us with a mobile device, we're always plugging 'em in. Yeah, it's 'cause when we grew up in the '90s, you couldn't trust a battery at all. And that was the world that was coming, a mobile one. When power was really going to matter. And the funny thing is, back in the '90s, Intel was actually perfectly positioned for the mobile world to come. The types of chips that are these RISC chips, the reduced instruction set, the simplified ones, they're actually called ARM typically. And they owned a company that actually produced ARM chips called StrongARM. And well, they were about to screw it up. In 2006, the year before they would get that fateful call from Steve Jobs, they sold off their entire ARM business, the one that would be perfect for mobile, to a competitor for $600 million. The reason? They didn't see the point. Tech is an industry with some fascinating dynamics. One of them is called low marginal cost. It's the idea that typically you do a design once and you build a chip out of it, and once that chip works, it costs you very little relative to the selling price to produce more chips. So you might spend $800 or $900 for an Intel chip, but it might cost you a couple bucks to make the thing. The expense was all from the beginning when you spent billions of dollars to write the thing. Same thing goes for software, but in chips it's especially pronounced. And what this means in practice is he who gets an early lead or she who gets an early lead tends to often win. And that's because winning and being ahead compounds. More people buy your stuff, you get more money to reinvest in it, your technology and feature lead, like, it advances. And here, just at the moment when Reduced Instruction Kit, and arms were gonna be important, well, Intel was starting, you know, 25 yards behind everybody else. So because of this decision in 2006, when in 2007 they got a call to produce chips for the new iPhone, well, they didn't have anything to base it off of. They'd just given up entirely. They were gonna have to start from way behind everybody else. Over the next decade, Intel would invest billions to try to catch up in mobile chips. And by 2016, they had wasted all that money. They gave up entirely on their series of chips called the Atom. By 2019, Intel would surrender entirely. They would sell their entire chip business to Apple for just a cool billion dollars. They'd lost. So why did they make this mistake? Well, they were missing the forest for the trees. PC chips, when you sell them, tend to have 60% margins. That kind of margin is pretty addictive. The problem is mobile chips only have a 20% margin. Totally different business and tough to make work. When you're somebody coming in like Intel was. And this is where they ran into a thing called the innovator's dilemma. In pursuing the future, they were potentially going to cannibalize their existing business. And at this moment, incumbents have to make the decision that they're going to cannibalize their own business or somebody else is going to do it for them. In this case, Intel made the wrong choice. But despite this lead, things weren't over for Intel. They still had another chance. Remember, unlike other manufacturers who had been outsourcing their chip production overseas and to other folks and weren't vertically integrated and didn't control it end to end. Intel had that advantage and they still had the best fabs in the world to be able to produce these chips. For decades under Noyce and then Grove, Intel's manufacturing had been legendary. They kept pushing the limits on what was possible with chips, making them smaller and smaller, reducing the amount of distance and time it takes for electrons to travel around inside of those things. For a while, it was down to a science, like a— well, basically like a metronome. Every 2 years, another improvement like clockwork. They were at 14 nanometers. That came a couple of years late. Then 10 nanometers came a couple of years late. And to be fair to them, this all gets really hard once you're down to nanometers, which are entirely, hugely small parts of, of meters. Like weird stuff happens with physics. But even still, competitors were starting to navigate these micro scales better than Intel would. Their lead in manufacturing started to show in their quality production. Nearly half of every wafer and chip produced was starting to show flaws. It's a long distance from where they'd been in the decades before. But instead of getting better at getting smaller, Intel responded by compensating with other ways. Instead of basically getting shorter distances to travel, they would throw in more tricks and hoops. And so the entire Intel stuff was stuck around this 40-nanometer stuff. While the competitors were starting to go smaller. Intel's biggest competitor was a Taiwanese company, and if you look today why people are freaked out about something happening to Taiwan, a lot of it is because so much of the chip building these days comes out of one company, Taiwan Semiconductor Manufacturing Company, or TSMC. And while Intel got stuck at 14 nanometers, uh, basically TSMC got it down to 7 nanometers. They were pushing way ahead. This created an opening for Intel's biggest competitor. And back in the '90s, a company called AMD had sued that allowed them to compete and offer processors that were offering the same APIs, the same interfaces as Intel. This company had been AMD, but they hadn't been able to compete for decades. In fact, they'd gone nearly bankrupt. And to save themselves, they'd partnered with TSMC out of Taiwan, with AMD doing the designs and Intel basically standing still. By 2017, AMD was basically pushing into the heartland for Intel. 25% of PCs, 33% of servers. Basically, they were competing and dominating on Intel's home turf. And the crazy thing, they had for a long time been at basically 0%. Suddenly Intel's design and manufacturing was gone from being in the lead to way behind. But things were about to get worse again. And that thing was the rise of artificial intelligence. And the way AI works is totally unsuited for the way Intel classically made processors. Intel had optimized for doing one thing at a time or a few things at a time really, really fast. AI requires exactly the opposite. You do a bunch of things all at once and reasonably fast. Back in 2006, a technology and chip company called NVIDIA had launched a thing called CUDA, which was an evolution of their attempts at a thing called a graphical processing unit, basically a board that would fit into your computer to help it do things with graphics really fast, like play games or do, you know, CAD and drawing and stuff like that. And so in the early 2010s, as basically Silicon Valley companies and folks in academia started to work on AI using the types of things that have become pretty prevalent today, they had two options. One, We go look at CPUs, which are not going to be very good at what we're trying to do, or go choose things like the chips from NVIDIA. And during that period of time, because they were so focused on just playing the game they'd always played, Intel was spending nearly 5 times as much on R&D as NVIDIA was at the time. And the reality was, sadly, none of that was spent on chips that would be good for an AI world. And in proof of the pattern that in a post-Andy Grove era, they constantly missed wave after wave, In 2017, Intel had been offered to invest in OpenAI and get 15% of the company for a relatively small amount of money. They passed. They would have gotten 30% of OpenAI for $1 billion and agreeing to sell them some chips. Like, nuts. In November 2022, ChatGPT launches and demand goes parabolic for the types of chips that will be used to run AI models like theirs. ChatGPT. ChatGPT. ChatGPT. And the chips they'd chosen were NVIDIA. In 2022, NVIDIA had done $27 billion in revenue. In 2023, they did $60 billion. A company of that size over doubling in revenue in just one year. Fast forward to 2025 and into 2026, NVIDIA's market cap has occasionally passed $3 trillion, often making it the most valuable company in the world at any given time. Intel once again was behind the curve and mis-executed. They had bought a company called Habana Labs with their own idea of producing AI chips. Uh, and those have basically floundered. In fact, until I researched this video, I didn't even know they existed. That's how little people in the AI space talk about Intel. And in the most evidence of corporate speak and what's wrong at Intel, uh, CEO Pat Gelsinger went out and said basically— well, I'll read it here because it's horrible— our revenues have not grown as expected and we've yet to fully benefit from powerful trends like AI. Yeah, we get the message. Intel continued their tragedy of what's going on with their C-suite and the CEO suite specifically. They have hired a guy named Pat Swan to come in. They brought back in Gelsinger. And again, like a company like Intel is so big and takes so long to bring things to product, to market. Like things do not shift on a dime for a company like this. And when you see like basically the company changing leadership every few years or even every year, like It's a sign things are going horribly wrong and aren't gonna get better anytime soon. Gelsinger, well, if you're gonna pick somebody, had actually been a pretty good choice to come back and try to fix Intel. He had started there in 1979 at age 18, and then he left to go work at a hugely successful software company called VMware, and now he was back to try to save the day. His vision was called IDM 2.0, and the idea was to go catch TSMC, become a vertically integrated and beautiful foundry, work on designs not only from Intel and from other folks. Remember how I told you the measurement was this kind of nanometer thing about how much progress Intel would be making, or any chip manufacturer for that matter? Well, in the past 5 years, Intel had made exactly one jump to a smaller measurement, and basically he was promising that they would do 4 in the next 5 years. They started to throw on the money— $20 billion for fabs in Arizona, $20 billion for ones in Ohio, uh, basically trying to throw money at the problem of being way behind. Gelsinger went and lobbied for $8.5 billion from the CHIPS Act that passed during the Biden administration. The idea was to help supplement their war chest to try to catch up. But behind all this, there was something else going on. Between 2011 and 2021, Intel spent nearly $70 billion going out in the open market and buying back their stock, which was enough money on its own to build 3 or maybe even 4 fabs. But instead of trying to keep up and fight with Nvidia and the other things going on in the market and reinvest like they needed to, uh, basically they were giving money back to shareholders. This turnaround was not going to be cheap, but it turned out to be more expensive than anybody thought. Uh, they lost $7 billion from the Foundry Unit in 2023 alone. Intel revenue went from $71 billion in 2021 to $54 billion in 2023, nearly a 30% decrease. But here's the problem. They were hiring more people. They were producing fewer dollars per head year after year. Q3 2024 brought a nearly $17 billion loss for the company. Uh, and a few weeks later, they announced that they were going to lay off 17,500 people from the company. Intel stock, despite all the buybacks, was down 60% since Gelsinger arrived. It was not going well. On December 1st, 2024, Gelsinger was asked to retire by the board, which is code for fired. They just give that to people who are in their 60s. It's the way it sometimes works. And in the kind of saddest moment, the once kind of picking envoy of NVIDIA replaced Intel on the S&P 500. As kind of one of the stalwarts of the American economy. And in a funny twist of fate, Mr. Tan had been on the Intel board and had resigned in August 2024. The reason? He had been frustrated with Gelsinger's strategy and didn't want to be in the room if they were going to keep doing it. This is one thing to know about boards as they get bigger and there's lots of shareholders around the room. They transform from kind of what you expect in an entrepreneurial environment where people are tight and super aligned to making stuff to when they get to be companies this size. Those boards become like basically political war rooms and you can kind of see the stuff that happens. The signs are there of all that happening. Intel, way overstaffed, announced another round of layoffs, 21,000 more, bringing the total staff down to 75,000. And that just happened earlier in 2025. It started to slow with all this growth of these plants, Germany, Poland, Ohio. They slowed down the bleeding of the cash going out. And in an ignominious kind of move, the Trump administration has taken all that money that they gave to Intel as part of the CHIPS Act. They converted that into equity. Now, you and me, the taxpayers, oh well, we now own 10% of Intel. And NVIDIA, the company that made its money from the first graphics boom and now the AI boom, also invested in Intel, having put nearly $6 billion in the company. Between them and the government, they have some new shareholders. Under Mr. Tan, the stock is up 79% from what it was. But it's still a fraction of what it was just 25 years before. After Andy Grove, it's pretty clear Intel lost its way. And structurally, I think I know why. The reason is, is Grove thought like an owner. He had a long-term mindset. He'd been there early on with the founders. He had a big ownership stake. After that, you had a bunch of people that were just hired guns. They were much more interested in what happened in any given quarter than what was going to happen in the future. You can see that in the way they missed the mobile revolution, the way they missed the AI revolution. They would have had to sacrifice the next couple quarters or years in order to get to the pot of gold at the end of the rainbow later on. And those structural problems set up with people with the wrong incentives caused the wrong decisions. And today, 25 years later, what was once a great company isn't so much that anymore. And an interesting twist, preparing for this video, I looked at how many Intel chips my family had bought before the past couple years and how many we bought recently. And now since we're a Macintosh family, running iPhones. We haven't bought anything from Intel, well, in a long time. A lot of our customers are just the same way. All right, this is a bit of a longer one, and I tried to make it not very technical. Let me know what you think in the comments below, uh, and, uh, we'll see you next time. Bye!