The two major chip companies that are leading the AI wave, NVIDIA and AMD have doubled their share prices from October last year to March this year and have become two monster stocks. The two big bosses behind them, Huang and Su coincidentally are both Chinese Didn’t we talk about NVIDIA last year? Today we are going to talk about another monster stock, AMD There are many discussions focused on AMD and NVIDIA graphics cards many people think that AMD only makes game graphics cards Actually, AMD is not only one of the most important in the independent graphics card market it is also one of the most important in CPU market In its more than half a century of business ups and downs the one opponent it can never avoid is not Nvidia but Intel. AMD has risen and fallen in the competition with Intel. For the last eight years, they staged an epic comeback The stock price increased more than 100 times, and the market value exceeded $300 billion Intel was instantly left behind, and its market value was even higher than that of Moutai. So how did AMD make a comeback How did the legendary Lisa Su make it rise from the ashes and ride on the strong wind of AI to reach its peak. Let's get into it The story begins with what is known as the greatest invention of the 20th century. In 1947, three scientists at Bell Labs successfully created the world's first transistor and won the Nobel Prize in Physics for it. There is a person named William Shockley a master-level figure in scientific research but he had a problem His management skills are very poor there were problems like stubbornness, racial discrimination, paranoia and etc a lot of problems This master had eight disciples who were also the top scientists in the industry. But because they couldn't bear it anymore they all left the master together in 1957 and set up their own company called Fairchild Semiconductor. Fairchild Semiconductor Many people may not have heard of this company before but it's no exaggeration to say that it is the cradle of talents in the semiconductor field and can even be said to be the origin of Silicon Valley This talents from this company kept flowing out In ten years, the eight founders all left after one another The last to leave are two brothers are Noyce and Moore, they left in 1968 to establish the famous Intel. A year later, Sanders, the sales director of Fairchild Semiconductor, himself left and established Advanced Micro Devices, AMD the flight of executives and scientists basically announced the elimination of Fairchild Semiconductor. However, it also officially started the commercial war between AMD and Intel that continues to this day. Actually, in the beginning, AMD and Intel are not at the same level at all. Intel The company was owned by the disciple of the great master Shockley they have the talents and the funds When they set up the company, people just stuffed money into their hands and they easily get two and a half million dollars and launched IPO in just two years and it continued to launch chips that led the industry standard. However, AMD did not have such a perfect start. In the 1970s, it was a melee for entrepreneurship in the semiconductor industry. There are so many small fished and shrimps like AMD that no one is willing to invest in it. So do you know who is the earliest individual investor who invested in Sanders? It was actually the head of Intel at the time, Noyce. What? So you see, AMD and Intel were actually very closely entangled from the very beginning. Later, Sanders was quite incredible. Three years after their establishment in 1972, AMD also went public. At that time, there was a very interesting phenomenon in the US For example if I am a manufacturer and want to purchase chips if I ask to buy chips from Intel, I will force Intel to give this drawing to another manufacturer to build it with you the so-called second supplier This will prevent from being stuck by one manufacturer Imagine at that time AMD and Intel actually had a good relationship so they formed a small group. AMD would often serve as Intel's second supplier. AMD's move is like I don't know if you have heard of it there is a very important strategy when riding called Drafting If you follow the leader, wind resistance will be greatly reduced and you will save a lot of energy And for the leader, having someone following behind, actually save more effort So they formed a team, a small fleet. Their most successful deal was to win over IBM, which was just about to enter the personal computer market in the 1980s. They teamed up and took off Intel's X86 architecture formed the CPU industry standards at that time. They also ushered in a glorious period in the 1990s. People who worked on computers at the time were basically familiar this sound Intel's market value in year 2000 $275 billions and became the sixth largest company by market value Meanwhile AMD its stock price has increased 15 times in five years. However, as the PC market gradually become saturated at that time, the two of them also had some licensing lawsuits. AMD and Intel also had a rift. AMD felt that it was not okay to be tied to Intel forever They have to start riding against the wind. In 1995, AMD began to significantly increase R&D expenses and spent $900 million to acquire NexGen, a semiconductor design company. And guess what, the COO of NexGen happened to be someone who was just poached from Intel The person that lead the design of the Pentium processor was called Dam. In other words, the father of the Pentium processor was poached by AMD by accident. Under Dam's leadership, in 1996 AMD launched the self-developed K architecture and finally ushered in its highlight moment. In June 1999, the Athlon processor equipped with the K7 architecture was launched and then quickly upgraded. Less than ten months later Intel released the world's first Athlon processor with a clock frequency exceeding 1GHz. it was directly going against Intel’s Pentium III processor at that time I'm not going into the parameters. To put it simply this chip and processor are far ahead of Intel’s chips of the same price so it is more cost effective It brought a profit of $1 billion to AMD in just one year While the year before, their net loss was at $89 million. In the past two years, they lost almost $100 million each year. In 2000, when Athlon came out, it suddenly became a profit of $980 million Athlon was indeed speedy In the next few years, AMD counterattacked all the way. In 2003, it launched the first 64-bit X86 processor equipped with K8 architecture With a series of breakthroughs, AMD's share in the CPU market increased to 48.4% in 2006 It's almost equal to Intel the desktop market even surpassed Intel. At this time, AMD is gaining momentum so it began to think about its next big thing and was ready to get married Preparing to make a big move, it is eyeing the field of graphics cards. We know that the two most important chips in computer are CPU and GPU. Central processing unit and graphics processing unit, which is graphics cards. What was the situation like on the market in 2000 ? Basically, on the CPU side only Intel and AMD were left. As for GPU, we talked about it in the previous episode regarding NVIDIA After a big fight in the 90s only Nvidia and ATI were left. What did AMD think at the time? If I could integrate CPU and GPU, if I could span the fields of these two major chips, then there'll be synergy between technology and customers Maybe it could even beat Intel. So at that time, it wanted to to choose between NVIDIA and ATI pick one and acquire one AMD executives began to think about it Nvidia's market value at that time was about $7.5 billion Meanwhile AMD's value was about $8.6 billion If I want to acquire Nvidia I'd need to pay a premium then Nvdia might have acquire AMD and get swallowed up by Nvidia, then it's all over Oh No, ATI value was about $5 billion even though it is very big. But still a distance from AMD, so this acquisition still possible So at that time, AMD just gritted its teeth emptied $1.8 billion in cash borrowed another $2.5 billion in foreign debt, and spent $1.1 billion in stock almost gave out half of their live to acquire ATI. Since then, AMD has become the only giant with a considerable market share in the two major markets of CPU and GPU. At that time, AMD was so powerful in terms of CPU, they share the market almost equally with Intel On GPU they swallowed up ATI in one breath, so they should be able to compete with Intel right/. But in fact, after this, AMD ushered in its darkest decade. Let me tell you this After AMD acquired ATI, in the next two years when they are doing the accounting they recorded their Goodwill $2.6 billion less What does this mean? It means that all the money they gathered total of $5 billion to acquire ATI they found out a year later that it is only worth about $2 billion Actually, AMD’s idea or strategy of entering the GPU field is very reasonable and even very smart. Now we can see that it is exactly because they acquired ATI that they could compete with Nvidia in the GPU field. But the problem is that it was really too expensive at the time and it was not even a little bit expensive. This made AMD's cash flow become particularly tight, and at the same time, Intel felt threatened. It has also begun to exert its strength. If you just look at the market share, you may think that AMD and Intel will share the market equally. But in fact, the gap between the two is quite wide. How can I put it? Because AMD's market share is mainly concentrated in the mid- to low-end Didn't we just mention that it focuses on cost-effectiveness, so its profit margin is not high At that time, Intel's revenue was almost ten times that of AMD and its market value was thirty times that of AMD After all, Intel had deep pockets, When it felt threatened, it started to get nervous, and started to terminate employees and restructure Just 3 days after AMD announced its acquisition of ATI Intel released the Core 2 processor using the new Core microarchitecture. Both performance and energy efficiency ratio were excellent, once again widening the gap with AMD. Moreover, Intel also has a tactic called Tick-Tock Note that it has nothing to do with TikTok. Tick updates the chip process every year and Tock updates the processor architecture every year. To put it bluntly, it is fighting against itself In the next ten years, the process was upgraded to 14 Nano instantly reoccupied the CPU market with its i3 i5 i7 processors. You see at this time, AMD spent a lot of money to buy ATI and found that it was too expensive On the other hand, Intel began to exert its efforts and AMD continuously lose its market share In the CPU field it has to fight with Intel in the GPU field it goes against Nvidia Moreover, it had to design and produce at the same time, so its not able to cope cash flow began to become very stretched and its debt accumulation gradually reached $5 billion Out of desperation in 2008, AMD split off its foundry and sold it which became GlobalFoundries. On one hand, it could get some money and on the other hand, it could focus more on chip design. In the next few years, AMD launched a series of products but market's respond had always been unsatisfactory. Sales performance has continued to shrink since 2010. The market share of CPU has dropped from nearly 50% at its peak to only 20% while the market share of server CPU has dropped from 30% to only 1 % market value fell to less than $2 billion Leave alone competing with Intel they could barely survive. AMD can only keep laying off employees and cutting costs. It even sold its central building for $160 million. Many Wall Street analysts believed that AMD might gradually withdraw from the stage of history. But in this endless dark moment, there was a person who descended from the sky with colourful auspicious clouds She was Lisa Su Zifeng , the new CEO of AMD in 2014. She is also affectionately called Su Ma Lisa Su is definitely a technical elite. She studied electrical engineering at MIT and got a doctorate. Afterwards, she worked in chip-related jobs at companies such as Texas Instruments and IBM. By the way, if you look at interviews of Jensen Huang and Lisa Su you'll find a very interesting phenomenon These two people speak very fast. They might control the speed during official launch but if you watch their private interviews, they are really like machine guns they don’t speak like ordinary CEOs. you'll be the one getting anxious and have to go twice as fast to listen today Let’s get back to AMD. You see, when Lisa Su took over she was facing a very difficult situation At that time, AMD was already in very difficult position She started with a wave of classic cool operations, such as simplifying company processes, maintaining customer relationships, cutting costs, etc. These are all important but meaningless so we won’t talk about it, but it’s not over yet. The most important thing, I think, is she made two strategic changes. She made two strategic changes and became the key to saving AMD. Each of them can be said to be hell-level difficulty. First one is high-performance computing. What is high-performance computing It sounds like a buzzword but the key lies in this high performance. Didn’t we talk about it before? It seems to you that AMD can compete with Intel but in fact, it mainly revolves around the mid-to-low end. Lisa Su think that this is not right the mid-to-low end had no future, moreover the added value and profit margin were too low So she refocused AMD on the development of high-performance chips. I've watched her countless interviews and the one word she kept talking about all day long was High performance computing. After two years of strategic adjustments, In 2016, AMD sounded the first clarion call to counterattack They launched Ryzen chip whch is equipped with the new Zen architecture. This Zen architecture is a brand-new framework that is completely different from Intel's process technology. Simply put, it uses a lot of small chips called Chiplets to be integrated into large chips. This allows for greater flexibility and higher yields. So if you look at AMD’s recent CPUs and GPUs they are all based on the underlying architecture of Chiplet. Originally, AMD’s goal for the new chips was to have a performance that would exceed 40% of the previous ones. Many people in the market thought it was a fantasy. It is difficult to achieve this in one step. No one expected that the first generation Ryzen was actually 52% faster than Intel’s eight-core processor i7 6900K. It was not only faster but also less than half the price. This was only the first generation at that time, it's not to say it can defeat Intel in terms of tech but the optimisation of the Zen architecture in the next few years can be said to make AMD unstoppable. In 2019, it launched the Ryzen 3000 series equipped with a seven-nanometer process equipped with Zen 2 architecture. In the gaming field, the PS5, Xbox X and S series including Steam Deck all uses this Zen 2 architecture APU chip. They made a big bet on seven nanometers and started to take off. In 2020, Ryzen 5000 series with Zen 3 architecture was launched In 2022, Ryzen 7000 series with Zen 4 architecture was launched It's not only chips quality just take a look at the quality of AMD's press conference the improvement is very visible With the strong return of the Ryzen series, AMD's market share in desktop CPUs has once again returned to a situation where it is evenly matched with Intel. However, this is far from over. The Zen architecture that AMD finally developed is not just for PCs and for consumers to use. This brings us to Lisa Su's hell-like task which is to expand business lines. Lisa Su knows very well that AMD has actually always been too dependent on CPU, so in 2015 she proposed three key growth areas gaming, data center and immersive devices. Her vision was indeed quite vicious Let’s go back a little bit. Before she became CEO, she had already saved AMD by promoting the gaming business. The life-saving straw in the darkness after AMD acquired ATI was precisely its ability to straddle the fields of CPU and GPU In 2011 they launched APU Accelerated Processing Unit Accelerated Processing Unit You can simply understand it as integrating CPU and GPU on one chip. When Lisa Su came to AMD in 2012 to be in charge of products she took the lead and made all the major game console giants to choose AMD's chips so that the game business can bring $2 to $3 billion in profits to AMD every year. To a certain extent, it has made up for the huge losses caused by the PC side, which has allowed AMD to survive. So the 3 fields Lisa Su mentioned for gaming segment AMD already monopolised the market they just need to develop accordingly Second segment, data center with the strongest growth momentum and the greatest potential. Actually data center is like cloud service provider supercomputers, they have to go to Intel and AMD to buy CPUs and then go to Nvidia and AMD to buy GPUs. In fact, this CPU segment was completely owned by Intel and this is also the lifeblood of Intel. Why do we say that? For companies cloud computing or gigh-performance computing has long been the basis of these technology companies even before AI came out. Mobile applications and game servers may have half of their budget placed on the servers So for these companies they need higher-end chips with better energy efficiency. AMD could not compete with Intel before. So if they want to seize the data center business, the necessary prerequisite is that you have the ability designing these top-notch chips It's just what Lisa Su did high-performance computing. You see, the high-performance Zen architecture is amazing. so the potential for data centre is also amazing But what’s even more amazing is that you have to complete these two points at the same time to make them match up. only then AMD can become what it is today After the launch of the Zen Architecture Lisa Su also attached great importance to data centre business She herself said that the data center business is the growth point with the greatest potential for AMD. It is AMD's Number One Strategic Priority AMD soon launched an EPYC chip for enterprise users, called Xiaolong in Chinese After several generations of upgrades and optimisations, it can now be said that the performance has surpassed Intel's benchmark Xeon series. We know that for enterprise users especially those high-tech data centre business they not only buy a lot of CPU but they use it immensely One indicator they care about is called TCO. Total Cost of Ownership Total Cost of Ownership This is completely different from what people usually talk about in the consumer industry. You may be done with these consumer goods after buying them, but for things like servers, The cost of your initial purchase is just the tip of the iceberg, because after you buy it, it will continue to work uninterrupted, its power consumption, maintenance costs and even cooling costs, etc. are the bulk of the cost underneath the iceberg it is also important for companies to control costs. Actually TCO is not very easy to calculate for enterprises. You can only roughly estimate it. And let me tell you that AMD is really "insidious". It has made a very convenient comparison tool on its own website You can pick any AMD's EPYC chip, pick an Intel Xeon chip, and compare them like the price speed, power consumption, and TCO. You can pick products and parameters at will and compare them There may be restrictions on advertising laws in China but in foreign countries, they can really compare like this. I simply played with it and concluded that chips of the same level AMD chips are cheaper than Intel by about 20% to 50% The maintenance cost is 10% to 30% lower. Those data centre cloud service providers may buy tens of thousands of servers a year and save hundreds of millions? In short, AMD just wants to tell you that you can just pull out any chip same level as Intel AMD will be cheaper and better performance lower power consumption and lower total cost. Some people may worry that the data on AMD’s official website are deliberately beautified their data or deliberately suppressing Intel? I don’t think that is a problem, otherwise Intel would have to sue AMD So although AMD is blatantly trying to trample Intel but Intel did not react at all Of course, this also shows that AMD must have the confidence to dare to compete with Intel’s products like that After all, the customers of the data center business are enterprise users so they cannot change at will like consumers. Look at EPYC's market share has not soared in a leap-like manner but Amazon's AWS, Microsoft's Azure cloud, Google cloud, IBM cloud and etc All these manufacturers have begun to partially use AMD's EPYC chips And I heard that the proportion of EPYC purchases by domestic Internet companies is still quite high. because everyone loves to wtch videos. when you watch 4K videos and the video is a little stuck it means that the company may need to purchase additional servers and data centers. This is not over yet. In order to expand the business, Lisa Su also spent a huge sum of $49 billion to buy the chip giant Xilinx which is no.1 player in the FPGA chip field. FPGA, is called a programmable logic array. Simply put it is another kind of chip different from CPU and GPU It can be used in devices such as electric vehicles, Mars rover communication base stations, etc. Through acquiring Xilinx AMD can significantly expand its business scope. We will not expand on this FPGA business. But what is the point I want to expand on? Let’s take a look at the acquisition part. Let’s see how they acquire it Think about it, this is $49 billion. Although AMD was booming at the time she didn’t have much spare money. It's not Microsoft In fact, she only had about $2 to 3 billion in cash as a buffer. So how did she buy Xilinx worth $49 billion? Let's take a look at today's financial knowledge. AMD uses a stock-for-share acquisition, which means I exchange my shares for your shares. Lisa Su said, otherwise after our acquisition is completed, I won't have any money to give you, All shareholders of Xilinx can get 1.7234 shares of AMD stock. Let’s do a simple conversion. When the acquisition was announced, AMD’s market value was about $100 billion In exchange, Xilinx’s acquisition price was about $35 billion. After the acquisition, it will become a $135 billion merger, so Xilinx shareholders will account for about 26% of the overall shares. This is not over yet. The fun part is that such a big deal we talk the deal today and settle tomorrow. Just like Microsoft's previous acquisition of Activision Blizzard, it took months or even years to complete. In other words, the acquisition value of Xilinx changes in real time with AMD's stock price, so you can see as soon as AMD announces the acquisition of Xilinx, the stock prices of the two companies stayed in same frequency As a result, the acquisition was not officially completed until more than a year later on 14th February 2022 They made it through. At that time, AMD's stock price has increased by nearly 40% compared to when it was first announced. In other words, the acquisition price of Xilinx has automatically risen from the initially negotiated $35 billion to $49 billion. In over a year it skyrocket to $10 billion So you see, this is actually the so-called getting something for nothing. Because AMD's stock has risen sharply in the past few years so their share has become as valuable as Tang Monk's meat. In this way, I won’t have to spend money when I go out to buy things. I can just cut off some of my flesh and exchange it with others. This is the largest acquisition in the history of chips. As for why ordinary companies sometimes choose to exchange shares instead of loaning to perform the acquisition what are the methods involved in the exchange of shares? The water here is actually quite deep. We will talk about it later when we have the opportunity In short, AMD is riding high with the EPYC series based on the Zen architecture and the acquisition of the Xilinx data center business is developing rapidly and the stock price is also soaring. Lisa Su was named the most powerful woman in business by Fortune magazine for two consecutive years in 2020 and 2021. In 2022, she became the chairman and CEO of AMD. Now it’s Intel’s turn to feel uncomfortable. To be honest AMD can grow so quickly and lead the way is actually due to Intel's own mistakes. Since 2015 it has been upgrading year after year like squeezing out toothpaste, which has made many consumers very unhappy. A big reason for this is that Intel due to some technical problems the 10 nanometer process was very difficult It got stuck and stood still at the 14nm stage for several years. In 2019 they were forced to apologise in writing to all its customers saying that it encountered bottlenecks in production and could not manufacture it. Sorry, I'm sorry. Its Tick-Tock strategy Tick-Tock all the way from 90 nanometers to 14 nanometers However, at 14 nanometers, Tick Tock the Tock could no longer go on so Intel had to abandon its Tick-Tock strategy. This is actually why Many people think that AMD made a very wise decision when it sold its foundry business, not because the manufacturing business was not good but because design and manufacturing have very different management models in terms of operation and core technology It’s difficult to manage both of them and let them be at the same frequency. Look at Intel , the entire chip line is delayed because of problems in manufacturing. On the other hand, if you look at AMD, it is very flexible. It has been cooperating with Global Foundries but later felt that it could not keep up with the pace of their progress. There was no pressure to switch to TSMC easily. So let me tell you that for many companies, it is not that the bigger the better. In fact, many professional managers some fund managers advice to most companies is that you can split off some businesses with relatively low synergy efficiency so that they can operate more efficiently. The time has come to 2023, year of AI when AMD almost killed itself when it acquired ATI It has ushered in a good harvest. As the only independent graphics card company on the market, it has ushered in its secondary boost. The series of AI chips they launched last year were snatched away by Microsoft, Meta, OpenAI as soon as they came out. The data center business In the fourth quarter of 2023, it surged 43% from the previous quarter to $2.3 billion, surpassing the gaming business and becoming AMD's first source of revenue. The Ryzen series on the consumer side are all equipped with AI engines. Compared with the previous generation, the NPU has improved performance by 60% and is now on the market in large quantities. AMD can be said to be all in AI I went to their AI summit two days ago. It can be said that all products are optimised for AI. In just eight years, AMD's share price has increased more than a hundred times, from the brink of bankruptcy to its peak, with a market value of $300 billion. It has almost doubled Intel's and even surpassed Toyota and Moutai. Although in absolute terms Intel's current revenue market share is still the largest but AMD's potential and growth momentum for the first time has firmly surpassed Intel. As for the competition with NVIDIA on the GPU side some people may think that NVIDIA is the greatest, it is great, certainly otherwise we wouldn’t have been able to do a special episode last year to talk about it In fact, I think the competition between AMD and NVIDIA is now completely different from the zero-sum game between AMD and Intel. They are competing for customers and markets. Lisa Su mentioned last year that the overall potential market size of the AI field in the next three to five years will reach $150 billion per year Globally, AI production capacity can only supply less than 20% of demand. From a production capacity perspective, it still has a big gap to fill, so it is actually a market where demand far exceeds supply. As long as you can develop powerful enough chips and powerful enough product customers will come queing They don't care if it's Su or Huang or Wang from next door As long as it’s useful, I want it. So for AMD and Nvidia, it’s completely a blue ocean and not a zero-sum game. They are now actually more like racing against themselves and technology. As Lisa Su herself said, AMD does not have to beat Nvidia to achieve good results in the graphics card market. In fact, I think I should add to her that AMD does have to beat Intel to achieve good results in the CPU market. Now it's Intel's turn to be anxious. They changed a new CEO in 2021 and chose to take a completely different path from AMD. They bet all their money on chip manufacturing and dumped $20 billion to build two new foundries they plan to catch up with TSMC's process inferiority by 2027. Intel actually still has many cards to play. The war between AMD and Intel is far from over. You see, we talked about Nvidia before combined with AMD this episode and taking a look at Intel you may find that the chip industry is an industry with fast iteration. Whoever launches a star product can turn around the ship in one go. At first glance, you may think that this is a short-term sprint. It can be accomplished in one or two years Actually this is a mere representation of what consumers can see. AMD’s EPYC ROCm, Intel’s Core processors, including NVIDIA’s CUDA, behind every breakthrough are actually investment that take at least three to five years or even longer It is a big gamble. As a chip company you not only need to quickly push for the research and development of current products and optimize sales but more importantly, how will you go in the next five years? You need to constantly make judgments and decisions and then invest a lot of costs and bet on it For other industries the management and CEO may make some major decisions from time to time Generally, when they bet right they can comfortably lead for a period of time. But in the chip industry, you really can't rest on your laurels No matter how successful your product is it may be leading for 3 to 5 years but a new architecture will beat you down So every 5 years you have to take maybe half of your net worth to make big bets in the next five to ten years Sounds intermidating? Just look at the speed of Jensen Huang and Lisa Su's hair turned gray, and you will know how much brainpower it takes and how much pressure they must bear. The technology's rolling gears of progress can be said to be vividly amplified in the chip field Saying it is a winner is only temporary Intel spending money to build factories, AMD is all in AI chips, and its acquisition of Xilinx. These are all the prelude to the next round of competition.
