一口气了解AMD | 别光盯着英伟达了~

Summary

AMD和NVIDIA是引领AI浪潮的两大芯片公司，股价自去年十月至今年三月翻倍，成为两大热门股票。AMD不仅在独立显卡市场占有重要地位，也是CPU市场的关键玩家。在与Intel长达半个世纪的竞争中，AMD经历了起伏，特别是在Lisa Su的领导下，通过战略调整和技术创新，成功实现逆袭，市值超过3000亿美元。AMD的故事从20世纪最伟大的发明之一——晶体管的诞生开始，经历了从Fairchild Semiconductor的人才摇篮到与Intel的激烈竞争，再到收购ATI进入GPU领域，最终在AI的推动下达到顶峰。

Takeaways

• 📈 过去一年中，NVIDIA和AMD的股价翻倍，成为两大牛股，背后的关键人物黄和苏均为华人。
• 🔍 AMD不仅在独立显卡市场占有重要地位，同时也是CPU市场的重要玩家，其主要竞争对手是Intel而非NVIDIA。
• 🚀 AMD在过去八年中实现了史诗般的逆袭，股价增长超过100倍，市值超过3000亿美元，超越了Intel，甚至高于茅台。
• 💡 传奇CEO Lisa Su（苏姿丰）带领AMD从低谷中崛起，借助AI的强劲东风达到顶峰。
• 🌐 半导体领域的人才摇篮Fairchild Semiconductor是硅谷的起源之一，其创始人之一的诺伊斯和摩尔后来创立了Intel。
• 🛠️ AMD在1970年代的半导体行业混战中并不占优势，但通过与Intel的合作关系，逐渐发展壮大。
• 💸 AMD通过收购NexGen获得了Pentium处理器的设计者，从而在1990年代末推出了与Intel竞争的高性能处理器。
• 📉 尽管AMD在2000年代初期取得了巨大成功，但由于收购ATI的成本过高，公司进入了最黑暗的十年。
• 🔄 苏姿丰上任后，通过简化公司流程、维护客户关系、削减成本等措施，开始了一系列经典的扭转操作。
• 🎯 AMD在高性能计算领域的战略调整，特别是推出基于Zen架构的Ryzen系列处理器，使其在CPU市场重新与Intel分庭抗礼。
• 📈 AMD通过收购Xilinx，进一步扩大了其在数据中心业务和AI领域的市场份额，展现了其在AI时代的强劲增长潜力。

Q & A

AMD和NVIDIA的股价在过去的一年中有何变化？

从上一年度的十月到今年三月，AMD和NVIDIA的股价翻了一倍，成为了两大牛股。

AMD在CPU市场上的主要竞争对手是哪个公司？

AMD在CPU市场上的主要竞争对手是Intel，两家公司之间的商业竞争可以追溯到半个多世纪以前。

Lisa Su在AMD的转型中扮演了什么角色？

Lisa Su是AMD的首席执行官，她在公司面临困境时加入，并实施了一系列战略调整，包括专注于高性能计算和扩展业务线，从而使AMD实现了史诗般的复兴。

AMD收购ATI后面临了哪些挑战？

AMD收购ATI后，虽然在GPU市场上取得了重要地位，但也面临了财务压力，现金流紧张，并且Intel加大了竞争力度，导致AMD在CPU和GPU两个领域都遭遇了激烈的竞争。

AMD的Zen架构有什么特点？

AMD的Zen架构是一种全新的框架，它使用了许多小芯片（称为Chiplets）集成到大芯片中，这提供了更大的灵活性和更高的产量。

AMD的Ryzen系列处理器在性能上有何突破？

AMD的Ryzen系列处理器在性能上实现了显著提升，第一代Ryzen处理器的性能就超过了Intel的八核心i7 6900K处理器的52%，而且价格不到一半。

AMD在数据中心业务上的策略是什么？

AMD在数据中心业务上的策略是推出基于Zen架构的EPYC系列处理器，并强调高性能和低总拥有成本（TCO），以此吸引企业用户从Intel转向AMD。

AMD收购Xilinx的原因是什么？

AMD收购Xilinx是为了显著扩展其业务范围，Xilinx是FPGA芯片领域的领先企业，其产品可应用于电动车、火星探测器、通信基站等多种设备。

AMD如何通过股票交换方式收购Xilinx？

AMD通过股票换股的方式收购Xilinx，即用AMD的股票交换Xilinx的股票。这种方式使得AMD能够在没有大量现金储备的情况下完成大规模收购。

AMD在AI领域的发展前景如何？

AMD在AI领域的发展前景非常广阔。AMD推出了一系列针对AI优化的芯片，并且在数据中心业务中取得了显著增长，AI芯片的需求量远远超过了供应量，为AMD提供了巨大的市场潜力。

Intel在面对AMD和NVIDIA的竞争时采取了什么策略？

面对AMD和NVIDIA的竞争，Intel选择了全力投入芯片制造业务，并投入200亿美元建设两个新工厂，计划到2027年赶上TSMC的工艺劣势。

Outlines

00:00

🚀 芯片巨头AMD的崛起与挑战

本段落介绍了AMD公司的发展历程和面临的挑战。AMD是全球领先的芯片制造商之一，在CPU和GPU市场中占有重要地位。段落中提到AMD与Intel的长期竞争，以及AMD在与Intel的竞争中如何实现逆袭。同时，也讲述了AMD在收购ATI后经历的困难，以及在Lisa Su的领导下，AMD如何通过高性能计算和优化产品线实现转型和复兴。

05:00

🌟 AMD的转型与突破

这一段讲述了AMD在面临巨大挑战时的转型策略和取得的突破。AMD通过推出高性能的Athlon处理器和K8架构，成功在CPU市场中占据了一席之地。然而，随后AMD的收购策略和市场定位导致其面临财务困境。在Lisa Su的领导下，AMD进行了战略调整，推出了基于Zen架构的Ryzen系列处理器，不仅性能大幅提升，还成功夺回了市场份额。此外，AMD还通过收购ATI成功进入了GPU市场，并在游戏机和数据中心领域取得了显著成就。

10:01

💡 战略调整与市场扩张

本段落重点介绍了AMD在Lisa Su的领导下进行的战略调整和市场扩张。AMD将重点放在高性能计算上，推出了基于Zen架构的Ryzen系列处理器，并在游戏和数据中心领域取得了成功。特别是在数据中心市场，AMD推出了EPYC系列处理器，通过优秀的性能和成本效益，逐渐从Intel手中夺取市场份额。此外，AMD还通过收购Xilinx进一步扩大了业务范围，进入了FPGA芯片领域。

15:02

🔄 收购Xilinx与财务策略

这一部分详细描述了AMD收购Xilinx的财务策略和过程。AMD通过股票换股的方式，以自身价值上升的股票作为交换，成功收购了Xilinx，而无需支付大量现金。这种策略不仅使得AMD能够以较低的成本扩大业务范围，还体现了AMD对其股票价值和未来发展的信心。收购完成后，AMD的市场价值和股价均有显著提升，进一步巩固了其在芯片行业的领导地位。

20:02

🌐 AMD的AI战略与市场前景

本段落讨论了AMD在AI领域的战略布局和市场前景。随着AI技术的快速发展，AMD推出了一系列针对AI优化的芯片产品，受到了市场的高度认可。AMD的数据中心业务在2023年取得了显著增长，成为公司的主要收入来源。同时，AMD在GPU市场与NVIDIA的竞争也呈现出不同于与Intel的CPU市场竞争的态势，双方更多是在共同开拓市场，而非零和博弈。AMD的成功不仅体现在产品上，还在于其对市场趋势的准确判断和战略决策。

25:03

🛠️ 芯片行业的未来与AMD的挑战

这一段分析了芯片行业的快速迭代特性和AMD面临的未来挑战。随着技术的不断进步，AMD需要持续进行研发投入和战略规划，以保持其在行业中的领先地位。Intel作为AMD的主要竞争对手，也在积极调整战略，投资建设新工厂，以期在工艺上赶超。AMD和Intel之间的竞争远未结束，双方都在为下一轮的技术竞赛做准备。

Keywords

💡NVIDIA

NVIDIA是一家领先的人工智能芯片公司，与AMD一同在AI领域取得了显著的市场份额增长。在视频中，NVIDIA的增长和其在AI领域的影响力被提及，展示了它在芯片行业中的重要地位。

💡AMD

AMD是另一家在CPU和GPU市场上具有重要地位的芯片公司。视频详细讲述了AMD的历史、与Intel的竞争、以及在Lisa Su的领导下取得的显著成就，包括股价的大幅增长和在AI领域的突破。

💡Lisa Su

Lisa Su是AMD的首席执行官，她的领导和战略调整对AMD的复兴起到了关键作用。视频中强调了她在公司中的重要作用，包括推动高性能计算、优化产品线和扩展数据中心业务。

💡Intel

Intel是全球知名的芯片制造商，在AMD的发展史中一直是主要的竞争对手。视频中提到了AMD与Intel之间的长期竞争，以及Intel在市场和技术上的领先地位。

💡Zen架构

Zen架构是AMD开发的一种CPU架构，它标志着AMD在高性能计算领域的重大突破。视频中强调了Zen架构对AMD产品性能提升的重要性，以及它在AMD复兴过程中的作用。

💡EPYC芯片

EPYC是AMD推出的一系列面向企业级用户的服务器处理器，中文名为“霄龙”。这些芯片在数据中心市场上取得了成功，帮助AMD在该领域与Intel竞争。

💡Xilinx

Xilinx是AMD收购的一家芯片公司，专门生产FPGA（现场可编程门阵列）芯片。这次收购帮助AMD扩展了业务范围，进入了新的市场领域。

💡AI芯片

AI芯片是专门为人工智能应用设计的处理器，它们能够提供所需的计算能力和效率。视频中提到AMD在AI领域的投入和产品开发，展示了公司在这一新兴技术领域的战略布局。

💡高性能计算

高性能计算指的是使用超级计算机或大量计算资源来解决复杂问题的技术。视频中强调了AMD在高性能计算领域的战略转变，以及这一转变如何帮助公司在技术和市场上取得成功。

💡Tick-Tock战略

Tick-Tock战略是Intel公司采用的一种产品更新模式，每年交替更新芯片的制程工艺（Tick）和架构（Tock）。视频中提到，由于技术问题，Intel在14纳米工艺上停滞不前，不得不放弃这一战略。

💡市场份额

市场份额是指一家公司在特定市场中销售产品或服务所占的比例。视频中讨论了AMD和Intel在CPU和GPU市场的份额变化，以及AMD如何在Lisa Su的领导下增加了市场份额。

Highlights

NVIDIA和AMD这两家领先的AI芯片公司从去年十月到今年三月股价翻倍，成为两大怪兽股票。

AMD不仅在独立显卡市场非常重要，它在CPU市场也同样重要。

AMD在与Intel的竞争中经历了起伏，过去八年上演了史诗般的逆袭。

AMD的股价在过去八年增长了超过100倍，市值超过了3000亿美元，超过了茅台。

AMD的传奇CEO Lisa Su带领公司从废墟中崛起，乘着AI的强劲东风达到顶峰。

20世纪最伟大的发明——晶体管的诞生，为半导体行业的发展奠定了基础。

Fairchild Semiconductor被誉为半导体领域的人才摇篮，也是硅谷的起源。

AMD与Intel之间的商业战争从Fairchild Semiconductor的高管出走开始，至今仍在继续。

AMD在创业初期并不顺利，与Intel的起步阶段相比，缺乏资金和人才支持。

AMD通过收购NexGen，得到了Pentium处理器的设计者，从而在技术上取得了突破。

AMD的Athlon处理器在1999年推出，迅速占领市场，一年内为AMD带来了10亿美元的利润。

AMD通过高价收购ATI，进入了GPU市场，并最终在CPU和GPU两大市场都占有相当的市场份额。

Lisa Su作为AMD的新任CEO，面对公司困境，采取了一系列战略调整，包括专注于高性能计算。

AMD的Zen架构是全新的框架，使用Chiplets技术，提高了灵活性和产量。

AMD的Ryzen系列处理器在性能上超越了Intel的i7处理器，且价格更低。

AMD在数据中心业务上取得了显著进展，其EPYC系列处理器性能超过了Intel的Xeon系列。

AMD通过股票换股的方式，以490亿美元收购了FPGA芯片领域的领先企业Xilinx。

AMD的股价在过去几年中增长了超过100倍，从濒临破产到市值达到3000亿美元。

AMD和NVIDIA在GPU领域的竞争不再是零和游戏，而是共同开拓市场的蓝海。

Intel在2021年更换了新CEO，并决定投资200亿美元建设新工厂，以期在2027年赶上TSMC的工艺劣势。

Transcripts

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.