第四季 第二集：ARM与软银

他于是创办了一家公司，想要实现这种计算愿景。

He goes and he starts a company to try and pursue this vision of computing.

他把公司命名为Active Book，专注于开发个人数字助理——也就是移动设备。

He calls it the active book company and he wants to focus on making PDAs, personal digital assistants which are mobile devices.

对于移动设备来说，最关键的就是电池续航。

And what's super important for mobile devices, battery life.

这是1988年吗？

And this is 1988?

对，就是1988年。

This is 1988.

所以电池续航特别重要。

So battery life is very important.

没错，非常重要。

Right, very important.

我觉得锂电池每年都能提升15%的性能。

I think lithium ion batteries get like 15% better every year.

所以倒退三十年，那时候的电池差多了。

So compound that back thirty years, like they were bad.

是啊，没错。

Yeah, right.

你现在的iPhone能撑多久？

Like how long does your iPhone last today?

根本不够用，想想那时候得是什么样。

Like not long enough, imagine back then.

现在世界上没几个人还记得1988年的事了，说实话，我那时候才刚有第一台电脑。

But there are just not that many people in the world that are thinking like 1988, shoot, like I think I got my first computer at that point.

我才四岁，那电脑大得像块铁疙瘩，摆在桌上，耗电能把家里一半的电力都吞了。

I was like four years old and it was a massive like, you know, hunk of metal that sat on my desk and probably used half of the house's power supply.

x86是什么东西？

What was an x86?

你还指望怎样呢？

What do you expect?

对，就是这样。

Yeah, right, exactly.

赫尔曼当时专注于掌上设备，他说：我觉得让我们以前在ARM开发的那些处理器，或许能促成这件事。

Herman, he's he's focusing on PDAs and he's like, You know what I think could make this happen is these processors we developed back at ARM.

于是他回到ARM的芯片团队，找到索菲，说：你们能不能重新设计一下，让芯片在保持高性能的同时，大幅降低功耗？

So he goes back to the chip team at ARM and to Sophie and he's like, Hey can you rework this whole thing to really optimize for power consumption while still maintaining this high performance?

他们说：没问题，我们可以做到。

So he's like, Yeah, we can do that.

后来真的做成了，他们开始和长期的硅谷合作伙伴VLSI合作生产这些芯片，因为那时候Acorn已经没有能力自己制造半导体了。

And they do and it works and they start producing these chips with their longtime Silicon partner VLSI because again, Acorn didn't have the resources to be manufacturing their own semiconductors at this point.

所以他们找VLSI代工生产芯片。

So they used a fabricator VLSI to make their silicon.

但与此同时，还有谁在考虑掌上设备呢？

Well at the same time who else is thinking about PDAs out there?

那是1988年，苹果公司。

This is 1988, Apple.

这时候，另一个牛顿——不是艾萨克·牛顿——就登场了。

And this is where the other Newton not Issac comes into play.

当时苹果已经没有史蒂夫·乔布斯了。

So Apple, Apple's under Steve Jobs has left at this point.

他已经被赶出了公司。

He's getting kicked out of the company.

那时候的CEO是约翰·斯卡利，而斯卡利的目标和赫尔曼·豪泽一样。

John Scully is CEO and John Scully's, you know, great white whale is the same thing as Herman Hauser.

就是个人数字助手。

It's the personal digital assistant.

人们总说斯卡利一无是处，但其实他当时有个远见，是计算机行业里极少有人能想到的。

You know, Scully gets like such a bad rap but like he actually had you know a vision that like very few people in computing you know did at the time.

连史蒂夫都没想到。

Not even Steve.

史蒂夫那时候正忙着搞NeXT呢。

I mean Steve was off doing Next, right?

他想打造一个超强的、功能强大的设备

He's trying to make a super Come powerful

那犯错和太早行动之间有什么区别呢？

on, what's the difference between being wrong and being too early?

没什么区别。

Nothing.

但好在他犯了错、又太早行动，因为斯卡利让拉里·特斯拉负责牛顿项目，这是他的心头好，而拉里确实了不起。

But good for the world that he was wrong and too early because Scully puts Larry Tesla at Apple on this Newton project which is his pet project and Larry is amazing.

拉里来自施乐帕洛阿尔托研究中心。

So Larry came from Xerox PARC.

他是最早那批计算机领域的元老级人物。

He was like an OG original computer guy.

他发明了复制粘贴功能。

He invented copy paste.

他带领着苹果内部一支精英团队，正在打造未来的计算平台。

He's he's like leading the SWAT team within Apple that's gonna build this computing platform of the future.

他知道需要一款低功耗、高性能的芯片。

He knows he needs a low power high performance chip.

他先去找了AT&T，那时候他们正在研发一款叫‘霍比特’的低功耗芯片——你根本编不出这种名字，1988年AT&T出的芯片居然叫‘霍比特’。

He first goes to AT and T which was working on a low power chip called The Hobbit which you can't make this stuff up just like you would expect from a chip from AT and T called The Hobbit in 1988.

那东西简直烂透了。

It sucked.

在所有方面都差得离谱。

Like it was terrible on every dimension.

这事儿还挺有意思的。

And this is fun.

我觉得我前几天聊过杰里·卡普兰的那本书《初创》。

I think one of my carve outs a few episodes ago was Jerry Kaplan's book Startup.

杰里是硅谷Go公司的创始人，那家公司当时也在做掌上电脑。

And Jerry, he was the founder of the Go Corporation here in Silicon Valley which was also trying to work on a PDA at the time.

后来他们被AT&T收购了，成了‘霍比特’项目的核心团队。

They had ended up getting acquired into AT and T and formed the backbone of The Hobbit.

总之，拉里开始接触半导体行业的人，想找个更好的芯片。

Anyway, Larry gets introduced, he's chatting with people in the silicon industry trying to find better chips.

有一天他跟VLSI的人聊天，对方说：我们有个合作伙伴叫Acorn，在英国，日子不太顺。

He's chatting one day with someone at VLSI and they're like, Hey, we've got this partner ACQ1 over in The UK and they're not doing too well.

他们本来想跟你们抢电话黄页的生意，但其中一位创始人现在在搞移动计算，弄出个芯片还挺靠谱的。

They were trying to compete with you guys in the telephone directory but one of their founders is doing something kinda interesting right now also in mobile computing and they have this chip that's working pretty well.

你不妨去看看。

You might wanna check it out.

拉里一听，立马就想把这东西弄到手。

And Larry's like, I need to have this right now.

于是他过去见了索菲和那边的工程团队，一看就说：这正是苹果做Newton需要的芯片。

So he goes over, he meets with Sophie and the engineering team over there And he's like, this is great.

但我没法从Acorn和Olivetti那里授权这项技术。

This is the chip we need here at Apple for the Newton, but I can't license this tech from Acorn and Olivetti.

毕竟咱们是竞争对手。

Like, we're competitors.

我们需要建立一个能在这里运作的结构。

We need to we need to create a structure that can work here.

于是他们和奥利维蒂、VLSI达成了一项协议。

So they architect a deal with Olivetti and VLSI.

他们把Acorn的芯片部门拆分出来，成立了一家新公司，以便从这家新公司授权处理器设计。

They spin off the chip division of Acorn into a new separate company that they can then license this processor design from.

他们动作很快，刚见面六周内就把这部门独立出来，成立了一家全新的公司，叫ARM。

They do it super fast within like six weeks of when when they get together, they've spun out the company into a new division, a new new totally separate company called ARM.

我简直震惊了，原来ARM是苹果为Newton项目搞的合资企业。

This blew my freaking mind that ARM was started as a JV with Apple on the Newton.

太不可思议了。

It's crazy.

于是苹果投资了150万美元进入这家公司。

So Apple invests 1,500,000 into the company.

所以ACQ1和奥利维蒂负责提供工程团队。

So ACQ1 and Olivetti are bringing the engineering group.

他们派了12名工程师去这家新公司。

They assigned 12 engineers to the new company.

VLSI是负责芯片制造的合作伙伴。

VLSI is the silicon fabricator partner.

他们获得了股权。

They get an equity stake.

苹果出钱。

Apple brings the money.

他们用一百五十万美元拿到了公司43%的股份。

They get 43% of the company for 1 and a half million dollars.

这简直是最稀释的种子轮投资了。

Again, talk about a dilutive seed round.

为了说明这有多讽刺，我先跳一步剧透一下：所有A系列芯片都是ARM架构的。

To blow out the irony of this, I'm gonna jump forward just to foreshadow, like, all of the A series chips are ARM processors.

iPhone最独特的地方有很多，但其中一点就是，他们早在几年前就领先全球，用ARM指令集架构实现了顶尖的制程工艺。

Like, the thing that is so differentiating about the iPhone, and there's a lot of things, but one of the things that they're just years ahead on is being able to have some of the best processes in the world all based on the ARM instruction set architecture.

我们后面会讲到这些公司是怎么慢慢分道扬镳的，但天啊，他们当初真的深度参与了这家公司的创立。

We will get to how these companies have sort of parted ways over time, but oh my god, they were actually involved in this founding of it.

毫不夸张地说，如果没有苹果内部的Newton项目，ARM这家公司根本不会存在。

It is no overstatement at all to say that without the Newton project within Apple, ARM as a company would not exist.

你手上正在用的手机，或者其他任何设备，样子都会完全不同。

The phones, you know, whatever device you're listening to this on right now would look very different.

你可能还得插着电源线用，但几乎可以肯定，你现在不会这么做。

You might have it plugged into a wall, which you almost assuredly don't.

这太不可思议了。

It's crazy.

这真的太不可思议了。

It's crazy.

这12位工程师后来独立出去，成立了新公司。

So these 12 engineers, they do the spin off.

他们从苹果拿到了一百五十万美元。

They get the 1 and a half million from Apple.

这十二位工程师走到剑桥街对面，把一间改造过的谷仓当办公室，还请来了曾在摩托罗拉担任高管的罗宾·萨克比来当新公司的CEO，之后就开始和苹果合作了。

These 12 engineers go down the street in Cambridge, they set up shop in a converted barn and they bring in this guy Robin Saxby who had been an executive at Motorola, to be the CEO of the new company and they get to work with Apple.

他们合作得非常紧密。

They're working super closely.

他们用的是自己掌握的核心RISC处理器技术，但必须做深度定制，适配到一款能用在Newton这种小型设备里的芯片里。

They're taking the core risk processor technology that they own but it needs to be super customized and fit into a chipset that's gonna work within this small device, the Newton.

所以他们和苹果紧密配合，一起开发，最终做出了这款处理器。

So they're working hand in hand and at with Apple to create it and they make the processor.

这款处理器叫ARM610，所有首批出货的Newton设备都搭载了它，后来虽然有升级，但每一台Newton都是靠它驱动的。

It's the ARM six ten and that goes into the Newton and every every Newton the first Newton's that shipped all had the ARM six ten processor in them and then it would get upgraded over time, but, it would power every Newton.

讽刺的是，还记得当初创办Acorn、启动了这一切的赫尔曼·豪泽吗？他在1991年把ActiveBook卖给了AT&T，而AT&T居然把它和自己收购的Go公司资产合并，搞出了一个叫Eo的产品，这种操作也只有他们干得出来。

Now ironically, remember Herman Hauser, the original Acorn cofounder who had started ActiveBook and kinda set all these wheels in motion, he ends up selling ActiveBook to AT and T of all people in 1991, and AT and T demands, merges it in with the assets of Go that it had acquired, which is now Eo within AT and T, only they could do something like this.

他们还强迫Eo改用Hobbit处理器。

They force it to start using the Hobbit processor.

当然了，说实话，我也不太明白。

And of course, I don't know.

我觉得这些掌上电脑确实上市了，但质量太差，市场完全不买账。

I think I think these these PDAs did end up shipping, but like they were so bad and it completely fails in the market.

大卫，你为这期节目挖的坑，我敢说已经刷新了新纪录。

David, the rabbit holes that you went down for this episode, I I believe, achieved new heights.

天哪。

Oh, man.

这真是太有意思了。

This is so much fun.

说到失败，AT&T用霍比特处理器，这操作简直蠢到家了。

Speaking of failure though, I mean, AT and T was like obviously stupid with, you know, the Hobbit processor.

你真该去读读《初创》这本书。

And and go read the book Startup.

书里把这段时期发生的各种荒唐事写得特别精彩。

Like, it's so good about all the ridiculousness that went on around this.

牛顿电脑当然也失败了，尽管它用的是ARM处理器。

The Newton also, of course, famously fails even though they have the ARM processor.

现在还太早了。

It's just too early.

这个世界还没准备好接受这个东西。

Like, the world isn't ready for this.

你给我讲讲具体情况吧。

So catch me up.

这部分我完全没做过研究。

Like I didn't research this part at all.

Newton失败之后，ARM那边发生了什么？

How does what happens to ARM post Newton failure?

Newton到1993年真正上市的时候，ARM已经整整两年只跟苹果合作了。

So when the Newton actually gets around to shipping in 1993, it's now been over two years that ARM's working exclusively with Apple.

Newton用的芯片，其实从很早就看得出来，销量根本上不去，这产品……

The Newton chips and like it's clear pretty early like there aren't gonna be enough unit sales here that like the products

这简直就是PDAs界的HomePod。

It's like the home pod of PDAs.

是的，这就像PDAs界的HomePod。

Yeah, it's the home pod of PDAs.

这时候，摩托罗拉的CEO罗宾上任了，他做出了几个极其出色的决定。

So this is where Robin the CEO from Motorola who came in, this is where he makes a couple really really brilliant decisions.

我们在之前的节目里提到过，我最近的一个科技主题是：当你能把一个巨大的技术浪潮和一个关键的商业模式创新结合起来——就像我们聊腾讯那次说的——那时候，奇迹就会发生。

And we've talked about in previous episodes that like one of my tech themes has been recently that like when you can marry a huge technology wave with a key business model innovation like we talked about on the Tencent episode like that's when something magical happens.

那么罗宾到底做了什么？

And so what did Robin do?

他想，好吧，我们得跟其他合作伙伴联手。

He was like, Okay well we need to work with other partners here.

我们不能只依赖苹果和Newton来大量采购我们的芯片，因为他们的授权费根本给不了多少。

We can't be dependent on just Apple and just Newton to buy a lot of chips from us because they're not gonna pay us a lot, in licensing fees here.

但他也意识到，让这款处理器真正成功的关键，不只是ARM设计好芯片然后直接交给苹果，而是他们深度嵌入了Newton的工程团队，为这个设备量身打造了高度定制化的解决方案。

But he's also realized that like part of the magic of what made this processor really work wasn't just that like ARM designed this chip and like handed it over a whole cloth to Apple, it was that they really embedded with the engineering teams on the Newton and made something like pretty customized that worked specifically for this device.

所以他想，我觉得我们完全可以对更多公司这么做：把我们的核心处理器技术提供给他们，深入他们的团队，帮他们开发出专属于各自产品的定制芯片。

And so he's like, you know, I bet we could do this with lots of people and we give them this core processor, this core risk processor technology we could embed within their teams and we can help them develop essentially custom silicon for their use cases.

你知道这里最有趣的是什么吗？

And you know what is really interesting here?

如果我们能建立一个跟这个模式契合的商业模式，比如这么说：当我们这么做的时候，你付给我们一笔预授权费，获得我们核心架构技术的授权，同时为我们要派驻你们那里的工程时间付费，我们就靠这个赚钱。

Like if we could create a business model that aligns with this, what if we say like, okay, when we do this, you pay us an upfront licensing fee for the rights to our core risk technology and you pay us for our engineering time for embedding with you and we'll make money on that.

但让我们把重点放在实际出货的台数上

But let's align, get aligned on like actually shipping units

我希望你们能多出货

want you to the to ship

大批量的设备。

bunch of units.

那不如这样，每卖出一台搭载我们技术的设备，我们都收一点点授权费，怎么样？

So how about we take a small royalty on every device you ship that has our technology in it?

而且说实话，这在当时还挺有意思的。

And then at the time like this is like, somewhat interesting.

毕竟，市面上潜在能出货的移动计算设备数量可是相当可观的。

Like how many devices could potentially ship that are like mobile computing devices?

所以ARM收了三次钱。

So ARM paid three times.

比如，他们收取许可费，因为你有权使用ARM的技术。

Like they get the license fee from, hey, you get to use the ARM technology.

你知道，这套指令集和架构是你可以在芯片上使用的特权。

You know, this instructions and architecture is one that you have the privilege of using on your chip.

这种模式大概有两种方式。

And there's sort of two ways that that can work.

要么他们为制造商设计芯片，要么他们说：你是个不错的设计师，这就是苹果现在的方式。

Either they design the chip for that that, manufacturer, or they say, here, you you feel like you're a good designer, and this is sort of how Apple's relationship works today.

你使用我们的指令集，然后自己去实现。

You use our instruction set, and and you do it.

所以他们第二次收费，是通过深度参与你的开发过程。

So then they get paid the second time for actually embedding with them.

这属于他们的软件与服务业务线。

That's their sort of software and services line of business.

然后他们第三次赚钱，每卖出一个单元，就能从每个出货的CPU成本里分到一小笔钱。

And then they get that third time which is every unit sold, they get a tiny little piece of the cost of each CPU that shipped out.

是的。

Yeah.

事实上，这正是科技行业神奇的地方。

Well it turns out, I mean this is one of the things that makes the technology industry magical.

哪怕只占一块巨大蛋糕的极小部分——这块蛋糕的规模，比全世界所有人加起来还要大上百倍——而如今全球设备数量已经多到让这一小块也变得极其可观。

A very very tiny piece of a pie that is literally almost a 100 times bigger than every person in the world combined, which, the number of devices in the world is at this point, turns out to be a very, very large slice of pie.

你知道到现在为止一共出货了多少颗ARM芯片吗？

Do you know how many ARM chips have shipped to date?

我知道，不过你来说吧。

I do, but why don't you go for it?

一千三百亿。

130,000,000,000.

对。

Yeah.

这是十亿，B-I-L-L-I-O-N的那个十亿。

That's billion with a b.

那可是一大堆硬币了。

And those That's lot of pennies.

你想想看。

And and think about that.

这些全是ARM芯片，每一块都是某个设备的核心处理器。

Those are ARM chips, each of which is the core processor in a device.

也就是说，总共有1300亿台设备。

So that is a 130,000,000,000 devices.

不一定这么算。

Not necessarily true.

很多设备里头有

Lots of devices have

不过很多设备本身装了好几个芯片。

Well, many devices have multiple chips within them.

是的。

Yes.

但即便如此，你知道，也还行。

But still, like, you know, it's, okay.

就算减掉四分之一、五分之一，或者随便多少。

Reduce that by, you know, a factor of four or five or whatever.

那还是远远超过地球上的人口总数。

Like, that's still, like, way more than there are people on Earth.

是的，他们签的第一份合同就很聪明，能从出货量中分得收益。

Yeah, that was a thoughtful first contract to sign that they get some upside on units shipped.

没错。

Yeah.

而且更厉害的是。

And what's also cool.

所以说，ARM这个例子太棒了，技术和商业模式相互促进，彼此提升。

So again like I think this is such a cool example ARM is of technology and business model playing together and and like making each other better.

所以很明显，这背后还有财务方面的考量。

So like obviously there's the financial aspect of this.

有趣的是技术层面，现在ARM和其他芯片公司不一样，比如英特尔。

What's interesting on the technology side too because ARM now like all the other chip companies take Intel for example.

他们就说，用我们的ARM7或者什么核心吧，诸如此类。

They're like use our, you know, Adam seven whatever chip or core I blah blah blah.

我会给你这个芯片，你把它放进你的设备里。

I am going to give you this chip, you are going to put it in your device.

但ARM的态度是：不不不，我们和你是一条心。

Because ARM is like, no no no, we're aligned with you.

我们希望你做出最好的产品，推出最棒的设备，我们会把技术团队深度融入你的项目。

We want you to make the best products and ship the best devices and we'll embed our technology and our teams with you.

你可以完全按自己的想法来设计，我们全力配合。

We can collaborate on designing it however you want.

正是ARM这种模式，真正推动了系统级芯片的快速发展。

So this is what really it's ARM and this model that starts to enable systems on a chip to be really take off.

那么，什么是系统级芯片呢？

So what is the system on a chip?

回想以前玩电脑的时候，就像我们前面说的，一台PC有主板、处理器、显卡、声卡，我还记得当年自己动手组装电脑，把一堆东西插上去。

Like back in the PC days when that we were talking about earlier, you know a system, a PC, you had a motherboard, you had a processor, you had a graphics card, you had a sound card, you had a I remember plugging all this stuff in and building my own PCs back in the day.

是啊，没错。

Yeah, man.

全都插在PCIe插槽里，

Put it all in your PCIe slots and

没错，就是这样。

call it Exactly.

天哪，

Oh, man.

太不可思议了。

Incredible.

这是PCI Express吗？

Is that PCI Express?

Express。

Express.

对吧？

Right?

是的。

Yes.

那是上一代的规格了。

That a was a louder generation.

没错。

Yeah.

那时候真是，天啊。

That was like, oh, man.

太疯狂了。

Crazy.

于是人们开始意识到，特别是在移动设备上，如果能把所有这些都集成到一块芯片上，而不是用一堆独立的芯片、总线和主板，会怎么样？

So people start to realize especially in a mobile environment like, what if you could just put all of that just on one chip instead of having separate chips and buses and motherboards and whatnot?

ARM那边就说，行啊，没问题。

ARM's like, yeah, cool.

就把我们的处理器集成到一块芯片里。

Like put our processor in one chip as part of it.

我们可以帮你搞定这个。

Like we'll we'll help you with that.

所以苹果现在用的A57，也就是那个仿生超人芯片，就是这么来的。

And so Apple now with the core, you know, the a 57, you know, bionic Superman chip.

更进一步了。

Extra form.

没错。

Yeah.

三星、高通、德州仪器这些芯片，你手机里以前一堆独立芯片，但现在核心功能基本都集成在一块硅片上了。

All the Samsung chips, Qualcomm chips, TI chips, like every chip your phones have multiple chips, but at the core most of the technology is being done on one actual piece of silicon.

这标志着从独立CPU时代过渡到了如今的系统级芯片时代。

This was the transition from the era of sort of the just discrete CPU to two system on a chip, which is you know where we are today.

对。

Yep.

就在同一年，1993年，我要给ARM公司和CEO罗宾点个大大的赞，他们真的扭转了局面。

And so actually that same year in 1993 and like again so much kudos to ARM the company and to to Robin the CEO for turning this around.

就在牛顿项目失败的那一年，他们和德州仪器达成了一项重要合作，为诺基亚提供处理器核心，用于诺基亚0610或6110手机。

The same year that the Newton ships and they realized this isn't gonna work, they signed a landmark deal with Texas Instruments to provide the core of a processor that Nokia has contracted with TI to go into the Nokia 06/10 or 6110.

我们会在节目笔记里放个链接，方便你们查看。

We'll try and put a link to this in the show notes.

你一旦看到这部手机，就绝对忘不掉。

Once you see this phone, you are going to remember this phone.

我觉得我人生第一台手机，就是这款的改良版。

This is like, I think the first cell phone I had was a variant on this.

是的。

Yeah.

这就是经典的直板手机。

This is the candy bar phone.

这是第一款在全球，包括美国，大规模面向消费者的GSM手机。

This is the first major consumer GSM phone that is sold certainly in America and all over the world.

正是从这款手机开始，手机出货量的曲线发生了转折，最终走向了今天的智能手机时代。

And this is what starts the kink in the curve of cell phone shipments that ends up with smartphones and where we are today.

大卫之前提到过，为什么ARM芯片尽管在很多方面更优秀，却难以打破微软、英特尔和Windows系统的双头垄断。

David was talking earlier about why these ARM chips were even though they were better in a lot of ways, they couldn't penetrate the duopoly of Microsoft and Intel and Windows running there.

你现在可以看到，运行在这部诺基亚手机上的软件——我相信大家都玩过贪吃蛇——和Mac或Windows上的软件完全不同。

You can start to see now the software that ran on that Nokia phone that I'm sure everyone played Snake on is nothing like the software that ran on Macs or on Windows.

贪吃蛇简直就是杀手级应用。在诺基亚6110之前，手机里的处理器根本跑不动贪吃蛇。

Snake was like the killer app ARM processors were literally like before the Nokia six thousand one ten, the processors in cell phones weren't good enough to even run Snake.

但现在，你有了这种低功耗、高性能的处理器，完全可以运行游戏了，对吧？

But now you have this low power high performance processor that can run games, right?

从那时起，一直到《农场小镇》《部落冲突》，再到今天的一切，都能直接追溯到这个起点。

And you can like draw a direct line from there to Farmville to Clash of Clans to everything.

没错。

Right.

但还有一座桥没跨过去，就是我们怎么从那种设备——那部手机更像嵌入式设备，而不是我们今天熟悉的手机——演变成现在的样子。

The bridge that hasn't gotten crossed yet is how we went from that, which really, I mean, that phone felt more like an embedded device than really like phones as we know today.

而今天的手机，本质上就是个人电脑。

And phones today are are effectively PCs.

如果你想想当初开发第一版 iOS 所做的工作，其实就是在精简 macOS。

And if you think about the work that was done to create the first version of iOS, it was really to strip down macOS.

大家总爱说 iOS 就是把 macOS 瘦身来的，还用着同样的 Darwin 内核什么的。

And people always harp a lot on this like stripped down macOS to to create iOS and it still uses the same Darwin kernel and all that stuff.

但更了不起的是，要把一个电脑操作系统搬到手机上，还得让它适配 ARM 芯片。

But the other crazy piece of work that had to be done to bring a computer operating system to to these mobile phones was adapting it for the ARM chipset.

要把原本为 PC 设计的操作系统，改到那些功耗极低、指令集完全不同的芯片上运行。

Like to bring PC operating systems to something that would operate on these chips that weren't getting a lot of power and had completely different instruction set.

iOS 和 Android 现在能在这种当初根本没打算支持的芯片上跑得这么好，简直让人难以置信。

The fact that iOS and Android works the way it does today on this chip that was nothing like what they were originally architected for is mind blowing.

说起来挺有意思，我到现在才反应过来——我敢打赌，如果你想想微软为什么错过了移动时代？

You know it's interesting, I didn't put two and two together till right now but I strongly suspect if you think about why did Microsoft miss mobile?

原因有很多，对吧？

There are lots of reasons, right?

文化上的，还有技术上的各种因素。

Cultural, what have you technology.

但我觉得，微软在移动领域失败的一个关键原因，是他们当时用的 Windows Mobile。

But actually this specific technology reason I think one could of the major points that Microsoft had Windows Mobile, right?

微软确实有 Windows CE 和嵌入式 Windows 之类的系统，对吧？

And Microsoft had Windows CE and embedded Windows and whatnot, right?

但那些系统

But those were

完全是另一套代码。

Completely different code base.

完全不同的架构。

Completely different code

基础，对吧？

base, right?

那根本不是微软的Windows，因为Windows和DOS只能运行在复杂指令集架构上，对吧？

That wasn't Microsoft Windows because Microsoft Windows and DOS only ran on complex instruction set architectures, right?

他们不可能为了在RISC架构上运行，就把整个系统重新架构一遍。

They're not going to re architect that whole thing to run on risk architecture.

没错。

No.

我的意思是，他们现在确实做了。

I mean they do now.

是的，现在确实做了。

Yeah, they do now.

但那是最近几年的事了。

But the last couple of years.

智能手机兴起的时候，为什么苹果能做到这一点？

Smartphones were taking on why could Apple do this?

苹果一向对他们的架构更开放，Mac OS和OS X支持的硬件平台一直都很开放，对吧？

Apple had always been much more open about their architectures that Mac OS ran on, that OS X ran on, right?

而且呢，

And like,

哦，他们是不是已经从摩托罗拉的处理器，也就是摩托罗拉的PowerPC，过渡到……嗯，是的，

oh, did they already do the transition from the Motorola processors, the Motorola Power PCs to Well, had yeah,

他们得把OS X移植到x86的英特尔芯片上，对吧？

they had to port into the x86 Intel with OS X, right?

所以他们对Mac OS和OS X移植到不同架构这件事，没那么死守，也更开放。

So they were much less wedded to and open to porting their Mac OS and OS X into different architectures.

而且他们本来就有Newton的技术底子。

And indeed they had this Newton DNA as well.

对不太了解背景的人解释一下，iPhone是2007年推出的，他们居然把macOS搬到了ARM芯片上。

For folks who don't know specifically what we're talking about, so the iPhone came out in 2007 which holy God they put macOS on ARM chip.

就在那之前两年，我们刚才提到的这件事，发生在乔布斯一场堪称传奇的发布会上，至今都被奉为经典表演。

Two years before that, the thing that we're referencing here is in a very Jobsian keynote that continues to go down in history as just an amazing piece of showmanship.

史蒂夫·乔布斯走上台说：我们要把所有新Mac里的芯片，从一直用的PowerPC，换成英特尔的。那场面，真是被他演得极具戏剧性。

Steve Jobs came out and said, so we're changing the, chips that are in all the new Macs away from the PowerPC that we've been using to using Intel, which of course was so dramatized.

保罗当场就走上了台，天哪，苹果居然用上了英特尔芯片，看看那些旧日纷争留下的残骸，战火至今未熄，而现在事情又有了新进展。

And Paul audibly walked out on stage and oh my god, Apple is using Intel, look at all the old wars that are still the ground is burning from all the carnage and wreckage of those old wars, and here's what's happening now.

史蒂夫当然说了，最疯狂的是，你们根本没意识到，这种情况已经持续一年了。

And Steve of course says, and the crazy thing is that you guys don't even realize it's been that way for a year.

过去一年的所有操作系统

And all the operating systems dating back a year

或者说是两年，我原以为是好几年呢，是的。

or I two years didn't know it was multiple years, yeah.

是的，这些系统其实早就具备这个能力了，只是你们一直不知道。

Yeah, have been capable of doing this and you guys just didn't know it.

所以赶紧去买台新电脑吧，你以前的所有东西都能在上面正常运行。

So go buy yourself a new computer and all your old stuff will work on it.

大卫，你说得对。

David, you're right.

我觉得这确实给了他们信心，让他们开始考虑，也许我们可以重新设计操作系统，让它适配另一款芯片。

I think that did give them sort of the confidence to say like, well, I guess we could start re architecting our operating system to work on yet another chipset.

那时候iPhone项目早就启动了，而且你们应该快出货了，我记得那会儿是2006年吧。

Well, of course at that point in time the iPhone project was well underway but you're probably getting close to shipping because I think it was 2006 right when that happened.

我想知道这些努力有多相似。

I wonder how similar those efforts were.

是的。

Yeah.

挺有意思的。

Interesting.

不过咱们还是回头说说ARM吧。

Well, anyway to rewind back to ARM.

事情就是这样发生的。

So this is happening.

他们跟TI做了这笔交易，诺基亚6110发布了。

They do this TI deal, the Nokia sixty one ten launches.

从那以后，ARM的发展势头就一路向上，再没停过。

Things are all basically up into the rate indefinitely from then, for ARM.

到1997年底，ARM的年收入已经超过两千五百万英镑，而且已经盈利了。

So at the end of nineteen ninety seven, ARM is doing over £25,000,000 in revenue and they're profitable.

他们同时在伦敦证券交易所和纳斯达克上市，别忘了，他们有英国、意大利的股东，还有苹果这样的美国股东。

They do a dual IPO both on the London Stock Exchange and the NASDAQ because remember they have British shareholders, Italian shareholders, and an Apple, American shareholders.

1998年4月17日，他们以每股5.75英镑的价格定价上市，公司估值达到两亿六千四百万英镑。

On 04/17/1998, they priced the IPO at £5.75 per share which translates to a market cap of £264,000,000.

那时候汇率比现在强多了，但即便如此，天啊，真希望我当年能参与IPO。

Now the exchange rate was stronger back then, but still, my god, I wish I could have invested at the IPO.

在整个科技泡沫期间，ARM的市值暴涨，这简直太棒了。

It grows hugely throughout the tech bubble especially, and this is this is so awesome.

我都不敢相信，这件事在这集里可能就一笔带过，但我真想特别提一下。

I can't believe this is gonna be like a footnote in this episode, but like I wanna highlight here.

正是这件事救了苹果。

This saves Apple.

如果没有ARM的这次上市，苹果在1998年几乎肯定会破产。

The ARM IPO had it not happened, very likely Apple would have gone bankrupt because this is 1998.

史蒂夫·乔布斯刚刚重返苹果。

Steve Jobs had just come back into Apple.

斯卡利被赶下了台。

Scully was ousted.

公司当时亏得厉害，真的濒临破产。

The company's bleeding cash like they are like seriously facing bankruptcy.

但十年后，他们还持有ARM的那部分股份吗？

And what it's a decade later but they still own that share of ARM?

没有。

No.

没有。

No.

没有。

No.

没有。

No.

他们开始逐步抛售ARM的股份，幸运的是，上市时并没有一下子全卖完。

They start liquidating the ARM shares and thankfully they don't liquidate all at the IPO.

接下来几年里，他们慢慢卖出，而ARM的股价却一路飙升。

They start selling slowly over a couple years as the ARM stock price is going crazy.

别忘了，苹果当初只投了一百五十万美元。

Apple, remember they invested 1 and a half million.

在接下来的几年里，他们靠出售ARM股票赚了七亿九千二百万美元的利润。

They make $792,000,000 in profit from selling their ARM stock over the next couple of years.

这简直就是救了公司一命。

And literally that is what saves the company.

要是没有这笔钱，公司早就破产了。

Like it would have gone bankrupt without that.

想象一下，你是个分析师。

Imagine being an analyst.

现在满大街都是苹果分析师，但你想想，当年当苹果分析师是什么感觉？

Like now there's an army of Apple analysts but like imagine being an Apple analyst and you're like, okay.

又一个季度，他们的利润根本不是来自正常运营，而是全靠卖完这批迟早要耗尽的资产。

Yet another quarter where all of their it wouldn't even be operating income, but all of their profit is coming from liquidating this thing that's going to run out.

说真的，正是这笔钱让公司撑了下来。

I mean, was literally what kept the company afloat.

就在这期间，史蒂夫正忙着打造后来的iMac，整合产品线，讽刺的是，还亲手终结了Newton。

Kind of amazing while while Steve was preparing what would become the iMac and consolidating the product lines and ironically killing the Newton.

等到2001年科技泡沫、互联网泡沫和电信泡沫接连破裂时，ARM当然也受到了一点冲击，出货量基本停滞了一年左右，甚至可能都没到一年。

So when the tech bubble bursts and when the internet bubble bursts and the telecom bubble bursts in 2001, ARM of course you know takes a blip and shipments basically like plateau for you know a year maybe not even.

但话说回来，这股浪潮实在太大了——到2002年，ARM的出货量就重新爬升，到了2010年，ARM的合作伙伴每年出货量已经超过50亿台设备。

But again like this is such a huge wave like by 2002 they pick back up to the point where in 2010 ARM is now ARM partners are now shipping over 5,000,000,000 devices every year.

也就是说，几乎全球每个人都能接触到一台由ARM合作伙伴生产的设备。

Again, so almost as many people as there are on the planet, ARM's partners are shipping devices.

当然，这些不只是手机，还有微波炉、冰箱、汽车——汽车里头可藏着不少ARM芯片。

Like you know these are these aren't just of course cell phones, these are you know microwaves, these are refrigerators, these are cars, like cars have a lot of ARM chips in them.

还有各种传感器和设备，只要是需要低功耗、高性能嵌入式处理器的东西，都离不开ARM。

These are sensors and devices, anything that needs a low power, high performance embedded processor.

好了，各位听众。

Alright, listeners.

现在正是感谢我们节目的老朋友ServiceNow的好时机。

Now is a great time to thank our longtime friend of the show, ServiceNow.

如果你在管理一家大型企业，AI代理很可能已经遍布各个部门，而部署它们已经不再是难题。

If you are running a large enterprise, AI agents are likely spread across every team, and deploying them is, no longer the hard part.

对。

Yeah.

真正的难点在于搞清楚它们拥有哪些权限，员工们用它们来做什么，以及AI到底在做哪些决策。

The hard part is knowing what permissions they have, what employees are using them for, or what decisions AI is making.

在企业规模下，AI安全可不是个小问题。

AI security for an enterprise at scale is not a small concern.

风险是实实在在存在的。

Like, the risks are real.

没错。

Exactly.

AI的挑战在于如何管理、保障安全、评估效果，确保它真正创造价值。

And the challenge with AI is governing it, securing it, measuring it, and making sure that it actually delivers value.

正因如此，ServiceNow推出了AI控制塔。

That is why ServiceNow built the AI control tower.

没错。

Yep.

AI控制塔为企业提供了一个统一平台，可以查看、管理、规范和优化全公司的AI应用。

AI control tower gives enterprises a single place to see, manage, govern, and optimize AI across the entire business.

而且它兼容任何AI系统，不只是他们自家的。

And it works with any AI, not just theirs.

你网络上的每台设备、每个系统的每项权限、每个AI代理，都能在一个地方看得清、管得住、保得安。

Every device on your network, every permission across every system, every AI agent visible and secure in one place.

ServiceNow能做到这一点，是因为他们花了二十多年，打造了企业运营的核心骨架——包括工作流、治理机制、审批流程、安全控制，以及IT、人力、客服、财务和安全部门日常运作所依赖的系统知识。

And ServiceNow can do this because they've spent more than 20 years building the operational backbone of the enterprise, the workflows, governance, approval, security controls, and institutional knowledge that power how work actually gets done across IT, HR, customer service, finance, and security.

ServiceNow每年已经处理超过一千亿个工作流，为超过85家《财富》500强企业完成数万亿笔交易。

ServiceNow already runs more than a 100,000,000,000 workflows annually and trillions of transactions for more than 85 of the Fortune 500.

所以当企业需要一个能够大规模管理AI的平台时，他们选择的正是已经融入自身业务核心的系统。

So when companies need a place to govern AI at enterprise scale, they're building on a platform at the center of how their business already operates.

而未来的趋势并不是只有一个AI，而是成千上万个AI代理会遍布公司的每一个部门。

And in a future that isn't going to be one AI, it's going to be thousands of AI agents working across every function of the company.

但问题是，谁来统一管理这些AI代理呢？

But the question is, who's managing them all?

如果你正想把AI的愿景转化为实际的业务成果，并确保它安全、可靠地规模化落地，不妨去 servicenow.com/acquired 看看，记得提一下是Ben和David推荐的。

So if you're trying to turn AI ambition into real business outcomes and make it work safely, securely at scale, go check out servicenow.com/acquired and tell them that Ben and David sent you.

通常这时候，我们会总结一下公司的历史和关键数据。

So typically this would be where we would wrap up our history and facts.

公司发展得不错，但还没上市。

Companies just growing nicely However IPO.

我们会对他们的上市表现做个评价。

We'd grade the IPO.

我们会说，要是苹果当初没抛售股票，一直拿着该多好啊。

We'd say like, man, if only Apple had hold on to their shares and didn't need to liquidate like

这简直有点疯狂。

It's kind of crazy.

如果不是软银收购了他们，我都不确定我们会不会做这一期，因为如果你一直听到现在，大概也同意我的看法。

I'm not sure we would be doing this episode if SoftBank hadn't bought them because it is sort of like, if you've bared with us this long in this episode, you probably agree.

这是一家非常冷门的技术公司，商业模式却荒谬得离谱。

It's this sort of obscure technology company that has an absolutely inane business model.

在继续之前，大卫，我们得把他们的商业模式说清楚：他们自己并不生产芯片，大约一半，或者略少一点的收入来自芯片设计和授权，还有一大部分则是纯粹授权指令集和架构。

Before moving forward, David, it worth putting a fine point on the business model where they don't manufacture chips, about half of their business or a little bit less I think comes from designing the chips and then licensing those out, and then, you know, they have this this large component of their business that is just licensing the instructions and architecture.

他们通过自己创造的知识产权，再加上围绕指令集构建的生态系统，成功锁定了市场价值。

Like, they have now managed through the value they've they've created through the intellectual property, but also the lock in they've created by building an ecosystem around their instruction set.

他们只是把指令集授权出去，就能从所有使用其技术的芯片中分得一杯羹。

They just license their instruction set, and it means they get a cut of all chips that are manufactured using their IP.

我到现在还是觉得不可思议，怎么没人站出来说：喂，这饭碗可不是你的。

And that sort of continues to blow my mind that somebody doesn't say like, sorry, that's not your lunch.

这可不是小数目，拿iPhone来说，我印象中他们每台能赚大约35美分，占芯片售价的1%到2%，具体要看是哪种商业模式生产的。

And this isn't like trivial amount of money, think it's something like if you look at the iPhone, I think they make like 35¢ per unit, it's something like one to 2% of the chip selling price depending on which of their business models it was manufactured under.

但每颗A系列芯片里那35美分，他们是怎么做到一直稳稳拿走这份分成的？

But like 35¢ of every you know A Series chip, why is it that they've managed to to hold on and actually get a cut of every single chip manufactured?

我觉得可能有两个原因。

I think it's probably two things.

第一，他们始终拥有顶尖的处理器和芯片工程设计人才，对吧？

One, they continue to have truly excellent processor and chip engineering and design talent, right?

在如今性能和功耗的提升比以往任何时候都更重要的时代，他们确实是全球最顶尖的处理器设计公司之一。

So they truly are one of the world class you know best processor design companies in the world at a time where advances in both performance and power consumption for processors are more important than ever.

但我觉得，这跟当年Wintel双寡头垄断的原因其实是一样的，对吧？

But I think also it's it's the same reason why the Wintel duopoly existed, right?

虽然苹果在架构切换上比当年的微软更灵活，但Android和iOS都跑在RISC架构的芯片上。

Like even though Apple is more flexible about moving architectures than you know Microsoft was back in the day, Android and iOS run on RISC architecture chipsets.

他们真会为了别的架构重新设计吗？

Are they really gonna re architect them for something else?

而且根本就没什么别的架构可换。

There even is there is no other thing to re architect to.

是的

Yeah.

没错

That's true.

你得自己设计一套指令集，再开发出

You would have to come up with your own instruction set, build your own

一套编译器来支持操作系统。

Set of compilers to enable operating systems.

其实苹果早就已经在做这些了。

So Apple does a lot of this anyway.

苹果是少数几家真正有能力这么做公司之一，因为他们完全可以自研指令集。

Like Apple is one of the companies that actually could probably uniquely do it because they could write their own instruction set.

这已经有点过度延伸到科技趋势和未来预测了。

I mean, is like way bleeding into tech themes and random future forecasting.

但他们完全可以开发新的编译器，集成到Xcode里，让所有编译出来的应用都能直接运行在他们新推出的非ARM芯片上。

But like they could write a new compiler, build it into Xcode, make it so that every app that's compiled just works on their new chipset that's not ARM based that they launch.

就像他们2006年对付英特尔那样。

Just like they did with Intel back in 2006.

没错。

Right.

这我一点都不意外，因为他们早就掌控了整个链条——虽然不自己生产芯片，但芯片都是他们自己设计的。

And it wouldn't surprise me because they've they've taken over not only their own I mean, they don't manufacture their chips, they design their own chips.

当然，他们还是从ARM授权了指令集。

They of course license the ISA from from ARM.

但要说他们下一步干脆用自己研发的架构，这倒是很符合他们一贯的垂直整合风格。

But like it does just seem like it would be one more expanding of their vertical integration to say actually it's gonna be based on our own architecture now too.

我赌未来三年内他们真会这么做。

Color me in for that prediction in the next three years.

有意思。

Interesting.

而且只有苹果有这个条件这么做，因为谷歌不可能对Android这么做，Android本身是个碎片化的生态。

And also Apple is uniquely positioned to be the only company to do it because Google's not gonna do it with Android because Android is a diverse ecosystem.

需要ARM成为芯片行业的统一标准，覆盖TI、高通、博通，还有所有合作伙伴，再加上谷歌，等等。

Need ARM to be the architecture to be the standard across the chip in TI, Qualcomm, whoever, Broadcom all the partners you need plus Google plus what have you.

谷歌和安卓的情况有点像微软，甚至比微软更糟，因为你没法保证它用的是Intel芯片。

Google with Android is kind of the Microsoft it's actually a little bit worse of a position than Microsoft because you can't guarantee that it's an Intel chip.

它可能是高通的芯片，但也可能是别的任何厂商的。

It's probably a Qualcomm chip but it could be a bunch of things.

好吧，咱们先别谈技术趋势了，回到

Well All right so To pull back from tech themes back to

咱们来聊聊这次收购吧，毕竟这档播客的主题就是这个。

our Let's talk about the actual acquisition because that's the name of the podcast.

这一集简直太分裂了，到目前为止，我 literally 一直在拉

This is like such a schizophrenic episode because until now Like I literally was pulling

我们走向未来。

us into the future.

这家公司和这项技术太重要了，希望我们已经足够清楚地传达出这一点了。

This is so important, this company and this technology and I hope we've done a good enough job communicating that.

但这期节目简直就是硬核科技专题。

But this is like a hardcore technology episode.

不过现在我们要转个方向，聊聊疯狂的软银和孙正义，简直像007电影一样。

But now we're gonna shift to grab the wheel and shift back to like crazy, you know, James Bond style SoftBank and Masa.

好的。

Okay.

这就带我们回到2016年。

So this brings us to 2016.

那时候世界一片大好。

All is well in the world.

ARM表现亮眼，总部在剑桥，工程团队出色，合作伙伴众多，商业模式也非常成功。

ARM is winning, you know, based in Cambridge, doing great engineering, working with partners, got their great business model.

前一年的营收达到了16亿英镑。

Did about 1,600,000,000 in revenue the previous year.

没错。

Yep.

股价一路飙升，因为金融圈开始意识到：如果你想搭上移动科技的顺风车，ARM真是个绝佳的投资选择。

The stock price is is having a huge run because the financial community is starting to wake up to this fact that like, oh hey, if I wanna ride the mobile wave and I want exposure to this, ARM is like a really good way to do it.

没错。

Yep.

2015年我们出货了150亿颗芯片，事情正在迅速发展。

We shipped 15,000,000,000 units in 2015 like it's a things are happening.

对。

Yep.

多亏了《快公司》、卡特琳娜·布鲁克，还有我们节目的朋友大卫·利德斯基，他们刚发表了一篇关于马萨和软银的精彩报道。

So thanks to Fast Company and Katrina Brooker and friend of the show David Lidsky and their great piece on Massa and SoftBank that just came out.

所以我们知道接下来会发生什么。

We know what happens here.

2016年夏天，马萨在加州伍德赛德占地九英亩的豪宅里，举办了一场科技界名流晚宴，那地方简直——

So it's summer of twenty sixteen and Massa is hosting a dinner for tech industry luminaries at his amazing nine acre estate in Woodside, California, which is like

一个

a

超级有科技氛围的社区，就在帕洛阿尔托外面，紧挨着280号公路。

super Tony community, you know, right outside Palo Alto, right right over two eighty.

这是史蒂夫·乔布斯那栋一直没建完的豪宅所在地。

This is where Steve Jobs' mansion was that he never built out.

这是拉里·埃里森整个庄园的所在地，而马萨的庄园也毫不逊色。

This is where Larry Ellison's whole compound is, and Massa has, like, an equally amazing compound there.

他正在举办一场晚宴。

He's hosting a dinner.

晚宴上有一位嘉宾叫西蒙·萨加斯，他早在1991年ARM刚分拆出来时就加入了公司。

One of the attendees at the dinner is Simon Sagars who had joined ARM back in 1991 right after the spinout.

他当时是工程师，也是公司的第十六号员工。

He was an engineer and he was the sixteenth employee.

这些年来，他一路晋升，现在成了CEO，他的角色更像是这个生态系统的守护者——人人都在用ARM，而软银作为全球最大的电信运营商之一，是至关重要的合作伙伴。

At this point he's risen through the ranks over the years, he's become the CEO and he's kinda like you know his job is like steward of this ecosystem and like obviously everyone uses ARM but like SoftBank is an important partner because they're one of the world's biggest telecom operators.

所以他希望软银能真正意识到，ARM芯片在它们销售的手机里有多关键。

And so like he wants SoftBank to, you know, make sure that they understand how important ARM is in the chipset of the phones that they're selling.

好吧，挺好的。

It's like, okay, great.

我去参加这个晚宴。

I'll go to this dinner.

我去跟马塞聊一聊。

I'll, you know, hang out with Masa.

大家坐下来吃晚饭，马塞就开始专门问西蒙，问ARM是做什么的，商业模式是怎样的，芯片用在哪些设备上。

So they're sitting there at dinner and Masa starts like focusing on Simon and he starts asking him a bunch of questions about you know what ARM does, what the business model is, know all the various devices that ARM chips end up being used in.

马塞就问：‘你们到底支撑了哪些东西？’

So Mass is like, So what exactly do you guys power?

西蒙回答：‘我们支撑的可多了，你知道的。',

And he's like, Well we power everything, you know.

不只是手机，还有汽车、咖啡机、冰箱、可穿戴设备。

We're not just in phones like we're in cars, we're in coffee makers, we're in refrigerators, wearables.

现在我们也开始进入服务器领域了，因为功耗问题，超级计算机的性能受限于电力，所以他们开始用我们的技术。

We're starting to get into servers like actually because of this power consumption issue like supercomputers like power is a limiting factor on performance so they're starting to use us.

Masa觉得这挺有意思。

And Maso's like interesting.

接着，Masa问他，就是现在广为人知的那个问题：如果钱不是问题，你会怎么做？

So then Masa asked him what, you know, now has famously become the the question that he asks every Vision Fund investment which is what would you do if money were no constraint?

Simon坐在饭桌上，想了想说：我想，我们还是会做同样的事，只是更快一点。

And Simon's just sitting there at dinner and, he's like, well I guess we'd probably, keep doing everything but faster.

Masa说：明白了。

And, Masa's like, okay.

Simon回到剑桥，回到英国，回到ARM，和刚才的环境截然不同。

So Simon goes back to Cambridge, back to The UK, to ARM, very different from, the environment he was just in.

几天后，他坐在剑桥的办公室里，接到了Masa从东京打来的电话。

He's sitting there a couple days later back in Cambridge and he gets a call from Masa in Tokyo.

按Masa一贯的风格，他说：我马上要见你。

And in typical Masa fashion, he's like, I need to see you right away.

不只是你，我还得见ARM的董事长Stuart Chambers。

And not just you, but I also need to see ARM chairman Stuart Chambers, chairman of the

这正是戴维报道中最精彩的部分。

And this is the best part of David's reporting here.

太棒了。

This is great.

上一期节目里我已经提到过这一点。

Well, alluded to this in the carve out on the last show.

有个问题，斯图尔特正在度假，他正坐在一艘游艇上，航行在土耳其的地中海海域。

So there's one problem which is that Stuart is on vacation and he's on a yacht in the Turkish Mediterranean.

但以马塞的样子，游艇在地中海这种事根本不算事儿。

But like Masa does it like Yachts in the Turkish med like that's that's his ballgame.

这种事根本难不倒他。

Like that doesn't faze him.

他就说，行。

He's like, okay.

等等，你稍等一下。

Well, you know, hang on.

我给他打个电话。

I'm gonna call him.

他给斯图尔特打电话，说：你得把船停到土耳其的梅尔梅，那是个位于地中海沿岸的度假小镇。

He calls him and he's like, I need you to dock your boat in Merme, Turkey, which, is like a resort town on the Mediterranean Coast in Turkey.

马萨随后派了一架私人飞机去剑桥接西蒙。

Masa then sends a private jet to Cambridge, picks up Simon.

他自己也登上了私人飞机。

He jumps on his own jet.

他们最后都在梅尔梅汇合了。

They all converge in Marmee.

我不确定是法语发音的‘梅尔梅’，还是土耳其当地叫‘马尔梅’。

I think I don't know if it's Marmee French pronunciation or Marmese in Turkey.

马萨包下了一整家餐厅，正对着码头，风景特别好。

Masa buys out an entire restaurant with, like, a view overlooking the marina.

那时候里面一个外人都没有。

So there's nobody else in there.

三个人在餐厅里碰头了。

They all converge, the three of them in the restaurant.

他们坐下后，马萨说：我想把你们整个公司买下来。

And they sit down and Masa's like, I wanna buy you guys.

我给你们开价三千二百亿美元，这比他们当时股价高出将近百分之五十。

And I'm gonna offer you $32,000,000,000, which was a almost a 50% premium to where they were trading at that moment in time.

我想

And I wanna

现在就谈妥。

do it now.

你知道，如果是两家上市公司之间的收购，通常溢价在百分之二十五左右。

You know, like, when there's a public to public acquisition, it sometimes is, like, 25%.

一般情况下，我们见过最高到百分之三十的溢价。

We've seen up to 30% in in in, you know, more typical situations.

但这次我觉得溢价达到了百分之四十三。

But like, I think it's a 43% premium.

是的。

Yep.

通常来说，上市公司之间的收购都要来回谈很久，还得有投行掺和，像Instagram这种干脆利落的交易，简直少之又少。

Typically with public to public acquisitions too, there's a lot of back and forth and there's investment bankers involved and like, you know, it's very rare that you get an Instagram style like done in there.

两周就搞定？

Closed it in two weeks?

是的。

Yep.

他们两周内就完成了，宣布了董事会批准，还同时公布ARM将继续作为软银旗下的独立部门运营。

They closed into they announced they got board approval and they announced within two weeks ARM to remain an independent division within SoftBank.

软银将支付三千二百亿美元收购整个公司。

SoftBank's gonna pay $32,000,000,000 to buy the whole thing.

市场对此非常看好。

The market, loves it.

股东一通过，交易就正式完成了。

It gets shareholder approval and it's done.

正如我们开头提到的，这个模式就是这样。

This as we alluded to at the top of the show, it was this model.

我敢说马塞当时已经在考虑愿景基金了，但这笔交易，还有‘计算无处不在’的愿景，真正成了软银如今通过愿景基金所做一切的原型。

I'm sure Massa was already starting to think about the Vision Fund but this deal and this vision of computing everywhere and what it'll enable becomes really the prototype for what SoftBank is now doing with the Vision Fund.

以至于到了第二年，也就是2017年，当愿景基金正式成立时，软银公司就把持有的ARM百分之二十五股份，按原价转让给了愿景基金。

So much so that the next year in 2017 when they actually get the fund set up, SoftBank corporate sells a 25 stake in ARM at cost to the Vision Fund.

于是愿景基金一上来就拿出八百亿美金，从软银公司手里买下了ARM百分之二十五的股份。

So the Vision Fund takes $8,000,000,000 right off the bat and buys 25% of ARM from SoftBank corporate.

那你觉得这背后是为什么呢？

So what do you think went on with that?

为什么会这样？

Why?

我觉得马塞从一开始就把这当作一项投资来看待。

Well, I think Masa was always thinking about this as an investment.

只是在愿景基金成立之前，唯一能做投资的方式，就是通过软银的资产负债表。

It was just that before the Vision Fund, only way you had to invest was through SoftBank balance sheet.

是的。

Yeah.

现在有个倒计时了，因为愿景基金的存续期是十二年。

I mean now there's a ticking time clock because the Vision Fund has a twelve year fund lifetime.

所以他不可能无限期地持有下去了。

So it's not like he can hold it indefinitely now.

迟早他们得把股份卖回给软银——但以马塞的立场，他肯定不想这么做。他更可能在未来五到七年里，让ARM获得不错的回报，然后卖给别的公司。

At some point they have to either sell that back to SoftBank, which I'm sure Masa doesn't want to do as someone who controls both entities, probably wants to in the next five to seven years get a nice return on ARM and sell it to someone else.

所以这事还没完呢。

So we haven't seen the last of this yet.

最有可能的情况是，这也是大家在转让之后开始讨论的：重新上市。

Well, think most likely and this is what people started talking about after transfer

不是20%的IPO，而是5%。

Not the 20 IPO 5%

把ARM转入愿景基金之后，最可能的就是让它重新上市。

transfer into the Vision Fund is re IPO it.

我觉得这是最可能的结果。

I think that's the most likely outcome.

但有意思的是，这也反映了愿景基金的策略。

But what's interesting and this also speaks to the Vision Fund strategy.

所以2016年软银收购这家公司的时候，公司里大概只有4000名员工，一方面这人数确实不少。

So in 2016 when SoftBank acquired the company there were only about 4,000 people working at the company which on the one hand is a lot of people.

但另一方面，这比优步少多了，也比现在的爱彼迎少多了。

On the other hand like that's a lot less than Uber, that's a lot less than you know I think less than Airbnb at this point.

像一家这么老牌的公司，它当时简直就是所有技术的基石。

Like for a company that old and that was like literally the foundation of you know, all technology at that point.

我的意思是，我觉得它在智能手机和嵌入式设备领域的市场份额高达96%。

I mean I think they have a 96% market share of all smartphone and embedded devices, embedded devices.

这可不是小数目。

That's not a lot of people.

从那以后，在两年到两年半的时间里，ARM招聘了超过2000名员工。

In the two plus two and a half years since then ARM has hired over 2,000 people.

自从那以后，他们的人数增加了超过百分之五十，要是上市公司根本做不到这点，因为他们现在亏钱了。

So they've grown by more than 50% headcount since they couldn't have done that if they were a public company because they're now losing money.

他们现在的净利润是负的。

They're now net income negative.

不过马萨，说到底，这是你的问题——如果钱不是问题，你会怎么做？

But Massa, again, you know, it's his question, what would you do if money were no object?

他们是净利润为负吗？

Are they net income negative?

我刚才看了一下相关数据。

So here's here's what I was looking at that.

他们的运营利润率是百分之五十二。

Their operating margin was 52%.

收购时他们的运营利润率是百分之五十二，现在降到了百分之二十四。

Their operating margin at the time of acquisition, their operating margin now is 24%.

尽管收入从十六亿美元增长到了十八亿美元，他们的EBITDA却下降了百分之四十。

Even though they've grown from $1,600,000,000 in revenue to $1,800,000,000 in revenue, their EBITDA dropped 40% despite the revenue increase.

所以我不确定他们是不是真的亏钱，这挺有意思的。

So I'm not sure if they're losing money, they that's interesting.

他们卖出了更多的设备，但每台赚的钱却没多多少，利润大幅下滑了。

They're selling a lot more devices that only make a little bit more money and their profits significantly dropped.

有意思。

Interesting.

他们的利润率也下降了。

And their margin dropped.

我刚又查了一遍。

I just pulled it up again.

他们发布过一篇报道，说2018年亏了两亿美元，不过原因可能有很多。

Said there's an article that they posted a loss of 200,000,000 in I believe 2018, which could be for lots of reasons.

这显然不是运营利润率，而是净利润。

And obviously that's not operating margin, that's net income.

他们现在对英特尔投资者强调的是：别急，我们正在大力投资增长，未来一定会回报的。

The drum that they're beating Intel investors right now is, hold on guys, we are way investing in growth and it will pay off.

等真发生了，我们再采取行动。

And then we're gonna do something about it when it does.

有趣的是，他们还在定期发布季度业绩报告。

And what's interesting, they still do quarterly presentations on company results.

他们现在重点谈的是如何大力投资下一代计算技术。

What they're talking about now is how they're investing heavily into the next computing waves.

ARM 已经开始设计专门用于人工智能场景，以及自动驾驶等用途的芯片。

So ARM has started now making designs and devices chips that are dedicated for AI use cases but also for like autonomous vehicle use cases.

你想想，英特尔当初花了大约一百七十亿美元收购了 Mobileye，而 Mobileye 专门生产汽车芯片，尤其是用于驾驶辅助和未来完全自动驾驶的计算机视觉系统。

So you think about Intel bought Mobileye for I think $17,000,000,000 and Mobileye makes chipsets for cars and in particular for computer vision for driver assistance and eventually fully autonomous use cases.

ARM 也在大力进入这个市场，同时还覆盖了各种嵌入式计算领域。

ARM is getting heavily into that market as well as well as all sorts of embedded computing.

这确实挺有意思，因为当初收购时，他们给出的理由是——等会儿我们还会快速过一下节目的其他部分。

Yeah it's interesting because the stated reason at the time of purchase, and at some point here we'll breeze through the other sections of the show.

但当时他们强调的是物联网，说物联网正在爆发，因为未来 ARM 芯片不再只出现在手机里，还会出现在你周围所有联网的设备中。

But at the time of purchase the drum they were beating was its IoT that, the Internet of Things is blowing up and because, you know, there's going to be an ARM device in much more than your phone, it's going be in all these other things that are communicating with the Internet around you.

所以我们相信，增长确实像孙正义所认为的那样，存在一些未被市场定价的潜力。他认为物联网设备的市场会比公众投资者——那些持有ARM股票的人——所预期的还要大得多，至少有理由相信它未来能产生更多的现金流。

That's why we believe that the growth is it's like Masa believed that there was some growth that was not priced in, that he believed was going to expand, the market for IoT devices was going to expand even greater than the public market investors who owned ARM believed, or at least that there would be some reason to believe that it would be generating more future cash flows.

当时他们反复强调的是物联网，物联网，物联网。但两年后的今天，回头看他们的一些投资者关系材料，发现重点其实已经转向了智能汽车和AI芯片。

And it was IoT, IoT, IoT, and it's interesting to see two years later now on some of their investor relations stuffs that it really is more about the connected car and AI chips.

当然，他们出货的设备数量依然是一个非常亮眼的故事。

So certainly their devices shipped has continued to be a really great story.

他们从2015年的150亿台，增长到我印象中2020年底的时候——

They went from I think 2015, they were 15,000,000,000 and now in I think at the end of twenty,

2017年就已经超过了210亿台。

seventeen they did over 21,000,000,000.

对，我看到的数据也是这样。

Yeah, that's what I'm looking at too.

指数级增长依然非常出色，但有意思的是，你得相信什么样的具体假设，才能认为低功耗ARM芯片的需求会持续增长？

Exponential growth continues to be really excellent but it is interesting, it's like on what thesis do you have to believe in a specific thesis about where there will be more CPUs, specifically low power ARM CPUs?

还是说，你干脆就不需要相信任何特定的假设，直接看数据就行？

Or do you just say, look, I don't have to believe anything in particular.

我很有把握，这些东西会越来越无处不在。

I'm just pretty sure that these things are gonna continue to be more and more everywhere.

是的。

Yeah.

挺有意思的。

It's funny.

我属于后者，不过我们要不要转到收购这个话题？

I'm in the latter camp, but should we jump to acquisition category?

是的。

Yeah.

真有意思，我们本来一直在为上市讲故事，结果突然就变成被收购了。

It's so funny that we were setting this up all along for an IPO narrative, but boom, acquisition.

因为我们最近做了好多起收购。

Because we've done so many recently.

我知道。

I know.