第48集：高通与博通

你在喝威士忌吗？

Are you drinking whiskey?

不。

No.

茶。

Tea.

我当时就想，这真是

I was like, that's

一大杯威士忌。

a big that's a big cup of whiskey.

欢迎回到《Acquired》第48集，这是一档关于科技公司收购与IPO的播客。

Welcome back to episode 48 of Acquired, the podcast about technology acquisitions and IPOs.

我是本·吉尔伯特。

I'm Ben Gilbert.

我是大卫·罗森塔尔。

I'm David Rosenthal.

我们是你们的主持人。

And we are your hosts.

今天，我们将讨论有史以来最大的科技收购案——博通收购高通。

Today, we are covering what would be the biggest tech acquisition of all time, Broadcom acquiring Qualcomm.

在我们录制时，这笔交易的最新进展是今天被否决了，交易估值为1030亿美元，不包括交易中涉及的债务。

And as we record, the current status of the deal was it was rejected today, for a, a price of a $103,000,000,000, not including the the, or I guess accounting for the debt that is part of the deal.

大卫，真期待看到这件事如何在我们面前展开。

And, it'll be really fascinating to see how this how this unfolds in front of us, David.

是的。

Yeah.

通信领域的较量。

The battle of the comms.

是的。

Yeah.

一比二。

One m versus two.

谁会赢？

Who will win?

让我想起了3Com。

Reminds me of 3com.

是的。

Yes.

这个行业的命名真是极其富有创意。

It's incredibly incredibly creative names in this industry.

是的。

Yeah.

嗯，我认为今天我们将会探讨的一个主题是：旧的又回来了。

Well, I think, one of the themes we'll get into today is what's old is new again.

所以，就像我们回到了公元2001年一样。

So, you know, it's like we're back in the year thousand and one here.

好的。

Okay.

对于那些可能不太了解半导体公司的听众来说，这两家公司本质上都是无晶圆厂的半导体公司。

So for for listeners who are, maybe not as, as deep in, at in semiconductor companies and, wondering sort of what these companies do, they're both effectively fabless semiconductor companies.

所以它们都没有晶圆厂，而晶圆厂是目前实际制造芯片的设施，除了可能的英特尔、三星和台积电。

So they neither of them have fabs, which are the fabrication facilities that actually manufacture the chips these days with the exception of maybe Intel, Samsung, TSMC.

大多数芯片设计公司实际上并不制造芯片，因为建造晶圆厂的成本太高了。

Most of the chip designers are not actually the the chip manufacturers because it's so expensive to create them.

因此，这两家公司负责设计，然后与合同制造合作伙伴合作，生产手机和其他计算设备内部的芯片。

So both of these companies design and then work with contract manufacturing partners to manufacture the chips that go inside your phone and other computing devices.

它们的产品涵盖从处理器本身到无线收发器的所有组件，我们将在节目中进一步探讨。

And, they they make everything from the actual processors themselves to wireless radios, and and we'll get into that more in the show.

但本质上，它们是手机、电脑、服务器等设备的零部件制造商。

But it's basically component makers for for phones, computers, servers, etcetera.

还有你的汽车、烤面包机，如今几乎一切设备都是如此。

And your car, your toaster, your everything these days.

没错。

That's right.

物联网的爆发。

The IoT explosion.

在深入讨论之前，先提醒大家几点。

Well, before we get too much into it, a couple of, of quick reminders.

我们有一个Slack群组，成员已超过一千人。

We've got a Slack where we are over a thousand strong.

如果你喜欢讨论并购、IPO、科技新闻，或者只是想再加一个Slack群组——毕竟谁会嫌Slack群组太多呢？

So if you if you like to talk about m and a, IPOs, tech news, or just add another Slack because lord knows we cannot be in in too many of those.

请访问 acquire.fm。

Go to acquire.fm.

你可以在侧边栏或移动设备底部加入。

You can join in the sidebar or on mobile down at the bottom.

另外，我们非常欢迎评价。

The other thing is we love reviews.

如果你喜欢这个节目，并认为其他人也会喜欢，请现在暂停节目。

So if, if you like the show and you think other people would like it too, pause the show right now.

你可以随时回来，在苹果播客上留下一个简短的评价，我想可以通过 iTunes 商店来操作。

You can always come back to it, and and go leave a quick review on, on Apple Podcasts, and you can do that through, I think, through the iTunes store.

好的，各位听众。

Alright, listeners.

我们的下一个赞助商是节目的新朋友——Rippling。

Our next sponsor is a new friend of the show, Rippling.

我先直接引用他们的投资者信函，因为我觉得这封信非常出色且清晰明了。

I'm just gonna start by quoting their investor letter because I think it is excellent and clarifying.

Rippling 的一个核心洞察是，大多数商业系统都包含大量关于员工的信息。

Rippling's one underlying insight is that most business systems are full of information about employees.

大家都知道人力资源系统确实如此，但我们也知道，这不仅限于人力资源部门。

Everyone knows that's true for HR systems, but we know that is true beyond the HR department as well.

我们认为，员工数据不仅仅是人力资源部门的专属领域，它实际上是商业软件的基本要素，尤其对于那些远超出人力资源范畴的商业软件而言。

We think employee data isn't just the domain of the HR department, it's the fundamental primitive of business software, including and most especially for business software well outside of HR.

这真是一个绝佳的论点。

That is a great thesis statement.

Rippling 正是如此，一个基于完全不同于其他任何产品的架构构建的全球人力资源、IT 和财务统一平台。

And Rippling is exactly that, a unified platform for global HR, IT, and finance built on a totally different architecture than anything else.

这就是员工图谱。

It's the employee graph.

大多数一体化的人力资源系统实际上并不是一开始就一体化的。

Most all in one HR systems didn't actually start as all in one.

它们最初只是薪酬供应商，然后通过添加新产品或收购来扩展功能。

They started as payroll vendors, then bolted on new products or acquisitions.

在底层，它们是一堆孤岛式工具，用脆弱的集成方式勉强拼凑在一起。

Under the hood, they're a patchwork of siloed tools duct taped together with brittle integrations.

每次有变化时，你都得手动更新五个不同的系统，或者完成一个二十步的检查清单。

Anytime something changes, you're stuck manually updating five different systems or working through a 20 step checklist.

Rippling 从第一天起就以员工图谱为核心构建了整个系统。

Rippling instead built their system from day one as the employee graph.

它是一个实时反映你整个员工队伍的知识图谱。

It's a real time knowledge graph of your entire workforce.

每位员工、职位、权限、设备、应用、地点和薪酬计划都完全同步，并集中在一个地方。

Every employee, role, permission, device, app, location, and compensation plan, totally in sync and all in one place.

如果莎拉获得晋升并从纽约搬到加利福尼亚，Rippling 会自动更新她的薪资税务、开通新的应用权限、寄送新笔记本电脑、发放新的公司信用卡，并指派所需的管理者培训。

So if Sarah gets promoted and moves to California from New York, Rippling updates her payroll taxes, provisions her new app permissions, ships her a new laptop, issues a new corporate credit card, assigns required manager training, all automatically.

你开箱即得三十多种原生系统。

You get 30 plus native systems out of the box.

人力资源、IT、财务、全球薪资、设备管理、公司信用卡、账单支付，可整体使用或按需选择。

HR, IT, finance, global payroll, device management, corporate cards, bill pay, altogether or a la carte.

通常需要在四个不同工具和三个部门间来回流转的工作流程——如入职、晋升、权限管理、支出审批——现在都能在一个地方自动完成。

So workflows that normally bounce across four different tools and three departments, onboarding, promotions, access management, spend approvals, they all just happen in one place automatically.

这就是为什么 Rippling 的客户可以用同样的团队支持两倍数量的员工，也正因为如此，他们在 G2、TrustRadius 和 Gartner 上被评为排名第一的人力资本管理套件。

That's why rippling customers can support twice the number of employees with the same team and why they're the number one rated human capital management suite on g two, TrustRadius, and Gartner.

如果你、你的公司或你投资组合中的公司希望以员工为中心，在一个统一的平台上运行业务核心，欢迎访问 rippling.com/acquired，并告知他们 Ben 和 David 推荐了你，或直接点击节目笔记中的链接。

So if you, your company, or your portfolio companies wanna run the backbone of your business on one unified platform with people at the center, head to rippling.com/acquired and tell them that Ben and David sent you, or just click the link in the show notes.

好的。

Alright.

那么，大卫，你准备好带我们进入正题了吗？

And with that, David, are you ready to, to take us in?

当然。

Yeah.

但在我们深入今天的历史和事实之前，我觉得有些背景信息虽然可能与科技主题相关，但我想先提出来。

But before we get into history and facts today, I think there's some stage setting that'll probably come back in tech themes but, but I wanna do upfront.

我们刚才还在开玩笑，说硅谷现在流行‘旧物复兴’，而那个旧的东西就是‘硅’。

We were joking a minute ago about what's old being new again in in Silicon Valley right now and and the old that is new is the Silicon in Silicon Valley.

显然，目前有一笔交易——博通收购高通，可能是有史以来规模最大的科技收购案，但这远非半导体领域唯一正在发生的事。

There's there's obviously this deal which is, potentially going through largest technology acquisition of all time Broadcom buying Qualcomm but that's far from the only thing happening in the semiconductor world right now.

而且，说来好笑的是，硅谷——半导体的起源和创新之地，确实就在旧金山以南、圣何塞周围的区域。

Well, and hell and hilariously, Silicon Valley, I mean, the the, genesis and innovation of semiconductors is in in Silicon Valley, you know, the area South Of San Francisco around San Jose.

但总部位于新加坡的博通和位于圣地亚哥的高通，都不在

But Singapore based Broadcom and San Diego based Qualcomm are neither in in

硅谷。

Silicon Indeed.

在真正的硅谷以及精神上的硅谷，世界各地都发生了许多事情。

There's just a lot going on in true Silicon Valley and in in the rest of the world, Silicon Valley in spirit.

所以我想，也许我们可以回顾一下几件事，这取决于你们是否喜欢这种方式。

So I thought maybe we'd review just a couple things and this might be depending on how you guys like this.

也许这会成为我们在《Acquired》节目中深入探讨的另一个主题，即半导体领域正在发生什么，以配合我们之前关于旅行、体育等的迷你系列。

Maybe this will become another theme that we dig into here at Acquired which is what's going on in the semiconductor world to go with our sort of mini series that we have on travel and sports and other things.

但这一切其实早在几年前就开始了，就像许多事情一样，源于机器学习的兴起。

But it all started really a couple of years ago as with many things with the sort of boom in machine learning.

如果我们看一下一家名为英伟达的公司，它也是另一家芯片制造商。

If we could look at a company called NVIDIA, which is another another chip maker.

长期以来，他们主要生产用于个人电脑、游戏主机和工作站的显卡。

And for a long time they've made just graphics cards for PCs and video game consoles and workstations.

因此，如果你是个游戏玩家，你肯定知道英伟达，也许还持有他们的股票。

And so if you were a gamer, you definitely knew what NVIDIA was and maybe you own some of their stock.

但近年来，随着GPU在机器学习中的广泛应用，英伟达基本上经历了一段疯狂的增长期。

But then with GPUs over the last couple couple years becoming so prevalent in machine learning, NVIDIA has basically been on this crazy tear.

所以如果你回看2015年，当时他们的股价是每股20美元。

So if you go back to 2015, they were trading at $20 a share.

你和我为了准备这期节目做了同样的计算。

You and I did the same math to prepare for this episode.

我有

I I have

2016年1月，股价为每股26美元。

January 2016 at $26 a share.

对。

Yeah.

你知道，我们开始前从不提前对齐内容。

This is you know, we don't compare notes before we start here.

我们保持对话的真实自然。

We're keeping the conversation real.

所以两年前，英伟达的股价是每股20美元。

So two years ago, NVIDIA's at $20 a share.

现在，它的股价是每股212美元。

Today, it's at $212 a share.

所以两年内涨了十倍以上。

So that's over a 10 x in two years.

我的意思是，这种利润率是风险投资人梦寐以求的，而这是在公开市场上实现的。

I mean, this is like VCs would kill to have that kind of markup, and this is in the public markets.

事实证明，显卡的用途远不止图形处理。

It turns out, you know, graphics cards are good for more than just graphics.

谁想得到呢？

Who knew?

线性代数、矩阵变换，你知道的，特别适合游戏和渲染，而恰好这些技术也是如今每一种新技术的核心——机器学习。

Linear algebra, matrix transforms, you know, really good for for games and for rendering and it just so happens for the core technology that's a piece of every new technology now, machine learning.

没错。

Yep.

没错。

Yep.

所以你有英伟达，这家公司正以一种非常好的方式蓬勃发展。

So you've got NVIDIA which is on fire in a good way.

你还有一家名为Nirvana的公司，它是一家专注于GPU和机器学习的专用芯片制造商和设计公司，是一家初创企业，后来被英特尔以3.5亿美元收购。

You've got a company called Nirvana which was a another GPU and and machine learning kind of dedicated chip manufacturer, chip designer that was a startup that was acquired by Intel for $350,000,000.

在很多方面，我认为这标志着半导体领域初创企业重新兴起的开端，而这种情况已经很久没有发生了。

And that was kinda in many ways I think the start of a renaissance in in terms of actual startups getting started in the in the semiconductor world, which hadn't happened in a long time.

就在本周，红杉资本领投了总部位于英国的Graphcore公司5000万美元的融资，该公司也生产专用于机器学习和深度学习应用的专用芯片。

So just this week, Sequoia led a $50,000,000 round in a company called Graphcore out of The UK, which also makes specialized chips for for machine learning and and deep learning applications.

令人难以置信的是，这家公司才成立大约一年。

And what's crazy, this company has been around for only about a year.

红杉资本刚刚投资了5000万美元。

Sequoia just invested $50,000,000.

他们此前已经筹集了6000万美元。

They had already raised $60,000,000.

因此，这家公司在短短一年多的时间里筹集了1.1亿美元，以在如今这个竞争激烈的市场中与众多初创企业竞争，这些企业都在努力开发专用的机器学习芯片，以挑战英伟达。

So this company has raised a $110,000,000 in just over a year to compete in what's a very crowded market now of lots of startups, going out there trying to compete with NVIDIA, make specialized machine learning chips.

这里值得暂停一下。

It's worth pausing for a moment here.

是什么推动了这种创新？

Like, what's enabled that that innovation?

除了市场对机器学习等新型计算的需求外，硅行业中有少数公司整合并掌握了大部分晶圆厂，这也有其合理性。

I Aside from the market demand for new types of computing with with machine learning, the need for just a few companies to consolidate and have most of the fabs in the silicon industry made sense.

这已经持续了一段时间了，我不知道，大概十年、二十年、三十年了吧。

You know, it it's been around for a while, I don't know, ten, twenty, thirty years.

这是因为当我们缩小了晶圆上所需的纳米尺寸时——我说得不太准确。

And that happened because as we shrunk the the number of nanometers that were necessary or that we could we could put on a wafer, not being very eloquent there.

但基本上，随着我们在给定晶圆上集成越来越多的晶体管，制造这些芯片的能力变得越来越昂贵。

But, basically, as we got more and more sophisticated at, putting more and more transistors on a given wafer, it became more and more expensive to produce the the, ability to manufacture those.

因此，只有少数几家公司实际从事制造，因为建造一座这样的晶圆厂需要超过一百亿美元的投资。

So only a few companies are actually doing the manufacturing because it costs 10,000,000,000 plus dollars to to create one of these fabs.

于是，突然间，你有了一个平台，无需自己投入巨额资本进行制造，就可以成立一家小公司，专注于芯片设计，要么授权你的设计，要么与这些制造商合作。

So suddenly, you have this platform where you don't need the actual capital to to do the manufacturing yourself, so you can start a smaller company to just do designs, to either license your designs, work with these manufacturers.

这实际上让我想起，有趣的是我们在谈论科技主题之前还没谈历史，但这让我联想到杰夫·贝佐斯曾说过，如果没有电网、UPS、互联网以及互联网之前的所有电话公司——那些为创新搭建平台的先行者，亚马逊根本不可能存在。

It it actually reminds me it's funny we're doing tech themes before we even talk history, but it reminds me a lot of of Jeff Bezos talking about how Amazon wouldn't be possible if it weren't for the electric grid and for UPS and for the Internet and for everyone, the phone companies before the internet, everyone that came before to basically build platforms that you could do innovation on top of.

没错。

Yep.

是的。

Yes.

我会说没错，但这类初创公司的特别之处在于，它们处于两种极端之间：过去创办半导体公司需要大量资金来建造晶圆厂，而软件公司则不需要。就像我们看到的Graphcore这家公司，它们正在筹集巨额资金，我认为这是因为设计这些芯片需要大量工程师和巨额技术投入。

I would say yes but what's interesting about this class of startups is they kind of fall somewhere between the amount of capital that you used to need to create a semiconductor company when you were fabbing the chips and like a software company in that as we're seeing with this company Graphcore, they're raising a ton of money and I think that's because you need so many engineers and so much just technology investment going into working on these chip designs.

我的意思是，这些芯片的复杂程度极高，因此需要大量的资金投入。

I mean these things are so massively complex, that it just takes a ton of investment into them.

而且这还只是技术投入，别忘了销售方面的投入，对吧？

And that's also not to mention sales investment, right?

因为现在有太多公司正在争夺这一细分市场。

Because there's so many companies out there now competing for this segment.

你看到机器学习热潮正在兴起，大量初创公司和大公司都在这一市场领域追逐增长。

You've got the machine learning sort of craze going on and all of the growth that's being chased by lots of startups and big companies there in one corner of the market.

然后，本，正如你提到的，在市场的另一个角落是晶圆厂本身。

Then Ben as you mentioned in another corner of the market, have the fabs themselves.

那就是台积电，我认为它是最大的，然后是三星和英特尔，它们是全球主要的晶圆厂。

So that's TSMC, which I believe is the biggest and then Samsung and Intel are the major fabs out there in the world.

而台积电的市值已经超过2000亿美元。

And and TSMC is over a $200,000,000,000 market cap company.

正如你所说，这仅仅是纯粹的、为他人设计的量产，包括许多这些由机器学习驱动的公司。

And and as you said, like, this is just pure commodity production of designs that other people are are making and including many of these machine learning driven companies.

所以这是市场的另一个角落，随着对它们生产的芯片需求不断增长，这些公司也在扩张。

So that's in another corner of the market and those companies are growing, as all this demand is coming online for chips that they're producing.

然后，市场上还有第三个角落，这正是我们今天要探讨的内容。

And then you've got this third corner of the market and that's what we're gonna explore today.

这些公司在很大程度上是老牌的大型芯片设计公司，在过去十年移动时代兴起时经历了大幅增长。

And these are kinda in many ways legacy large chip designers that really had had a big run up over the last ten years as the mobile era got installed.

你知道，博通和高通这两家公司都生产了大量芯片，几乎进入了每一部手机。

You know both of these companies Broadcom and Qualcomm specifically were making tons of chips that were going into just about every mobile phone out there.

但发生的情况是，苹果开始将大量芯片设计内部化，正如我们在PA Semi和Authentic那一集讨论的那样，以至于我现在认为iPhone和iPad中的几乎所有芯片都已经是苹果自主研发的了。

But what's happened is that you have Apple which has started to bring a lot of their chip design in house as we talked about on the PA semi and authentic episode, to the point where I believe almost pretty much almost every chip within an iPhone and an iPad is is Apple designed in house at this point.

啊，不是的。

Ah, no.

并不是这样的。

It is not.

不是。

No.

哦，有意思。

Oh, interesting.

这倒是个好话题，我来岔开一下，因为当我试图自己弄清楚高通和博通到底是做什么的时候，我查阅了iPhone X和Pixel 2 XL的物料清单和拆机分析。

This is a good I'll take a little segue here because I, as I was trying to understand myself exactly what is Qualcomm and what is Broadcom, I went in and looked at the, basically, bill of materials and the teardowns for each of the iPhone X and the Pixel two XL.

呃。

And Ah.

事实上，令人惊讶的是，这些公司生产的元件不仅大量出现在安卓手机中，也出现在iPhone里。

It is actually pretty surprising how many components are made by these companies that are in not only Android phones, but but iPhones.

所以，真的吗？

And so Really?

虽然我们都清楚苹果为A11处理器以及整个A系列处理器设计了芯片，还有运动协处理器，我认为他们现在也自研了音频和数字信号处理器。

While while we all know that Apple creates their own design for the the a 11 processor or the whole a series processors, the motion coprocessor, the I think they have their own audio, the digital signal processor now.

因此，苹果设计了所有这些组件，但交由台积电和三星进行合同制造。

So Apple's designing all these things, and they're having them contract manufactured by TSMC and Samsung.

实际上，iPhone中包含大量高通和博通的元件。

There's actually plenty of Qualcomm and Broadcom components in in the iPhone.

在iPhone 10内部，高通制造了千兆级LTE收发器。

So inside the iPhone 10, Qualcomm makes the gigabit LTE transceiver.

他们还制造了LTE调制解调器，这实际上是骁龙系列的一部分，而苹果在这两者之间双源采购，即英特尔和高通。

They also make the LTE modem, which is actually part of the Snapdragon series, and Apple dual sources this between Intel and Qualcomm.

所以他们会轮换使用，一些批次的手机用英特尔的，另一些则用高通的。

So they they sort of switch off which, some batches of phones get an Intel, some some get the the Qualcomm.

iPhone 10还使用了博通的芯片，用于无线充电、功率放大器模块和触摸屏控制器。

The iPhone 10 has, Broadcom chips for wireless charging for the power amplifier module and for the touchscreen controller.

所以所有这些东西都是博通的。

So all those things are are Broadcom.

而在由LG制造的Pixel 2 XL以及HTC One（其实就是Pixel 2）中，高通制造了多种组件。

And in the Pixel two XL, which is actually manufactured by LG, and the the HTC one is just the the Pixel two, Qualcomm makes a variety of components.

他们制造了骁龙处理器，这可以说是苹果A11仿生芯片的对应产品。

They make the Snapdragon processor, which is the, sort of counterpart to Apple's a 11 Bionic chip.

高通还制造了千兆LTE射频收发器、电源管理集成电路、快速充电集成电路。

They, Qualcomm makes the gigabit LTE RF transceiver, the power management integrated circuit, the quick charge integrated circuit.

有趣的是，我认为Pixel 2 XL中没有任何博通的组件，至少那些拆机分析的人没有发现。

And interestingly enough, I don't think there's any Broadcom components in the Pixel two XL, at least not that could be identified by people that are ripping the phone apart.

所以这里的关键是，iPhone 10和Pixel 2 XL都包含了大量高通和博通的零部件。

So the bottom line here is the iPhone 10 and the the the Pixel two XL both have a ton of Qualcomm and Broadcom parts in them.

我认为它们唯一不直接竞争的主要领域其实是主CPU本身。

And I think the only major area where they don't compete is actually on the the sort of main CPU itself.

它们在所有无线电、蓝牙、Wi-Fi等组件上都存在竞争。

They compete on on all the radios and all the the Bluetooth and the Wi Fi and all that stuff.

有意思。

Interesting.

有意思。

Interesting.

这表明了这个领域的复杂性。

Well, goes to show how complicated this space is.

是的。

Yeah.

每个人都是亦敌亦友。

Everybody is a frenemy.

每个人都是竞争对手。

Everybody's a competitor.

每个人都处于竞合关系中。

Everybody is in in coopetition.

是的。

Yeah.

而且这些设备中芯片的数量本身就令人震惊。

And even just the sheer number of chips in these devices is crazy.

好了，听众们。

Alright, listeners.

是时候谈谈另一家

It's time to talk about another one

我们

of our

最喜欢的公司，Statsig。

favorite companies, Statsig.

自从你们上次听到我们提到Statsig以来，他们有了一个令人兴奋的更新。

Since you last heard from us about Statsig, they have a very exciting update.

他们完成了C轮融资，估值达到11亿美元。

They raised their series c, valuing them at $1,100,000,000.

是的。

Yeah.

巨大的里程碑。

Huge milestone.

祝贺整个团队。

Congrats to the team.

而且时机很有趣，因为实验领域正在迅速升温。

And timing is interesting because the experimentation space is, really heating up.

是的。

Yes.

那么，为什么投资者将Statsig估值超过10亿美元？

So why do investors value STAT SEG at over $1,000,000,000?

这是因为实验已成为全球顶尖产品团队产品栈中的关键部分。

It's because experimentation has become a critical part of the product stack for the world's best product teams.

没错。

Yep.

这一趋势始于Facebook、Netflix和Airbnb等Web 2.0公司。

This trend started with web two dot o companies like Facebook and Netflix and Airbnb.

这些公司面临一个问题。

Those companies faced a problem.

如何在扩展到数千名员工的同时，保持快速、去中心化的产品和工程文化？

How do you maintain a fast, decentralized product and engineering culture while also scaling up to thousands of employees?

实验系统是这一问题的关键解决方案。

Experimentation systems were a huge part of that answer.

这些系统让这些公司的每个人都能访问一套全局的产品指标，从页面浏览量到观看时长再到性能。

These systems gave everyone at those companies access to a global set of product metrics, from page views to watch time to performance.

每当一个团队发布新功能或新产品时，他们都可以衡量该功能对这些指标的影响。

And then every time a team released a new feature or product, they could measure the impact of that feature on those metrics.

因此，Facebook可以设定一个公司级目标，比如增加应用内时长，然后让各个团队自行探索如何实现它。

So Facebook could set a company wide goal like increasing time in app and let individual teams go and figure out how to achieve it.

将这种模式扩展到数千名工程师和产品经理身上，就能实现指数级增长。

Multiply this across thousands of engineers and PMs and boom, you get exponential growth.

难怪实验现在被视为必不可少的基础设施。

It's no wonder that experimentation is now seen as essential infrastructure.

是的。

Yep.

如今，像Notion、OpenAI、Rippling和Figma这样的顶尖产品团队同样依赖实验机制。

Today's best product teams like Notion, OpenAI, Rippling, and Figma are equally reliant on experimentation.

但他们并不自己搭建，而是直接使用Statsig。

But instead of building it in house, they just use Statsig.

而且他们不仅仅用Statsig来做实验。

And they don't just use Statsig for experimentation.

在过去几年里，Statsig增加了快速产品团队所需的所有工具，比如功能开关、产品分析、会话回放等。

Over the last few years, Statsig has added all the tools that fast product teams need, feature like flags, product analytics, session replays, and more.

所以如果你希望

So if you would like

帮助你的团队工程师和产品经理更快地构建产品并做出更明智的决策，请访问 statsig.com/acquired，或点击节目说明中的链接。

to help your team's engineers and PMs figure out how to build faster and make smarter decisions, go to statsig.com/acquired, or click the link in the show notes.

他们提供非常慷慨的免费套餐、5万美元的初创企业计划，以及面向大型企业的实惠企业合同。

They have a super generous free tier, a $50,000 startup program, and affordable enterprise contracts for large companies.

只要告诉他们，是本和大卫介绍的。

Just tell them that Ben and David sent you.

好的。

Okay.

我们来聊聊高通和博通吧。

Let's get into Qualcomm and Broadcom a little bit.

这两家公司现在都可以说是‘混血’公司了，就像我们节目开始前开玩笑说的那样，‘嗯，没错’。

Both companies are are at this point sorta franking companies as we were joking before the show of Yep.

多年来它们进行了大量并购，以至于很难理清这团乱麻，追溯它们的起源。

Having done so much m and a over the years that it's hard to hard to even untangle the rat's nest and go back and figure out where they originated.

不过，关于博通，我觉得这一点挺有意思。

But real quick on Broadcom, I thought this was interesting.

去年，它们实际上与另一家公司艾维戈合并了。

They actually merged themselves with another company called Avago last year.

而如今被称为博通的这家公司，实际上就是艾维戈。

And actually the the company today known as Broadcom really is Avago.

他们采用了Broadcom这个名字，而Avago本身最早起源于惠普的半导体部门，之后经历了多次剥离和并购，最终变成了现在的样子。

They took the Broadcom name and Avago, started life itself way back in the day as HP's semiconductor division, has then been through a bunch of divestitures and mergers over the years and ended up, ended up here.

但这其实可以追溯到硅谷的起源，即惠普在车库里创业的年代。

But it goes really back to the the origin of Silicon Valley with Hewlett Packard in the garage.

是的。

Yeah.

想想看，如果他们没有把这部分业务剥离出来，会发生什么，这真的很有趣。

It's fascinating to think about what would have happened if they hadn't spun that out.

这或许可以成为一个有趣的话题，以后可以专门做一期。

That could be sort of a fun episode to do at some point.

但没错。

But Yeah.

你知道吗，如果惠普成了最大的智能手机组件制造商，会怎么样？

You know, what if what if HP was the largest smartphone component manufacturer?

是的。

Yeah.

他们现在可能比现状要好得多。

Well, they might be in a better place than they are now.

但这得留到下一期再说。

But that's for another episode.

不过高通，我们要多花点时间聊聊，因为我觉得这个故事特别有趣。

But Qualcomm, we'll spend a little more time on this, because I think this is such a fun story.

高通成立于1985年，由圣地亚哥大学（或加州大学圣地亚哥分校）的教授厄温·雅各布斯与其他几位曾在学术界和芯片行业共事的同事共同创立。

So Qualcomm was founded back in 1985 by University of San Diego or UC University of California San Diego professor Erwin Jacobs along with several other folks that he had worked with in the colleagues from from academia and in the in the chip industry.

而我觉得有趣的是，它实际上……

And it actually was, and this is what I think is fun about it.

它本质上是一家原始的网络效应公司，而高通最初围绕的网络效应正是电话网络。

It was actually like an OG network effect company And the network effect that Qualcomm was started around even though it became a semiconductor company was the original network effect of the telephone.

我的意思是，如果你回溯到贝尔电话公司时代，就会明白：随着越来越多的人加入电话网络并相互连接，网络的价值和壁垒也会随之提升。

And and what I mean by that is if you go back and and think about kind of the Bell Telephone Company and the idea of, you know, as you add more participants to the network of people who have telephones and are connected, then the more valuable the network becomes as you as you add more people and the more defensible it becomes.

而高通实际上无意中在蜂窝电话领域也实现了同样的网络效应。

Well, Qualcomm actually sort of stumbled into doing the same thing with cellular telephones.

我们都记得，或者很多人还记得，在2000年代那十年里，曾发生过运营商标准之争，比如GSM和CDMA之间的竞争。

So we all remember that or many of us remember there were sort of the carrier standard wars that were happening over the decade of the February and most of the 20 tens of, you know, GSM versus CMA.

我们之所以选择GSM，是因为我父亲当时深入研究了这个问题，并决定押注GSM。

We we went with singular specifically because my dad, like, was looking into this and and decided that he wanted to make a bet on GSM.

GSM才是未来。

The GSM was the future.

我认为GSM的架构使得它所需的基站更少，但覆盖范围更广。

And I think something about the GSM architecture made it so you could have there were fewer towers right now, but they had a wider range.

因此，这是一个面向未来的稳妥选择——当它们部署更多基站时，会更加高效。正因如此，我们至今仍使用AT&T，即使在Singular与AT&T合并之后。

So it was the safe bet for the future that when they put more of them out, they're gonna be more efficient, and we're still with AT and T to this day post Singular AT and T merger for that reason.

没错。

Yep.

因此，在美国，Singular和AT&T采用的是GSM标准，T-Mobile也是如此；而Verizon和Sprint则使用CDMA，两者彼此不兼容。

And so Singular and AT and T were with the GSM standard as was T Mobile in The US and Verizon and Sprint were CDMA and they were incompatible with another one another.

这就像移动电话世界的HD DVD和蓝光之争。

This was like the HD DVD and Blu ray of you know the mobile telephone world.

消费者想要理解这一点简直是一场噩梦。

And consumers trying to understand that is a complete nightmare.

哦，简直是噩梦。

Oh, complete nightmare.

但更糟糕的是，这些标准背后的实际技术都是CDMA，由高通公司商业化了。

Well, and and to make it even more of a nightmare though, the actual underlying technology behind these standards was all CDMA and Qualcomm, commercialized.

他们并没有发明CDMA技术，但他们是第一个将这种码分多址技术应用于手机世界的人。

They didn't invent the technology of CDMA but they were the first to commercialize the application of this to the cell phone world of co division multiple access.

这项技术的作用是，它是一种允许多重接入的通信标准，MA代表的就是多重接入。

And what that technology did, it was a standard that allowed for communication multiple access, the MA stands for.

它允许在单一波长或单一频段上同时传输许多不同的私有信道。

It allowed for the communication of many many different private channels over one wavelength or one frequency in the airwaves.

因此，当你知道有数百万用户同时使用蜂窝网络，每个人都在进行数据流媒体、通话和短信，但所有这些都通过相同的无线频谱传输时，

And so that's how when you know you have however many millions of people all on a cellular network and they all have their own, you know, data that they're streaming and calls that they're on and text messages that they're doing but it's all going over the same airwaves.

这就是如何将所有通信划分开来，让每个人都有自己的私有信道。

That's how it's all divided and everybody has their own private channel.

而实现这一技术的发明者是高通，因此他们开始将这项技术应用于手机基站，随后又制造了与之配套的手机芯片和手机终端。

And the technology to do that was invented by Qualcomm and so they started putting it into cell phone base stations and then they started making handsets and chipsets for handset, cell phone handsets that went with it.

随着基于这项技术的基站越来越多，能够利用这些基站的手机终端数量也随之增加。

And then the more base stations that got out there, that operated on this technology then the more handsets that were put on that could take advantage of those base stations.

于是，这演变成了一种良性循环，并形成了网络效应。

And then it just turned into this virtuous cycle and became a network effect.

尽管存在GSM和官方CDMA这两种不同的版本，但它们都基于高通的核心技术，而高通则通过授权从所有这些技术中获利。

And so even though there were these two different flavors of it with GSM and and official CDMA, it was all based on the underlying Qualcomm technology and Qualcomm was making money from licensing from all of it.

即使我们向前看，审视他们现在的授权收入，我认为这些收入大多来自专利，我相信我们稍后会深入讨论这一点。

Even if we flash forward now looking at their their licensing revenues, I think a lot of these come come from patents, and and and I'm sure we'll get into that.

但大约有三分之一的高通收入来自于授权协议。

But it's something like, maybe a third of of Qualcomm's revenue is is from licensing deals.

我想，我可以在这里查到确切的数字。

I think, I can get the actual number here.

是的。

Yeah.

高通上个季度或去年的10-K报表显示，设备和服务收入为154亿美元，许可收入为80亿美元。

Qualcomm's 10 k last quarter or last year was 15,400,000,000.0 in revenue from equipment and services and 8,000,000,000 from licensing.

我认为，其中很大一部分要归功于他们许可的这项底层技术，这项技术几乎存在于每一个手机无线电和基站中。

And and a lot of that I think goes back to this, this underlying technology that they're licensing out that's in just about every cell phone radio out there, base station.

我们是怎么从那里走到今天的？

How do we get from there?

从九十年代中期到九十年代末，一直到两千年代，他们一直专注于这项技术。

This is in the late ninety mid mid nineties through late nineties all the way through the 2 thousands that they're focusing on this.

我们是怎么从那里发展到今天 Broadcom 的局面的？

How do we get from there to where we are today with Broadcom?

嗯，首先他们制造基站，而高通也是，然后他们开始制造手机。

Well, you know, first they were making the base stations and Qualcomm was and then they started making phones.

我记得我上中学的时候，确实用过高通品牌的手机，这么说有点暴露年龄了。

So I remember actually having Qualcomm cell phones, Qualcomm branded cell phones back in the day when I was in middle school, dating myself there.

后来他们最终卖掉了那项业务，转而专注于生产手机用的芯片组。

And then they they they end up selling that business and they they go into making chipsets that go into the phones.

他们将手机业务出售给了京瓷。

They sold the the handset business to Kyocera.

我也记得用过京瓷品牌的手机。

I remember also having Kyocera based phones.

于是他们开始制造芯片组，这就是他们推出骁龙处理器并让大众熟知的时候。

And, so they go into making chipsets and this is you know when they get into like the Snapdragon processors that get branded that people hear about.

但近年来，随着手机市场变得极度饱和，许多并非由苹果和三星自主研发的处理器和芯片开始日益商品化。

But then over the past few years as the phone market has gotten really saturated and and a lot of the processors especially the non, processors that aren't designed, the chips that aren't designed by the Apples and Samsung's themselves start to become more commoditized.

正是在这一时期，我们开始看到这一领域的整合。

That's when we start to see consolidation happening in this space.

这就是去年博通与安华高并购交易发生的时候，如今博通正试图收购高通，以进一步整合市场。

This is when the Broadcom and Avago deal happened last year, and now Broadcom is coming in and attempting to buy Qualcomm to further consolidate.

是的。

Yeah.

还有一个快速补充：我想回到之前提到的，博通并不生产真正的CPU。

And one quick note I wanna rewind earlier where I said Broadcom doesn't make actual CPUs.

他们确实制造CPU。

They do make CPUs.

我想到的是，我不认为博通制造LTE天线。

The thing I was thinking of is I don't believe that Broadcom makes LTE antennas.

他们只是制造蓝牙、Wi-Fi之类的芯片。

They just make the chips that are Bluetooth and Wi Fi and all that.

我以前根本不知道高通曾经制造过手机。

First of all, didn't know that Qualcomm ever made a phone.

我们得去查一查，或者找一部那样的手机。

We're gonna have to, I don't know, dig up one of those or something.

好的。

Okay.

所以我们现在在这里。

So here we are.

现在正发生着大量的整合。

There's all this consolidation going on right now.

我的一个科技主题其实是，我们现在正处于一个时代，这些东西已经变得高度商品化，价格也被压得极低，因此我们必须看到行业整合，因为作为芯片制造商的研发成本极其高昂。

One of my tech themes was was actually that we're we're in this era now where this stuff has gotten so commoditized and and gotten pushed down so far in price that we need to see consolidation because the r and d costs to be a a chip manufacturer are enormous.

制造成本也非常高昂。

The manufacturing costs are enormous.

开发下一代新产品的周期漫长且昂贵。

The, cycle to to create the next new model is is long and expensive.

所以，我的意思是，这些公司要想真正与垂直整合者、手机制造商竞争，就必须合并，因为它们希望整合研发成本、制造成本等各方面资源，将这些优势捆绑在一起，然后在向手机制造商销售产品时也进行捆绑，从而可以说：看。

So, I mean, these companies, in order to really compete with with the vertical integrators themselves, with the handset manufacturers, have to combine because they they wanna take advantage of combining their r and d costs, their manufacturing costs, all that, and basically bundle those together and then bundle their products when they're selling them to the handset manufacturers so they can say, look.

我们不再让你从两家不同的公司采购，而是从一家公司全部采购，这样我们可以给你更优惠的价格。

We'll give you instead of buying from two different companies, you're buying from one, so we'll we'll cut you a deal.

我们可以把这部手机的所有内部元件都卖给你。

We'll sell you all the internals to this phone.

它们正感受到来自手机制造商的极大压力，因为这些制造商开始自行承担部分业务，为了维持利润，它们必须通过合并来应对。

And they're really just feeling a lot of pressure by the the handset manufacturers starting to take on some of this on their own where they need to preserve their margins by combining.

是的。

Yeah.

我的意思是，这可能是我们这个节目上第一次，除了阿拉斯加维珍那一期，真正看到了规模经济发挥作用，因为在软件领域，这通常并不成立。

Mean, is really, you know, I think this is one of the first times on this show maybe other than the Alaska Virgin episode where we've seen, you know, economies of scale like actually be a thing, because in software it usually doesn't work that way.

没错。

Right.

没错。

Right.

事实是，由于生产和开发成本如此之高，这种做法实际上是有道理的。

It so happens that this is so expensive to produce and so expensive to, to develop that it actually makes sense.

没错。

Yep.

显然，我们在讲述历史和事实时省略了很多内容，但为了把这一集控制在三小时以内。

Obviously, we've glossed over a lot there in the history and facts, but in the in an effort to keep this episode under three hours.

我打算，呃。

I'm gonna yeah.

嗯。

Yeah.

在我们进入收购类别之前，我想从历史和事实中挑出几件事。

There's a couple things on history and facts that I wanna pull out before we go into, into acquisition category.

其中一个就是这笔交易，我们之前说它被拒绝了。

So one of them is that this this deal so we we said it was rejected.

这笔交易金额为1030亿美元，不包括假设的债务。

A $103,000,000,000 deal, not including the assumed debt.

我认为，包括债务在内，收购价格是1300亿美元。

I think it's a $130,000,000,000 purchase price including the debt.

每股价格为70美元，以现金和股票形式支付。

It's a a $70 per share in cash and stock.

我预测这笔交易最终会在某个时候以略高于此的价格完成。

I'll make a prediction that this deal will go through at some point somewhere slightly higher than this.

所以，假设每股价格为80美元，以现金和股票形式支付。

So let's say it's, you know, at $80 per per share in cash and stock.

让我们基于这个前提来评估本集的其余内容，即交易最终会完成，获得监管批准——这也是我们接下来应该讨论的内容——并且价格大约在这个水平。

Let's evaluate the rest of this episode on that basis, that that it actually does go through, that it gets the regulatory approval, which is another thing we should talk about, and that that it's around that price.

了解这些公司目前发生的其他事情很有趣，这些事情塑造了它们股价现状的环境，以及它们所面临的外部压力。

It's interesting to know all the other things going on with these companies right now that are shaping the environment of why their share prices are where they are, why there there are external pressures.

目前正有诉讼进行中。

There's lawsuits going on.

联邦贸易委员会正在起诉高通公司。

The FTC is currently suing Qualcomm.

苹果公司目前正卷入与高通公司的诉讼中。

Apple is currently embroiled in a lawsuit with Qualcomm.

由于这些外部因素，它们的股价目前处于低迷状态。

Their share price is depressed right now because of these external factors going on.

所以，如果我是博通，我看到的是：哇，这是一个绝佳的机会，可以以相当便宜的价格收购一家优秀公司，因为它们正经历一段艰难时期。

So if I'm Broadcom, what I'm seeing here is, boy, there's an opportunity to to get a great company for pretty cheap because they're going through this sort of rocky time.

它们最终会走出困境。

They'll pull out of it.

它们的股价将反弹。

Their stock price will rebound.

但是，嘿。

But, hey.

我的意思是，一年前他们的股价是每股70美元。

I mean, they were trading at $70 a share a year ago.

现在他们的股价低于那个水平。

Now they're trading below that.

我们可以现在假装这还有一点溢价，等他们的内在价值实际上可能更高时再买回来。

We can pretend it's a little bit of a premium now and try and buy them at that again when their intrinsic value is actually probably higher.

所以，从时间框架来看，我们确实看到博通正在抓住机会。

So so we're definitely seeing, Broadcom being opportunistic, from a time frame perspective right now.

除此之外，还有银湖合伙公司，这是一家私募股权公司，实际上也是拥有Skype并将其出售给微软的那家公司，我们在本节目的Skype那一集中已经讨论过。

And then the other thing on top of that is that, Silver Lake Partners, a private equity firm and actually the private equity firm that owns Skype, and and sold that to Microsoft, which we we covered on the Skype episode of the show.

以及安德森·霍洛维茨作为一小部分。

Along with Andreessen Horowitz as a small piece.

没错。

That's right.

他们承诺提供50亿美元的可转换债务来为博通的这项收购融资，而博通的报价是现金加股票，每股包含10美元股票和60美元现金。

They committed $5,000,000,000 in convertible debt to finance this acquisition for Broadcom, and the Broadcom offer is cash and stock, where it's $10 of stock and $60 of cash per share for this.

因此，稍微审视一下这笔交易的结构及其周围的环境因素，挺有意思的。

So it's interesting to sort of look at the the structure of the deal a little bit and sort of the the environmental factors around it.

我认为，高通拒绝这个价格是正确的决定，因为他们也看到了博通只是在趁机行事，并且至少在初次报价时，并不愿意为这家公司的实际价值支付应有的价格。

And I think it's probably the right move for for Qualcomm to to reject it for this price because they're seeing the same thing that that Broadcom is is just being opportunistic and not not really being willing at least at the first go around here to to pay for what the company is really worth.

这引出了另一个关键点，即半导体领域与软件世界的巨大差异：这些公司普遍背负着大量债务。

That brings up another point that is a key difference in the semiconductor corner of the tech world versus much of the software world which is that these companies all have significant debt on them.

因此，它们都是高杠杆的，正如你所指出的，你看到了很多这样的结构安排。

So they're levered and you're seeing a lot of structure, know, as as you pointed out.

我的意思是，Silver Lake在这里通过可转换债务来帮助融资收购。

I mean, Silver Lake's involved here using convertible debt to help finance the acquisition.

这在软件界并不常见，比如Facebook通常直接用股票收购你。

This is not a not typically how things get done in the software world where, you know, Facebook buys you for stock.

为了进一步增加这笔交易的复杂性，博通目前正在以50亿美元收购Brocade，我想是这么念的。

And just to add even more hair to the deal on both side, Broadcom is currently acquiring Brocade, I think that's how you say it, for $5,000,000,000.

我稍后对这一点还有个评论。

I have another comment on that in a minute.

高通目前正在以470亿美元收购恩智浦半导体。

Qualcomm is in the midst of acquiring NXP Semiconductors for $47,000,000,000.

博通收购高通的报价并不以这两笔交易中的任何一笔完成为前提，这一点本身就很有趣。

The offer from Broadcom to buy Qualcomm was not contingent upon either of these things closing, which is is interesting in its own right.

如果它确实完成的话，抱歉。

If it does close or I'm sorry.

如果博通收购了高通，这意味着什么？

If Broadcom does buy Qualcomm, what does that mean?

比如，高通收购恩智浦的交易还会继续吗？

Like, does the does the m x NXP deal go away?

它们会尝试作为合并后的实体继续推进这笔交易吗？

Do they try and and continue doing that even as a combined entity?

这笔交易能通过监管审批吗？

Will that go through regulatory approval?