From 10 to 10,000

The content here is for informational purposes only, should not be taken as legal, business, tax or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any a16z fund.

本节目内容仅供参考，不构成法律、商业、税务或投资建议，也不可用于评估任何投资或证券，并非面向任何 a16z 基金的投资人或潜在投资人。

For more details, please see a16z dot com slash disclosures.

更多细节请见 a16z.com/disclosures。

Hi, everyone.

大家好。

Welcome to the a16z podcast.

欢迎收听 a16z 播客。

Today's episode features Chris Dixon, now a general partner on a16z Crypto, interviewing Elad Gil.

今天这期，由现在担任 a16z Crypto 普通合伙人的 Chris Dixon 来采访 Elad Gil。

The author of a new book just out called the High Growth Handbook Scaling Startups from 10 to 10,000 people.

他是刚出版的新书《High Growth Handbook》的作者，副标题是把创业公司从 10 人扩张到 10,000 人。

Elad is an investor or advisor to numerous tech companies.

Elad 是众多科技公司的投资人或顾问。

As well as cofounded Color Genomics.

他还共同创办了 Color Genomics。

He was also cofounder and CEO of a company that was acquired in the early days of Twitter, where he also then became a VP.

他也曾联合创办一家公司并出任 CEO，Twitter 早期把这家公司收购了，他也就此成了 Twitter 的 VP。

And he also worked at Google, where he started their mobile team and worked on AdSense.

他还在 Google 工作过，在那里组建了移动团队，也做过 AdSense。

The discussion that follows covers everything from executive hiring, to product management, to late stage financing, and more.

接下来的对话涵盖了高管招聘、产品管理、后期融资等方方面面。

The two also then discuss broader market and tech trends.

两人随后还聊了更宏观的市场与技术趋势。

Ranging from the continuation of mobile cloud, to machine learning, to crypto, and finally to longevity, both near term and further out in the future.

从移动和云的延续，到机器学习、crypto，最后聊到长寿——既有近期的，也有更远的未来。

But they begin with the challenges of scaling.

不过他们是从扩张的挑战聊起的。

There's sort of famous books like Peter Thiel's book, which is from 0 to 1, kind of the early stages of a startup.

有几本很有名的书，比如 Peter Thiel 那本《Zero to One》，讲的是创业公司的早期阶段。

And I think a lot of blogging and other kinds of stuff has been out there from like Paul Graham and Fred Wilson and things about fundraising and product ideation and things like this.

我觉得像 Paul Graham、Fred Wilson 也写了不少博客和别的东西，讲融资、产品构思这类事。

But there hasn't been much written about all the later stages.

但关于后面那些阶段，写的东西不多。

⟨?And to your point⟩, I think there's, you know, Peter Thiel's book is Zero to One, and really the hope is to have a book that's talking about when you go from 10 to 10,000.

顺着你的话说，Peter Thiel 那本书是《Zero to One》，而我真正希望的，是有一本讲从 10 人走到 10,000 人的书。

You know, something's working, you have product market fit, and then what are all the different things that break?

已经有东西跑通了，你有了 product market fit，那接下来会坏掉的都是些什么？

If you think about it the surface area of an early stage company is actually really simple.

想一想，早期公司的表面积其实非常简单。

It's don't run out of money, don't fight with your cofounder, and find product market fit.

别把钱烧光、别跟联合创始人吵翻、找到 product market fit。

And that's that's all you have to do, and you'll be successful to some extent.

要做的就这些，做到了你多少就能成。

Oh, it's of course very hard to do.

当然，这做起来非常难。

It it it's incredibly hard, incredibly hard, but You know, those are the three things.

难得要命，难得要命，但就是这三件事。

Yeah.

对。

At at the late stage, there's all sorts of things that can break.

到了后期，可能坏掉的东西五花八门。

Executive hiring, buying companies for the first time, whether you should go public, how do you do late stage fundraises.

招高管、第一次收购公司、要不要上市、后期融资怎么做。

How do you build a product management org, like there's all this complexity that starts to kick in, and so this book is really meant to try and address those later stages, something starting to work, what do you do?

怎么搭起产品管理这套组织——各种复杂度开始冒出来，所以这本书真正想处理的就是这些后面的阶段：东西开始跑通了，然后你该做什么？

You know, it's interesting, cause I used to be based in New York, and now based in Silicon Valley.

有意思的是，我以前在纽约，现在在硅谷。

And there's a lot of, you know, the companies that get to product market fit.

有很多公司能做到 product market fit。

They're not just in Silicon Valley, right?

它们不只在硅谷，对吧？

There's plenty of companies in New York and other cities, other places.

纽约和其他城市、其他地方也有一大把。

What I've noticed is different about Silicon Valley is There's a whole bunch of people who have done the scaling the one to n phase.

我注意到硅谷不一样的地方是，这里有一大批人做过从 1 到 n 的扩张。

And so, I kind of describe it as like the second stage rocket, you know, that sort of the first stage is finding product market fit, and then can you get the second one layered on there?

所以我把它比作第二级火箭：第一级是找到 product market fit，然后你能不能把第二级接上去？

And one of the advantages of Silicon Valley, I think people talk about how there's more engineers, and there's more investors, but those are frankly everywhere.

说到硅谷的优势，大家常说这里工程师更多、投资人更多，但说实话这些哪儿都有。

But the big difference is VP Marketing, VP Sales, VP Engineering, and then all the product managers, and just sort of all those layers in between.

真正的差别是 VP Marketing、VP Sales、VP Engineering，还有那些产品经理，以及中间的各个层级。

Somebody was telling me, a European founder was telling me that they felt that often European technology companies will cap out at a certain stage, and it's in part market dynamics around the EU market, but in part.

有人跟我说过，一位欧洲创始人跟我说，他觉得欧洲的科技公司常常会卡在某个阶段，一部分是欧盟市场本身的动态，但另一部分，

it's just driven by the fact that you tap out in terms of where the executive team is capable to take things given their experience.

只是因为，以高管团队现有的经验，他们能把公司带到的高度就到头了。

So, there's a lot of misconceptions or miss around scaling startups.

关于创业公司的规模化，有不少误解，也有不少没讲清楚的地方。

What do you think are the biggest changes in the company as the organization grows?

你觉得随着组织变大，公司里最大的变化是什么？

I mean, the naive view is you fifty employees, you have a hundred people, that's just twice as many, and you gotta do some, you know, more management.

天真的看法是：原来五十个员工，现在一百人，不过是人数翻倍，多做点管理就行了。

But of course, what really happens is the communication patterns Break down, and a whole bunch of new things have to be developed.

但真实发生的是，沟通模式垮掉了，一大堆新东西得从头搭起来。

There's three or four big shifts, so you you touched on a really key one, which is the way that you communicate is fundamentally different, particularly if you're high up in the company.

有三四个大的转变。你刚点到了很关键的一个：沟通方式会根本性地变样，尤其你在公司高层。

And part of the reason is, you know, if you're ten people, you can just sit in a room and talk to everybody.

原因之一是，十个人的时候，坐在一间屋里就能跟每个人说上话。

If you're a hundred people, there's two to three layers between the CEO and every other person.

一百人的时候，CEO 和其他每个人之间隔着两到三层。

But as you start hitting five hundred to a thousand people, the layers multiply.

但等你冲到五百人、一千人，层级会成倍增加。

And the ability for information to move up and down the stack shifts.

信息在这个层级里上下传递的能力，也跟着变了。

And that also means that the modes of communication has to change, right?

这也意味着沟通的方式必须跟着改，对吧？

You can't go and have as many skip level meetings.

越级沟通会没法开那么多了。

Not everybody's gonna ship to the all hands anymore.

全员会也不再是人人都到场。

You have people in international offices.

会有同事在海外办公室。

And so you really have to move the modes of communication in a pretty deep way.

所以沟通方式真得做一次很深的调整。

The second thing is you really have to focus on building out a strong executive staff, and not just that, you're gonna have entirely new functions.

第二件事是，你必须认真搭一支强的高管团队；不止如此，你还会碰上全新的职能。

that you as the CEO have never dealt with, and you have to figure out how do I hire a CEO for the first time?

这些职能你身为 CEO 从来没打过交道，你得想明白：第一次招一个 CEO，我该怎么招？

How do I think about customer support at scale?

规模化之后的客户支持，该怎么想？

And so you have to start understanding how are you gonna hire all these really talented people, and convince them to come in and You know, build out their respective orgs.

所以你得开始搞清楚：怎么招到这些非常有才华的人，说服他们进来，把各自的组织搭起来。

The third is that you start adding new processes, and you have to have a certain set of lightweight processes for strategic planning, for deciding how to budget.

第三是开始加流程，你得有一套轻量的流程来做战略规划、来定预算。

At a thousand people, it's more about how do you start allocating resources to the right areas.

到一千人的时候，更多是怎么把资源分配到对的地方。

And then lastly, the surface area of what you're doing expands so dramatically that you have to really learn um how to delegate and manage at scale.

最后，你手上事情的表面积扩张得太猛，你必须真正学会放权，学会在大规模下做管理。

So if you're internationalizing And you're dealing with different time zones, and a much larger orb, you have to deal with that.

如果你在做国际化，要面对不同时区、面对大得多的组织，这些你都得处理。

If you're buying companies, you now have to have people who can deal with the integrations of those companies.

如果你在收购公司，就得有人能把这些公司整合进来。

If you're launching multiple product lines, how do you think about deprecating the old ones?

如果你在推多条产品线，老的产品线怎么下线？

How do you do cross sales, bundles, et cetera?

交叉销售、打包销售这些又怎么做？

And so The complexity keeps sort of notching up across the board.

复杂度就这样在各条线上一格一格往上抬。

Yeah, one thing I've noticed, so for example, like if you take your typical like former programmer, engineer, founder, and let's say they're doing you know a B to B kind of sales company, and then they have to at some point hire a sales leader and marketing and all these other things.

是啊，我注意到一点。比如一个典型的程序员、工程师出身的创始人，假设做的是一家 B2B 销售型公司，到某个时点他必须招销售负责人、招市场，还有其他一堆职能。

They have never done that before, right?

这些他以前从来没干过，对吧？

And.

而且。

so, they just have no idea what greatness looks like.

所以他们根本不知道优秀长什么样。

And so, you know, how do they do that?

那他们该怎么办？

Number one, and number two, there's a tendency, I think, to apply some of the same filters that one might use in the first ten people when you're hiring product.

这是第一点。第二点，我觉得有个倾向：把招前十个人做产品时那套筛选标准直接套过来。

So, for example, I've heard multiple times engineer founders wanting sales people who know how to program or something, right?

比如我不止一次听到工程师出身的创始人想要会写程序的销售，诸如此类。

Which is probably not compatible with generally great enterprise sales skills.

而这跟通常意义上出色的企业级销售能力，大概是不兼容的。

And so, what kind of advice do you give for people who are hiring in these roles, for jobs I've never frankly even worked, maybe with, let alone hired for.

那招这类岗位的人，你会给什么建议？说实话，有些岗位我连共事都没共事过，更别说招人了。

I think there's basically two things that are sort of good generic advice, and obviously the only good generic startup advice is that there's no good generic startup advice, right?

我觉得有两条还算不错的通用建议——当然，唯一好的通用创业建议就是：没有好的通用创业建议，对吧？

So, take everything with a grain of salt, but The first is don't reinvent everything.

所以听什么都打个折扣。第一条：别把所有东西都重新发明一遍。

I think if you're an innovative, especially technical founder, you wanna rethink what's my org structure, and ultimately, you know, there's a reason that corporations have been around for hundreds of years and actually work, and There's certain org structures that you can innovate on.

如果你是个爱创新的创始人，尤其技术出身的，你会想把组织结构重新想一遍；但说到底，公司这个形态存在了几百年、而且确实管用，是有原因的，有些组织结构你确实可以创新。

but ultimately you should find certain experienced people to do certain things where the function needs that experience.

但最终，在那些真正需要经验的职能上，你该找有经验的人来做。

So that's sort of a meta point.

这算是一条元层面的判断。

I think in terms of finding an executive, there's a few steps that you can take.

具体到怎么找一个高管，有几步可以走。

The first one is You should go and meet all the best people of that function.

第一步，去见这个职能里最优秀的那批人。

It's not necessarily people to hire, it's people to learn from.

他们不一定是拿来招的，而是拿来学的。

So, what does a great CFO look like?

一个出色的 CFO 长什么样？

And then I would just ask them, what would you look for in a person that would fit my company?

然后我会直接问他们：要找一个适合我公司的人，你会看什么？

And What sort of experience would you look for?

你会找什么样的经验？

What interview questions should I ask them?

我该问他们哪些面试问题？

How should I get to know this person?

我该怎么去了解这个人？

How will I know if they're good in the function?

我怎么才能判断他在这个职能上到底行不行？

Second, once you get that feedback, I'd write up a job break, and I'd circulate it to All the people who are gonna be interviewing this individual, because they don't have any context either, and they have no idea what to look for.

第二，拿到这些反馈之后，我会写一份岗位说明，发给所有要面这个人的同事——他们同样没有上下文，不知道该看什么。

And so I found for example, when you're hiring a BD person or somebody to run BD, and early company will have a mix of people from engineering and product and other orgs.

我就发现，比如招一个 BD、或者招人来带 BD，早期公司来面试的会是工程、产品和其他部门混在一起的人。

And they may have no idea what to actually look for, and they may think it's a sales person versus a BD person, versus something else.

他们可能根本不知道该看什么，可能把这个岗位当成销售，而不是 BD，或者当成别的什么。

And so just defining that tightly so everybody knows these are the questions we're gonna ask, these are the characteristics we're looking for.

所以把这件事定义得足够紧：让所有人都知道我们要问哪些问题、要找哪些特质。

Allows both clarity of interviewing, but then it also allows for clarity of discussion of the candidate once you make that higher.

这既让面试更有章法，也让最后拍板时对候选人的讨论更清楚。

And then the last thing is you should figure out how to onboard them effectively.

最后一件事是，想清楚怎么把他有效地带上手。

And part of that will be working with them closely to figure out what they think their role should be, and what the function should be, what they think it should evolve as, and then mixing in your own opinion, and making sure that you don't feel overwhelmed by their expertise.

其中一部分是跟他紧密配合，弄清他觉得自己的角色该是什么、这个职能该是什么、以后该怎么演进，再掺进你自己的看法，同时也别让他的专业能力把你镇住。

But.

但是。

you're willing to defer when they really understand something you don't.

在他确实懂一些你不懂的东西时，你愿意听他的。

Yeah, one thing, one of the process I've observed is companies that don't think they need, let's say, a CFO.

对，我观察到的一个现象是：有些公司觉得自己不需要，比如说，CFO。

And what we like to do is we do what we call calibration where we introduce them to three great CFOs.

我们喜欢的做法是，做一次我们所谓的对标——给他们引荐三位顶尖的 CFO。

You probably aren't looking for jobs, and almost invariably the founder comes back and says, oh my god, I didn't know that someone could do those things, and like all those problems it would solve for me, and it's completely eye opening, right?

这几位大概都不在找工作，而创始人几乎无一例外会回来说：天啊，我都不知道有人能干这些事，能替我解决掉这么多问题——完全打开了眼界，对吧？

Because if your background is engineering or whatever, you've just never seen greatness in CFOs, VP Sales, et cetera, right?

因为你要是工程或者别的什么出身，就从来没见过顶尖的 CFO、VP Sales 长什么样，对吧？

Yeah, absolutely.

对，完全同意。

I think the services model that Andreessen Horowitz has built is really valuable for founders as they're scaling in the CEO role, cause to your point, otherwise they'll have no idea.

我觉得 Andreessen Horowitz 建起来的这套服务体系，对创始人在 CEO 这个位置上往上长特别有价值，因为正如你说的，不然他们根本无从知道。

Founders have a bunch of questions about hiring execs.

创始人在招高管这件事上有一堆问题。

When should they hire them?

什么时候该招？

What should they look for?

该看重什么？

There's lots of debate about How high the bar should be, and you know what kind of athletes versus veterans or something.

门槛定多高，争论很多，还有是要“运动员”型的潜力股，还是要打过仗的老将。

Where do you come down on those issues?

这几个问题上你倾向哪一边？

So, there's a few different sections of the book that talk about both executive hiring, as well as managing your executive team, and how do you go about it.

书里有好几节讲的就是高管招聘，还有怎么管理高管团队、具体该怎么下手。

And I think there's a few key tenants in terms of bringing on execs or alternatively promoting people from within.

我觉得引进高管、或者反过来从内部提拔人，有几条关键原则。

I think Ben Horowitz may touch about this, I can't remember in his book or in other articles.

Ben Horowitz 好像讲过这个，我记不清是在他书里还是别的文章里。

You should always be thinking of the person as the right person for the next 12 to 18 months.

你要始终把这个人看成未来 12 到 18 个月的合适人选。

Not the person for the next five years.

而不是未来五年的那个人。

And so, I think a common mistake is that people will end up hiring, you know, they'll have a 20 percent engineering team, and they say, hey, in five years will be 500 people.

所以我觉得一个常见错误是，大家会招过头——手上是 20 人的工程团队，却说，嘿，五年后我们会有 500 人。

So, let's hire that 500 percent engineering team personnel now.

那现在就招一个带得动 500 人工程团队的人吧。

They hire way too far ahead, and that's a person who's out of a big company, used to managing a big organization.

他们招得太超前了，而那种人是从大公司出来的，习惯了管大组织。

I'll basically show up, be incredibly bored, and either try to overhire and hire a super senior staff, or just turn out.

他来了基本上会闲得发慌，要么想着超额招人、招一堆特别资深的班底，要么干脆走人。

Second, people tolerate good versus great hires for too long.

第二，大家容忍“不错但不卓越”的人选，容忍得太久。

And so they'll have a good VP marketing, versus the person who's truly exceptional and a great fit for what they're doing.

他们会留着一个还不错的 VP marketing，而不是换上那种真正出类拔萃、跟手上的事高度契合的人。

And it's almost like being on a life raft, and you're like, I have five spots, and who are those five spots?

这几乎就像坐在救生筏上，你会想：我只有五个位置，这五个位置给谁？

It's not this this giant yacht, you know, cruising down the Mediterranean that you can have hundreds of people on, and so really your executive team should be this life raft.

这不是一艘在地中海上巡游、能装几百人的大游艇，所以你的高管团队真的就该是这只救生筏。

And you should ask if this person's good, but not great.

你应该问自己：如果这个人只是不错、算不上卓越呢？

How do I find great?

我上哪儿去找卓越的？

Sometimes, by the way, those people will be co-founders, other people who were early, you're loyal to them, they were really helpful in the beginning, and they may still be great in the organization, but maybe they're not the best fit, right?

顺便说一句，有时候这些人就是联合创始人，或者是早期就在的人，你对他们有情义，他们在最开始帮了大忙，他们在组织里可能依然很出色，但也许已经不是最合适的人了，对吧？

I don't know, that's attention I've seen sometimes.

说不好，这是我有时候会看到的一种张力。

I think in general, dealing with Sort of what are called oldtimers in the book is a key question, because they've proven their loyalty to the company.

我觉得总体上，怎么处理书里所说的元老，是个关键问题，因为他们已经证明了对公司的忠诚。

They were essential in their early days.

早期他们不可或缺。

They have all the cultural context, they have all the trust of the founders, which is really important and valuable currency in terms of getting things done.

他们掌握全部的文化脉络，也握着创始人的全部信任，这在推动事情落地时是非常重要、非常值钱的本钱。

And a subset of them are gonna act really badly, when all said and done.

而其中一部分人，说到底，会表现得很难看。

They're going to drag their feet, they're gonna argue with new hires, they're gonna effectively throw sort of a prolonged MiniFit, and part of that is just driven by the fact that they used to have enormous influence, and that influence is diminishing.

他们会拖后腿，会跟新人吵架，实际上就是闹一场没完没了的小情绪；一部分原因是，他们过去有巨大的影响力，而这份影响力正在缩水。

Part of that is it's a very chaotic.

还有一部分原因是，那个环境本身非常

environment, and part of that is not understanding what's coming.

混乱；另一部分原因是，他们没看懂接下来会发生什么。

And what's coming is enormous growth potential for everybody who's involved with the company.

而接下来会发生的是：凡是跟这家公司沾边的人，都有巨大的成长空间。

And so, I remember when I joined Google, it was about 1500 to 2000 people, and over 3.5 years grew to 15000 people.

我记得我加入 Google 时大概是 1500 到 2000 人，三年半里长到了 15000 人。

And the first time I went through hyper growth, I was caught up in all the chaos.

第一次经历超高速增长，我被那股混乱裹着走。

If I had a peer who got promoted above me, I'd worry, what about me?

要是有个跟我平级的人被提到我上面，我就会犯嘀咕：那我怎么办？

And the second time I went through it, it was at Twitter, and they bought my startup when Twitter was about 90 people.

第二次是在 Twitter，他们收购我创业公司的时候，Twitter 大概 90 人。

And over two and a half years, it went to fifteen hundred people.

两年半里，涨到了一千五百人。

And the second time through it, I just thought, you know what?

第二次经历的时候，我就想，你猜怎么着？

The company is gonna grow in all these different directions, and that means that from a career perspective, I'll be fine.

公司会朝各个方向长，这意味着从职业发展看，我不用担心。

And I should just do good work, and keep my head down, and try to be helpful.

我该做的就是把活干好、埋头做事、尽量帮上忙。

And so, there was uh product reorg that happened on the Twitter team, that was a little bit messy.

后来 Twitter 的产品团队做过一次组织调整，过程有点乱。

And Dick was asking different product managers what they wanted to do, and I said, well, I just wanna help the company.

Dick 挨个问产品经理想做什么，我说，我就想帮公司。

Like, put me wherever you want, I don't care.

把我放哪儿都行，我不在乎。

I just want to help the company.

我只想帮公司。

And then two months later, I got promoted to be a vice president there.

结果两个月后，我在那儿升任副总裁。

And I think it was because I just put everything else aside and said, where can I help?

我觉得原因就是，我把别的都放到一边，只问：我能在哪儿帮上忙？

So I think that's a really important thing.

所以我觉得这一点真的很重要。

Put aside the anxiety, focus on the big picture.

把焦虑放到一边，盯住大局。

And realize that every six months, if the company is in hyper growth, it's a different company.

还要意识到，公司如果在超高速增长，每六个月它就是一家不同的公司。

If it goes from fifty to two hundred people, to five hundred people over six to twelve months increments, it's a completely different company.

如果它以六到十二个月为一段，从五十人到两百人、再到五百人，那完全是另一家公司了。

There's new opportunities.

会冒出新的机会。

Somebody's gonna go and need to open international offices.

总得有人去开海外办公室。

Somebody's gonna need to launch new products.

总得有人去发新产品。

Somebody's gonna be needed for new functions.

总得有人去扛起新的职能。

And you can be that person if you do good work and keep your head down.

只要你把活干好、埋头做事，那个人就可以是你。

And the other thing you should expect as an old timer is Your role is gonna shrink, and then it'll expand.

作为元老，你还该预料到另一件事：你的职责范围会先缩小，然后再放大。

And you need to be OK with that shrinking period, and you need to be OK with you losing responsibility before you suddenly gain tons of responsibility.

你得接受那段缩小期，得接受自己先交出一部分职责，然后才会突然拿到一大堆职责。

Because if you're the only designer, you were designing everything.

因为如果你是唯一的设计师，那所有东西都是你在设计。

But you're gonna start hiring other designers, and you're gonna have to give things away.

但你会开始招别的设计师，你必须把一些活交出去。

But a year later, they're gonna have to promote somebody to run a chunk of the design team, and that may be you, if you act the right way.

可一年之后，他们得提拔一个人来带设计团队的一大块，如果你表现得当，那个人可能就是你。

And so, I think really keeping that in mind is crucial.

所以我觉得，把这一点记在心里非常关键。

So, let's talk a little bit about product management.

那我们聊聊产品管理。

What are some of the key learnings that people should know about there?

这方面有哪些关键经验是大家该知道的？

There's very few very strong product organizations in Silicon Valley.

硅谷里真正很强的产品组织没几个。

Google has a great product or Facebook has a great product or there's a handful of them.

Google 的产品很棒，Facebook 的产品很棒，也就这么一小撮。

And part of that I think is because often technical founders either create a pseudo role for that and Other functions are basically filling the gap, or in some cases, it's back to the earlier point around.

我觉得部分原因是，技术型创始人往往要么给这事设个名不副实的岗位，让别的职能来填这个缺口；要么就回到前面提到的那点。

you never see excellence in the roles, you don't really understand how to build it out.

你从没在这个岗位上见过卓越的样子，也就不真懂该怎么把它搭起来。

And so often the evolution of a product organization will be the one of the founders will be really the product manager for the product.

所以产品组织的演化常常是这样：先由某位创始人实际担任这个产品的产品经理。

And then it'll evolve into a state where there's gaps, and they're trying to fill the gaps around product, and so they'll pull in somebody from another team.

然后演化到出现缺口的阶段，他们想把产品这块补上，于是从别的团队调个人过来。

It could be marketing, it could be sales, it could be business operations, it could be something else.

可能是市场，可能是销售，可能是业务运营，也可能是别的什么。

And those people may be very exceptional at running process and running schedules, but they may not have the product insights, and the experience to really know, OK, this is how you think about product roadmap.

这些人也许非常擅长跑流程、管排期，但他们可能没有产品洞察，也没有那种经验去真正知道：好，产品路线图该这么想。

this is how you think about a PRD or writing a spec for the product for the first time.

第一次写 PRD、写产品 spec 该这么想。

And here's a set of processes that help me engage all the stakeholders around the company to really sort of drive a product forward.

还有这么一套流程，能帮我把公司里所有利益相关方都拉进来，真正把产品往前推。

So eventually what happens is if those processes start to break down or that approach starts to break down.

所以最后会发生的是，如果那些流程开始失效，或者那套打法开始失效。

They'll end up eventually hiring and somebody who has deep product experience to build out an org, and then it sort of gets professionalized.

他们最终会招一个有深厚产品经验的人来把这个组织搭起来，这块也就专业化了。

Facebook and Google, they obviously are hiring great people, but I think it's also the way they empower those people and design the organization that makes some great product organizations.

Facebook 和 Google 当然在招很优秀的人，但我觉得让他们成为强产品组织的，还在于他们怎么赋能这些人、怎么设计组织。

Yeah, they ended up hiring Jonathan Rosenberg and externally to run product for Google, and what he ended up doing is, I think three key things.

是的，他们最后从外部招了 Jonathan Rosenberg 来负责 Google 的产品，而他做的事，我觉得有三件是关键的。

Number one is, he really started mentoring a lot of the key product individual contributors who went on to actually run most of the product org at Google in the early days.

第一，他真的开始带教一批关键的产品一线执行者(IC)，这些人后来撑起了 Google 早期的大部分产品组织。

So uh, Susan Wojcicki who now is CEO YouTube, Salar who similarly had run YouTube and before that uh the ads team, Marissa Mayer who became CEO Yahoo, see these people who are sort of all stars in terms of the things that they've accomplished.

比如 Susan Wojcicki，她现在是 YouTube CEO；Salar 也一样管过 YouTube，更早之前管广告团队；Marissa Mayer 后来当了 Yahoo CEO——看看这些人，论做成的事都是全明星。

And very early on, they were basically individual contributors on the product org, and they were really learning the ropes, and some came from engineering backgrounds, and some came from marketing or other backgrounds back to that mix of people sort of joining.

而在非常早期，他们不过是产品组织里的一线执行者(IC)，真的是在摸门道，有些人是工程背景，有些来自市场或别的背景，又回到刚才说的那种人员混搭。

And so, one is he mentored people actively.

所以，一是他主动带教人。

Two is he really focused on the process side of it.

二是他非常聚焦流程这一面。

What are the checkpoints for launching a product?

发布一个产品有哪些检查点？

How do we approach it?

我们该怎么推进？

And then lastly, he actually built a training program called the associate product management, uh, product manager program.

最后一点，他还真搭了一个培训项目，叫 associate product manager 项目。

Where they basically recruited really really smart people out of in the Google case top universities, and they basically said we're gonna train these highly technical, really smart people to do product management, and we're gonna select for people who in our interview process Seem to exhibit very strong product instinct.

他们招的是非常非常聪明的人，Google 的情况是从顶尖大学里招，然后说：我们要把这些技术底子很强、非常聪明的人训练成做产品管理，而且我们会专挑那些在面试里看上去产品直觉很强的人。

and so that group ended up including a variety of people who've really gone on to impact product organizations all over Silicon Valley.

结果那批人里出了一大串，他们后来真的影响了整个硅谷的产品组织。

So, this will go to the question of what is a great product manager?

那这就引到那个问题：什么才是优秀的产品经理？

Just first, what is product management?

不过先说，产品管理到底是什么？

'Cause I think it can confuse sometimes with project management on the one hand, design on the other hand.

因为我觉得它有时候一头会跟项目管理混淆，另一头会跟设计混淆。

Yeah, that's a good point.

对，这问题问得好。

Ben Horowitz has a good blog around good product manager, bad product manager from sort of the nineties, the tries to encapsulate enterprise product management.

Ben Horowitz 有篇很好的博客，大概是九十年代写的，讲好产品经理、坏产品经理，想把企业级产品管理讲透。

You know, ultimately the role of a product manager is fourfold.

归根到底，产品经理的职责有四方面。

Number one is to really own the generation of the product requirements, and what's the user need?

第一是真正扛起产品需求的产出：用户的需求是什么？

How will this actually get used?

它实际会被怎么用？

How's it differentiated?

它的差异化在哪？

What are the key components that make this a good product?

哪些关键部分让它成为一个好产品？

So, one is sort of owning that view of the product and roadmap and sort of.

所以第一点，就是扛起对产品和路线图的这套判断。

And depending, for example, an enterprise company typically they'll sit between sales and engineering, and kind of mediate the two.

这也要看情况，比如企业级公司里，产品经理一般坐在销售和工程中间，居中调停两边。

Yeah, there's typically a natural tension between product and other organizations.

对，产品和其他组织之间通常天然有张力。

And a sort of a second role that they often play particularly in enterprise companies, is there almost a buffer between engineering and the rest of the world.

他们常常扮演的第二个角色，尤其在企业级公司里，几乎就是工程和外部世界之间的缓冲垫。

But they're also a communication device both in and out.

但他们同时也是一个双向的沟通渠道。

In other words, They'll help synthesize a lot of the different feedback from sales and directly from customers.

换句话说，他们会帮着把来自销售、也直接来自客户的各种反馈整合起来。

I want this feature, I want that.

我要这个功能，我还要那个。

And the sales people are kind of incented to say yes.

而销售那边多少有动力说“行”。

The engineers on the other hand want to keep the product focused, and get their job done.

工程师则相反，想让产品保持聚焦，把自己手上的活干完。

And so there's a tension, right?

所以这里就有张力，对吧？

And the product manager sits in between and kind of prioritizes and figures out what really matters.

产品经理夹在中间，排优先级，搞清楚什么才真正重要。

Exactly.

没错。

And on the engineering side, there's tension with product, because engineers may think a feature stupid.

而在工程那一侧，跟产品之间也有张力，因为工程师可能觉得某个功能很蠢。

Because they're not necessarily hearing the direct feedback from the customer, and they may not believe, oh, the customer really needs this.

因为他们不一定听得到客户的直接反馈，他们可能不信客户真的需要这个。

Or they may want to go down the road of building something that may be more complicated than what's really needed to future proof it or for other reasons.

或者他们可能想去做一个比实际需要更复杂的东西，为了给未来留余地，或者出于别的原因。

And if product pushes back on that, there may be tension there as well.

而如果产品这边把它顶回去，那边也会有张力。

And so, the role of a product manager in part is also to navigate all that tension between the different organizations and end up with a view of what's really needed for the product despite all these various inputs.

所以产品经理的职责之一，就是在各个组织的张力之间周旋，在这么多输入里最终形成一个判断：产品到底真正需要什么。

Great, so let's talk about in your book, you talk about late stage financings.

好，我们来聊聊你书里讲的后期融资。

There's been a lot written and sort of blogged and other things about early stage, and I think people at least in the startup community kind of understand series A and B, but there's a whole Kind of new.

关于早期，已经有一大堆文章和博客写过了，创业圈里大家至少大致明白 series A 和 B 是怎么回事，但现在还多出一整类新的东西。

uh, new I say, because in the old days, 10 years ago, you'd go public or something, and now there's this whole kind of other set of investors.

我说“新”，是因为放在过去、十年前，你差不多就直接上市了；而现在多出了一整批别的投资人。

What are the things, the main things that founder should know about that?

关于这块，创始人最该知道的几件事是什么？

Yeah, I think there's a few key things that founder should consider when doing a very late stage round.

我觉得创始人在做非常后期的一轮融资时，有几个关键点要考虑。

And many of these rounds in some cases retired secondary, and some cases are not.

这类轮次里有不少会搭上老股转让，也有不搭的。

So, the first is there are some value added late stage investors.

第一，后期投资人里确实有一些能带来增值。

And so, just like in the early stages, not all money is equivalent, and in the late stages, there may be people who will continue to buy stock as you go public, and therefore help support the price of the stock in the early days.

就像早期一样，钱和钱是不一样的；到了后期，有些人会在你上市时继续买入股票，从而在挂牌初期帮忙托住股价。

as things start.

就在刚起步的那阵子。

These can be the so called crossover investors.

这些大概就是所谓的 crossover 投资人。

Yeah, T. Rowe Price, Fidelity, and the like.

对，T. Rowe Price、Fidelity 这类。

They may have interesting insights in terms of how public markets will think about them, and therefore how they should be thinking about navigating their business really on or the sets of reporting or controls or other things that they wanna build in.

他们对公开市场会怎么看这家公司往往有独到见解，顺带也知道公司该怎么把生意往前带、该搭哪些汇报体系、内控或者别的东西。

or they may have real insights into markets.

也可能他们对市场有真正的洞察。

Or differential networks that you normally wouldn't get access to.

或者手里有你平时根本接触不到的另一层人脉。

And so, I do think that there's investors who can help in a more traditional sense, they just get involved later in the life of a company.

所以我确实觉得，有些投资人能用更传统的方式帮上忙，只是他们介入的时点更靠后。

So, the first question is just whether or not This money will be helpful.

所以第一个问题就是：这笔钱到底有没有帮助。

It doesn't have to be frankly, because in some cases, or in many cases, you have escape velocity.

说实话不一定非得有，因为很多时候你已经有了逃逸速度。

But optimally, you have capital on board, that's always gonna help.

但最理想的是，上船的资本本身就帮得上忙，那总归是加分。

The second thing is that You don't wanna get too far ahead of yourself in valuation, and I think often at the later stages, there's a temptation to do that, because the odd thing is, the higher the the valuation of the company and the better known the brand The more sources of capital exist.

第二点是，别让估值跑得太超前；后期常常有这种诱惑，因为有个奇怪的现象——公司估值越高、品牌越响，能拿到钱的来源反而越多。

and so what you see is on the late stages, some companies start building in very complicated structures, because they want that very high valuation, and the firm may be willing to give it, but then they build in Extra preference or participation for the investor.

于是你会看到，到了后期，有些公司开始往里塞非常复杂的结构性条款，因为他们想要那个极高的估值，机构可能也愿意给，但接着就会给投资人加上额外的优先权或参与分配权。

or things that effectively are really a debt instrument masquerading as equity.

又或者那些实质上是债务工具、只是披着股权外衣的东西。

They're really locking in a return.

本质上就是把回报锁死了。

Participating preferences, multiple liquidation preferences, ratchets, all sorts of kind of quote structure.

参与分配优先权、多倍清算优先权、棘轮条款，各种所谓的“结构性条款”。

Yeah, yeah, exactly.

对，对，正是。

The last thing that I think is interesting about late stage capital Is the degree to which it allows companies to keep going whether they should or not.

后期资本还有一点我觉得有意思：它能让公司一直撑下去，不管这家公司该不该撑下去。

And so, I would posit that about half of the unicorns are not worth a billion dollars anymore.

我个人的判断是，大概有一半的独角兽已经不值 10 亿美元了。

In other words, 50 percent of the companies that are valued in the private markets as being worth more than a billion shouldn't be worth more than a billion, and that day of reckoning keeps getting pushed out because of so much capital in the markets and so much Liquidity sort of pushing things ahead.

换句话说，在私募市场上被估到 10 亿美元以上的公司里，有 50% 不该值那么多；而这个清算之日被一再推后，因为市场上钱太多、流动性太足，把一切往前推着走。

So, there's.

所以说，

been big shifts in terms of late stage financing and how you get that done.

后期融资、以及这事儿具体怎么做成，已经发生了很大的变化。

There's also been big shifts in terms of M and A.

并购这块也发生了很大的变化。

What what do you think that, what what do you attribute that to?

你觉得这是——你把这归因于什么？

I think it's maybe three things.

我觉得可能有三点。

uh, One is private market valuations continue to go up.

一是一级市场的估值还在不断往上走。

Second, I think that there is a generation of founders who haven't really made the transition mentally, that often as a company scales, you're moving from a product-centric organization.

第二，我觉得有一代创始人在心态上并没有真正完成转变：公司规模一大，你就要从一个以产品为中心的组织，

To one that's both product and distribution centric.

转向一个既以产品为中心、也以分发为中心的组织。

And it's actually something that uh Marc Andreessen talks about a little bit in the book, and in terms of a discussion that we have where, you know, if you look at the old model, you look at Cisco or Microsoft or any of these companies.

这其实也是 Marc Andreessen 在书里谈到的一点，我们那段对谈里聊过：你看老模式，看 Cisco、Microsoft，这类公司里随便哪一家。

What they would do is they'd.

他们的做法是，他们会

have over than anything else, they'd get distribution due to that, and then what they do is they'd go and buy companies or build things, and push it against that distribution channel, and continue to sort of iterate.

把分发看得比什么都重，靠这个建立起分发能力，然后去买公司、或者自己做东西，推到那条渠道上，再不断迭代。

That also happen with the next wave.

下一波公司也是这样。

I mean Google did that really aggressively with everything from Google Maps to Android to Chrome, Facebook did it.

Google 就把这件事做得非常激进，从 Google Maps 到 Android 到 Chrome 都是，Facebook 也这么干。

And so the traditional model is you buy stuff and you distribute it to your channel, and you realize you're also a distribution company.

所以传统模式就是：你买下东西，通过自己的渠道分发出去，同时意识到自己也是一家分发公司。

And I think that a lot of folks today have less of that mindset of, hey, we're also a distribution company, and therefore we should be buying things aggressively to distribute.

而我觉得今天很多人身上这种心态弱了很多——我们也是一家分发公司，所以应该激进地买东西来分发。

I think a third area is internal pushback.

第三块我觉得是内部的阻力。

But maybe it's stronger now around two things.

不过现在这种阻力可能在两件事上更强了。

So, the argument is often, well you should wait, or we shouldn't buy it because it'll take us two years to integrate it, and often the answer may be don't integrate it.

常见的说法是，我们应该再等等，或者我们不该买，因为整合它要花两年；而答案往往可能是：别整合。

You know, it's okay, maybe if it runs independently, we have a series of APIs talking to each other.

让它独立运行也没什么问题，我们之间用一组 API 互相对接就行。

And then secondly, there's often arguments around comp.

其次，经常会有围绕薪酬的争论。

If we had 10 million dollars, instead of buying this company, we could hire 50 engineers, and therefore, we should hire those engineers.

如果我们有 1000 万美元，与其买下这家公司，不如招 50 个工程师，所以我们应该去招这些工程师。

But it's often really hard to find 50 incremental engineers, and then allocate them to that thing that you'd buy.

但要真的额外招到 50 个工程师，再把他们分到你本来打算买的那件事上，往往非常难。

Often, you allocate them elsewhere.

通常你会把他们分配到别的地方去。

I think I agree with all that, and I would add to it that some of these big tech companies have shifted their salaries so dramatic, like they basically broke, like traditionally incumbents would have sort of salary bands.

这些我都同意，我还想补一点：有些大科技公司把薪酬提得太猛，基本上把老规矩打破了——传统上，在位巨头都有自己的薪酬区间。

And that was one reason startups to leave, and then they would also, then they get re- sort of aquired.

这也是有人离开大公司去创业的一个原因，然后大公司再把他们收购回来。

But now you look at the salaries of companies like Google, some of this came out like the court documents of what they're paying like the Waymo engineers, and it's like you know it's like the price of an annual price of a startup acquisition or something.

但现在你看 Google 这类公司开的薪酬，有些是从法庭文件里流出来的，比如他们付给 Waymo 工程师多少钱，那光一年的薪水就抵得上收购一家创业公司的价钱。

So they've also just.

所以他们也就

Dramatically rethink their pay scales, which has hurt the kind of aqua hire market.

彻底重新想过了自己的薪酬体系，这也打击了人才收购(acquihire)那块市场。

I think there's probably different analysis at different stages, like aqua hire, and then sort of tech tuck ins would be more distribution.

我觉得不同阶段的分析大概不一样，比如人才收购是一类，而那种技术型的 tuck-in 收购更偏分发。

I think the takeaway though for founders, I think it was probably never a good idea to build a company to sell.

不过对创始人的启示，我觉得是：为了卖掉而做一家公司，大概从来就不是个好主意。

But it's even more so not a good idea today.

而今天更加不是个好主意。

You need to first hit product market fit, and then you need to figure out how to scale, and you just have to deal with those issues, right?

你得先做到 product market fit，然后想清楚怎么规模化，那些问题你只能自己去扛，对吧？

Absolutely.

完全同意。

But I also wonder how much of it is cyclical or sort of secular change, right?

但我也在想，这里面有多少是周期性的，有多少是长期的结构性变化？

Is it like a fundamental change in the market?

这算不算市场的根本性变化？

I think you're raising two really good points.

我觉得你提了两个非常好的点。

One is, what's the why now of why to buy a bunch of stuff?

一是，大举收购的 why now 是什么？

And often those mirror technology waves.

而这些 why now 往往和技术浪潮对得上。

And so, since we just went through cloud, mobile, and social as three separate waves, you actually saw companies subsequently buying talent for those areas.

我们刚刚经历了云、移动、社交这三波各自独立的浪潮，随后你确实看到公司在这些方向上买人才。

So, I remember When Twitter bought my startup back in 2009, shortly thereafter, there's a big wave of buying mobile companies, and every company was trying to buy mobile teams, cause iOS was new.

我记得 2009 年 Twitter 收购我那家创业公司，没过多久就来了一大波收购移动公司的潮流，每家公司都想买移动团队，因为 iOS 还是个新东西。

Yeah, yeah.

对，对。

And then it was AI, and then it was you know whatever.

然后是 AI，再然后是别的什么。

So maybe the only area where people are very aggressively buying right now is machine learning.

所以现在大家还在非常激进地买的，可能只剩机器学习这一块。

And so maybe there's a lack of why now.

所以也许问题是缺一个 why now。

I think part of it too is people aren't really hiring corporate development people early in the life cycle of these breakout companies, and of this last batch or last generation.

我觉得还有一部分原因是，这一批、或者说上一代跑出来的公司，在生命周期的早期都没怎么招企业发展(corp dev)的人。

And I think that may also be a reason, like there's nobody who's just thinking about it nonstop.

我觉得这可能也是一个原因：没有一个人是在不停地琢磨这件事的。

And often you see these transitions, or if you hire a great corporate person, selling the company will go and buy a bunch of companies, and It's a little bit chicken or egg, but I actually think having somebody who's thinking about it strategically.

你经常会看到这样的转变：招到一个很强的企业发展负责人，公司就会去买一堆公司，这有点像先有鸡还是先有蛋，但我确实觉得，有一个人在战略层面想这件事，

and can aggregate all the input from the different teams in the company, and take it to the executive team, and say, hey, here's the five things we should buy, and why.

能把公司内部各个团队的输入汇总起来，拿到高管团队面前说，我们该买这五样东西，以及为什么，

Actually makes a big difference, and so Coinbase is a good example, where over the last you know, 12 months, I think they've made five or six acquisitions, and in part that may be driven by them hiring Emilie Choi to come in and You know, really drive that function in a smart way.

这真的会带来很大不同，Coinbase 就是个好例子，过去 12 个月里我记得他们做了五六笔收购，其中一部分原因可能就是他们招了 Emilie Choi 进来，把这个职能真正聪明地做起来。

So, one thing I.

所以，有一件事我

like to think about a lot is sort of where are we in the history of tech?

我特别爱琢磨的一个问题是：我们现在处在科技史的什么位置？

We're still in the middle, I guess, of the sort of mobile revolution, but obviously sort of mobile phones and all that, and cloud computing and SaaS.

我们大概还在移动革命的中段——手机这一摊，还有云计算和 SaaS。

Social, what's the big thing over the next 10 years?

还有社交。那接下来十年，真正的大事会是什么？

Clearly those trends are gonna continue, but will we layer on more trends?

这些趋势显然会延续，但我们会不会再叠上新的趋势？

We do seem recently to be headed for this kind of consolidation around the big 5 tech companies.

最近看起来，行业确实在往五大科技公司集中。

Will that continue?

这会一直持续下去吗？

Will there be new things that kind of come out?

会不会冒出新的东西？

Obviously, you and I both have an interest in crypto as an example, as kind of this new thing that's kind of potentially quote unquote disruptive.

显然，你我都对 crypto 感兴趣，它算是可能带来所谓「颠覆」的那种新东西。

Yeah, it definitely feels in me like there's uh four areas that are really interesting right now.

对，我确实感觉眼下有四个领域特别有意思。

The first to your point is a continuation of the trends of the last decade around cloud and mobile, and then sort of social being a primary consumer platform at this point.

第一个正如你说的，是过去十年围绕云和移动那些趋势的延续，加上社交到今天已经成了主要的消费级平台。

The second area really is the crypto world, and at least for me, a lot of the really interesting areas in that are around forms of money, store of value, privacy tokens, et cetera.

第二个领域是 crypto，至少在我看来，里面真正有意思的方向集中在货币形态、价值储存、隐私代币这些。

As well as sort of the smart contracting world securitization of everything, and then lastly, the NFT or the non-fungible token.

还有智能合约那一块，万物证券化；最后是 NFT，也就是非同质化代币。

So, persistent digital goods that you can mix and match in different ways across Media, games, et cetera.

也就是可长期留存的数字物品，能在媒体、游戏这些场景里以不同方式组合搭配。

So crypto is super interesting.

所以 crypto 非常有意思。

Third area is the sort of systems and semiconductor layer of machine learning.

第三个领域是机器学习的系统层和半导体层。

So if you look at Nvidia, and Nvidia GPUs are really the primary basis.

你看 Nvidia，Nvidia 的 GPU 其实就是最主要的底座。

For a lot of sort of performant machine learning models, and Google obviously launch TPUs a couple years ago, Tensor Processing Units, a specialized ASIC that's really meant to help accelerate certain types of machine learning models as well from a hardware perspective.

很多高性能机器学习模型都跑在它上面；Google 几年前也推出了 TPU，即 Tensor Processing Unit，一种专用 ASIC，同样从硬件层面加速某些类型的机器学习模型。

And so I think that's a really interesting, pretty wide open area, and I think that's pretty transformative, and it's a way to effectively index machine learning.

我觉得这个领域很有意思，格局还远没定，变革性很强，而且它相当于给机器学习提供了一种指数化投资的办法。

Because I think it's very hard to invest in quote, quote, a machine learning company.

因为我认为，直接投一家所谓的「机器学习公司」非常难。

And then lastly, I'm personally very interested in sort of longevity and anti aging technologies, and by that I mean true biofarm.

最后，我个人对长寿和抗衰老技术非常感兴趣，我指的是真正的生物制药。

Which is translating existing science that's existed in labs for 10 20 years, and is trying to turn it into drugs that can extend lifespan pretty dramatically.

也就是把在实验室里躺了一二十年的科学成果转化出来，做成能大幅延长寿命的药物。

So, the continuation of mobile and cloud.

那说说移动和云的延续。

We've gotten through obviously kind of the first wave of that.

第一波我们显然已经走完了。

What would wave two would be the most exciting things about wave two?

第二波会是什么？第二波里最让人兴奋的是什么？

I think there's two or three components of wave two that are really interesting.

我觉得第二波里有两三块特别有意思。

Number one is on the saas side, I still think there's a dozen billion dollar plus companies to be created where you're just taking Something that everybody's building internally over and over again, and you're just turning it into an API.

第一块是 SaaS 这边，我仍然认为还能再诞生十来家十亿美元以上的公司，做法就是把大家都在内部反复重造的东西拿出来，直接做成一个 API。

To factor out a thing that people are building over and over and make it an API.

把大家一遍遍重复造的东西抽出来，做成一个 API。

Exactly, and I think if you took apart a Fortune 500 company and asked what are all the different things that They have to do repetitively, you could come up with a dozen of these.

没错。我觉得你要是把一家《财富》500 强公司拆开，问问他们有哪些事必须重复做，就能列出十几个这样的机会。

So, I think that's super interesting.

所以我觉得这块非常有意思。

I do think there's a lot of services that are sort of mobile first in really interesting ways.

我也确实觉得，有不少服务可以用很有意思的方式做成 mobile-first。

I think there's again reinvention of a lot different areas.

我觉得还有很多领域会再经历一次重新发明。

You and I could talk about crypto for a full podcast or more, so I won't go too much into that.

crypto 这个话题，你我能聊满一整期播客还打不住，所以我这次不往深里说。

But I guess one big question is where do you think we are in the evolution of it?

但有个大问题：你觉得我们现在处在它演进的哪一段？

Like what are the next kind of key milestones that you'd want to see?

接下来你最想看到的关键里程碑是什么？

I think of it as there's two things, there's sort of the prices, and then there's the real innovation.

我把它看成两件事，一边是价格，另一边是真正的创新。

And the real innovation has been strong, but I think there's a lot of things that need to happen.

真正的创新一直很强，但我觉得还有很多事得先发生。

Yeah, it's it's a great point.

是啊，这个点提得很好。

You know, I think the primary use case that is clearly working is some form of store of value slash speculative asset.

我觉得明确跑通的主要用例，是某种形式的价值储存 / 投机资产。

And there's lots of things that are claimed to be working that aren't.

还有一大堆被说成已经跑通、其实并没有的东西。

So, for example, nobody's really building scalable apps on a theorem right now, and I think the theorem community is clear that they need to be able to scale that platform.

比如说，现在其实没人在以太坊上做真正能扩展的应用，而且以太坊社区自己很清楚，必须让这个平台扩得起来。

And so one key thing is infrastructure and scalability, and so a lot of people are obviously focused on that right now.

所以一个关键点就是基础设施和可扩展性，现在显然很多人都在攻这块。

You know, ultimately, I guess the way that I would think about it at a macro level, to just abstract out a level, is it reminds me a lot of the late nineties, where there's a few fundamental core things that people are clearly doing on the internet.

说到底，往上抽一层，我在宏观上的看法是，这很让我想起九十年代末：当时有少数几件根本性的核心事情，人们确实在互联网上做成了。

And then there's tons of speculative stuff that's gonna go to zero.

然后还有海量投机性的东西，最后都会归零。

And so, in the early days of the internet, you needed PayPal to be able to buy things.

互联网早期，你得有 PayPal 才买得了东西。

You needed to be able to start doing commerce, Amazon and eBay and some of the early sites, you had Yahoo.

你得能把电商做起来，Amazon、eBay 和一些早期站点，还有 Yahoo。

And then later, Google for media and things like that.

再往后是 Google，做媒体之类的事。

But most dot com companies didn't work out.

但大多数 dot com 公司都没做成。

The interesting thing is that you basically had a speculative bubble coupled to real value creation.

有意思的地方在于，当时基本上是一场投机泡沫和真实的价值创造捆在一起。

I mean, Amazon the most controversial company that era, of course, is worth far more than its peak back then, so it took a long time to get back there, but yeah.

Amazon，那个年代最有争议的公司，现在的市值当然远远超过当年的峰值，只是花了很久才涨回去。

And then there was new companies like Google that ended up going public much later.

后来又冒出 Google 这样的新公司，上市时间晚得多。

Facebook didn't even exist, and then ended up becoming a major force.

Facebook 当时压根还不存在，后来却成了一股主要力量。

And you know, to your point about earlier, it said you had to have both apps and infrastructure.

照你前面说的，应用和基础设施得同时有。

And in that case, the infrastructure was things like broadband and mobile phones and things like that.

那一轮里，基础设施就是宽带、手机这类东西。

But You know, you couldn't have had Netflix and YouTube and a lot of just the rich experiences that people expect today without a lot of infrastructure build out, right?

没有那一大轮基础设施铺开，就不会有 Netflix、YouTube，也不会有今天大家习以为常的那些丰富体验，对吧？

I mean, Netflix was literally mailing DVDs, right?

Netflix 当年可是真的在寄 DVD，对吧？

And so, it's shifted pretty radically to becoming a major content producer.

后来它转得相当剧烈，变成了一家主要的内容生产方。

And I think in crypto we're gonna see the same thing, where A, most crypto projects from today are gonna go to zero.

我觉得 crypto 也会走一样的路：第一，今天绝大多数 crypto 项目都会归零。

There's a few that are gonna be incredibly valuable, and or D R, and that's gonna grow.

会有少数几个变得极其有价值，而且还会继续长大。

But I think most are gonna go to zero.

但我认为大多数都会归零。

But second I think as you have this technology built out, you're gonna see ideas from today that fail, ten years later work.

第二点是，随着这套技术铺开，你会看到今天失败的点子，十年后跑通了。

There's this game, I think it was called Dot Bomb.

有个游戏，我记得叫 Dot Bomb。

It was a board game that was popular, I don't know, popular, but somewhat popular among certain people in like the early 2000s.

那是一款桌游，当年挺流行的，也说不上流行，反正 2000 年代初在某些人里面还挺受欢迎。

And they had cards that were all of the, they're called bad idea cards, and they're all the bad ideas of uh the nineties.

游戏里有一堆卡片，叫“坏点子卡”，全是九十年代那些坏点子。

And it was literally like, if you go through the cards, it's literally every top.

你要是把那些卡片一张张翻过去，那简直就是今天每一个热门方向。

It was like ride sharing, yeah, like selling pet food online, internet money.

像是网约车，对，还有在网上卖宠物粮、互联网货币。

You know, group deals, like mobile gaming.

还有团购、手机游戏。

It's literally all the quote unquote bad ideas actually turned out to be really good ideas.

那些所谓的“坏点子”，最后真的全都成了非常好的点子。

They were just way ahead of their time and not build right.

只是它们太超前了，当时也没做对。

Yeah, I think we're gonna see the same thing in crypto where there's all these ideas that are early and there's no infrastructure, and then 15 years are gonna be amazing.

是啊，crypto 我觉得也一样，一堆点子出来得太早、没有基础设施撑着，然后 15 年后会变得非常了不起。

And then the third category, uh, machine learning, machine learning, silicon machine learning, that's interesting, cause historically you have some big new trend, and what the startups do, you build infrastructure software, et cetera.

再说第三类，机器学习、机器学习芯片，这个挺有意思，因为历史上每来一波大趋势，创业公司做的事就是搭基础设施软件之类的。

Now all of this stuff is either open source, or like the algorithms are published openly as papers.

而现在这些东西要么是开源的，要么算法都以论文形式公开发表了。

A lot of the code implementing those algorithms is very high quality kind of open source stuff that's in turn monetized through the cloud service providers, right?

实现这些算法的代码，很多是质量非常高的开源项目，而这些项目转过来又由云服务商拿去变现，对吧？

So it's kind of hard, you know, now as a startup you can go and you can take this stuff and you can apply it, so you can build Machine learning powered accounting software, which is a very interesting area.

所以现在挺难的——作为创业公司，你可以把这些东西拿过来直接用，比如做机器学习驱动的会计软件，那是个很有意思的方向。

We've got a bio fund that's doing a lot of AI meets bio investing.

我们有一支生物基金，在做很多 AI 与生物交叉的投资。

But you're right that there's no way if you want to just sort of bet on as an investor or as an entrepreneur, bet on machine learning.

但你说得对，如果你想作为投资人或者创业者，单纯押注机器学习本身，那是没办法做到的。

It's very hard to.

非常难。

Yeah, and I think every major technology wave also had a major silicon company created in it.

对，而且我觉得每一波重大技术浪潮里，都诞生过一家重要的芯片公司。

And right now, to some extent, that's Nvidia for not only graphics processing, but also for machine learning purposes.

眼下这家公司在某种程度上就是 Nvidia，它不只用于图形处理，也用在机器学习上。

But if you look at what a machine learning ASIC really needs, it's really fast I/O.

但你看机器学习 ASIC 真正需要什么，它需要非常快的 I/O。

And then it's a lot of matrix multiplication.

然后是大量的矩阵乘法。

And if you look at the surface of a GPU, that's actually dedicated to those functions is quite small relative to the overall chip.

而你再看 GPU 上真正用于这些功能的那部分面积，相对整块芯片其实相当小。

And so there is room to create things that are ten hundred times better or faster or more performant from a power perspective.

所以还有空间做出好十倍、一百倍的东西：更快，或者从功耗角度看性能更强。

By creating customASIC for ML, and so I think that's really exciting.

办法就是为 ML 做定制 ASIC，我觉得这件事真的很让人兴奋。

Well, the other thing with ML right is the training side, which usually happens in the data center, and then the deployment side, which will happen at some edge device.

另外 ML 还有一点，训练这一侧通常发生在数据中心，而部署这一侧会发生在某个边缘设备上。

And may have you know size and power constraints and other kinds of things that require specialized hardware.

边缘设备可能有尺寸和功耗上的限制，还有别的种种条件，需要专门的硬件。

Right?

对吧？

It's a great point.

这点说得很好。

The inference side of machine learning will have its own specialized chips potentially, and if you look at the companies in the market, they often segment themselves by whether they're involved with the training, the your pointer, or the inference.

机器学习的推理这一侧，可能会有自己的专用芯片；而你看市场上这些公司，它们往往就按自己做的是训练还是推理来给自己划分。

And if you get into a world where you really have broad based self driving cars or robotics or other things, then these chips become really important.

如果真的进入一个自动驾驶汽车、机器人大范围铺开的世界，这些芯片就变得非常重要了。

And then your fourth category of longevity.

然后是你说的第四类，长寿。

There's 20 years of evidence.

这方面已经积累了 20 年的证据。

The aging is a biological process you can perturb, and there's three or four main lines of evidence.

衰老是一个可以干预的生物学过程，主要的证据线有三四条。

Number one, there's caloric restriction.

第一，热量限制。

So, if you calorically restrict, you extend lifespan in multiple organisms.

做热量限制，在多种生物体里都能延长寿命。

Two is, there's certain genes that if you knock out in model organisms They'll live two, three times longer, they'll be healthier adults the whole time, and then they'll crash out.

第二，有些基因在模式生物里敲掉之后，寿命能延长两到三倍，而且全程都是更健康的成体，最后才垮掉。

So for example, in C elegans, there's a pathway called the DAF pathway that is involved in regulating lifespan.

比如在秀丽隐杆线虫里，有一条叫 DAF 的通路，参与调控寿命。

And I actually worked on that directly for my PhD.

我读博士时做的就是这个方向。

Third, there's what's known as parabiotic, which is exchange young blood into old animals and vice versa.

第三，是所谓的联体共生(parabiosis)，把年轻血液换进老年动物体内，反过来也一样。

And it's clear that you can increase the rejuvenative capacity of older animals by giving them young blood.

很明确的一点是，给老年动物输年轻血液，能提升它们恢复年轻的能力。

And actually, it looks like those factors in old blood that are screwing up functions.

而且看起来，真正把功能搞坏的是老年血液里的那些因子。

So if you give old blood to a younger animal It'll actually be less functional in terms of regenerating a damaged liver or muscle or the like.

所以把老年血液输给一只更年轻的动物，它在修复受损肝脏、肌肉这类组织时，反而会变差。

And then lastly, there's at least two FDA approved drugs that in multiple organisms will extend life span between 5 and 30 percent.

最后一条，至少有两种 FDA 批准的药物，在多种生物体里能把寿命延长 5% 到 30%。

One of them is rapamycin, which in mice, for example, extend life span 10~30 percent, and it also extends life span in flies and worms and other organism, so it's evolutionarily conserved.

一种是雷帕霉素，在小鼠身上能把寿命延长 10~30%，在果蝇、线虫和其他生物体里同样能延长寿命，所以它在进化上是保守的。

The other one is metformin, which is the tenth most prescribed drug in the world, the anti-diabetes drug.

另一种是二甲双胍，全世界处方量排第十的药，治糖尿病的。

So, it does look like there's clearly things that can modify the pace at which different types of organisms, including mammals, actually age.

所以看起来确实有些东西能改变不同类型生物体的衰老速度，哺乳动物也包括在内。

I think a lot of people think that's kind of science fiction or something.

我觉得很多人会把这当成科幻之类的东西。

They make jokes about the blood transfusions from young people or whatever.

他们会拿抽年轻人的血来输开玩笑。

I think it's as near term as any biotech product is in some sense, because it takes time to develop.

某种意义上，我觉得它和任何一款生物技术产品一样近，因为开发本来就要时间。

There's a lot of factors that have been cloned out of blood that are actually really valuable drugs today.

从血液中克隆出来的因子里，有不少今天已经是非常有价值的药物。

Two examples of that would be insulin, which again is just a factor in blood which impacts a biological process that People now inject as a standalone.

举两个例子，一个是胰岛素，它本来也就是血液里的一个因子，影响某个生物过程，现在人们把它单独拿出来注射。

And then there's epoGen, which is Amgen's big blockbuster drug, which helps you generate red blood cells.

另一个是 epoGen，Amgen 的重磅药，帮你生成红细胞。

And similarly, the way that was discovered was They literally looked for what's the factor in blood that lets anemic cows be non-anemic, as something that have red blood cells.

同样地，它当年被发现的方式，就是去找血液里到底哪个因子能让贫血的牛不再贫血、能长出红细胞。

And so, these aging molecules are the same thing.

这些跟衰老有关的分子也是一回事。

It's not Some magical thing about blood is just there's proteins floating around, and maybe some of them do good stuff.

血液本身没什么神奇的，不过是里面漂着一堆蛋白质，也许其中有些能干点好事。

The hard part has been that because you're looking at things like aging and longevity, it takes a really long time to measure it.

难的地方在于，你研究的是衰老和长寿这类东西，测量起来要花非常长的时间。

And so, the pace at which people have been innovating on these things is slow.

所以在这些方向上，大家的创新节奏一直很慢。

But also bio farm in general tends to be focused on oncology or cancer, and a few other areas.

而且生物制药整体上更愿意聚焦肿瘤学、癌症，再加上另外少数几个领域。

And they're not really thinking about how can I push disease back in general, you know, with a drug.

他们并没有真的在想：我怎么用一款药，从整体上把疾病往后推。

So, I do think there's a big market gap, which creates a really interesting opportunity.

所以我确实觉得这里有一块很大的市场空白，这就带来了一个非常有意思的机会。

What do you think about the societal implications of if we do develop longevity technology?

如果我们真做出了长寿技术，你怎么看它的社会影响？

The societal implications are massive, and I think there's all sorts of ethical conundrums, because if you look at the Kuhnian paradigm of scientific change, it basically says you need the old guard to die.

社会影响是巨大的，而且我觉得会冒出各种伦理难题，因为按科学变革的库恩范式，它基本上说的是：得等老一代人死掉。

In order for scientific progress to happen, and so imagine if no professor at any university ever retired, or imagine if any of these characters who you know have had a very negative impact on their country around the world.

科学进步才发生得了。所以你想象一下，如果大学里没有一个教授会退休；或者想象一下，世界各地那些给自己国家带来极大负面影响的人物。

You could flip it around, of course, and find all the positive characters.

当然，你也可以反过来看，把正面的那些一条条列出来。

Sure.

当然。

Yeah, yeah, yeah.

对对对。

But it does lock in societal structures, and it does lock in power structures.

但它确实会把社会结构锁死，也会把权力结构锁死。

And so, from that perspective, one can argue that it could really hurt self renewal in a global system.

从这个角度说，可以论证它会严重伤害整个体系的自我更新。

And therefore, is it you know, quote unquote bad?

那它算不算所谓的“坏事”？

The other sort of really interesting societal dynamic is around how you think about your own life.

另一个特别有意思的社会效应，是你怎么看待自己这一生。

So say that you had another 100 productive years, where you basically thirty something for 100 years.

假设你多出 100 年的高产岁月，基本上就是以三十几岁的状态活上 100 年。

You'd think about marriage and having kids differently.

你对结婚、生孩子的想法就会不一样。

When would you do that?

什么时候做这些事？

Would you do it early?

早点做？

Would you do it late?

还是晚点做？

Would you do it multiple times?

还是做好几次？

People would they'd stay in grad school to their like ⟨?…⟩.

人会一直读研究生，读到⟨?…⟩。

Yeah, I mean there are already people who do that.

对，现在就已经有人这么干了。

So the people who stay in grad school to their forties, where they stay ⟨?till they're⟩ 140.

现在有人研究生一直读到四十多岁，将来就该读到 140 岁了。

Do you try to make money early in life so you could spend it through the rest of your life?

你是想早早赚够钱，好用余生慢慢花？

Or do you wait until late and gather experiences?

还是拖到很晚再说，先去攒各种经历？

Do you start thinking about having multiple careers over the course of your life, and you also think about social mobility?

你会不会开始考虑一辈子做好几段职业？还有社会流动性也得一起想。

So You know, if you're disadvantaged early, does that extend for 200 years?

如果你早年处境不利，这份不利会不会延续 200 年？

And what does that mean?

那又意味着什么？

And so, I just think there's these really deep societal implications of the ability to extend life span.

所以我觉得，延长寿命这件事，对社会的影响非常深层。

Fascinating.

太有意思了。

Well, great.

好，那就到这儿。

Well, thanks a lot for joining us, Elad.

非常感谢你来我们节目，Elad。

Elad's the author of the new book, High Growth Handbook.

Elad 是新书《High Growth Handbook》的作者。

Great, thanks so much.

好的，非常感谢。

Thanks so much having me.

谢谢你们请我来。