马丁·卡萨多谈AI背后的需求驱动力

如果由代理来做决策，中央采购团队、平台团队和IT团队会怎么样？

What's gonna happen to central buyers and platform teams and IT teams if agents are making the decision?

很明显，编程基本上已经过时了，但工程领域却远未过时。

It's very clear that coding is pretty much dead, but engineering is very much not.

每次技术时代更迭，你都得重新做一遍所有事情，而我们每次都忘了这一点。

Every time you have a technical epoch, you have to redo everything, and we forget that every time.

我认为人们甚至对‘泡沫’都没有一个共同的定义。

I don't think people even have a common definition of a bubble.

如果AI的需求是真实且在加速的，为什么一切仍然感觉受限？

If AI demand is real and accelerating, why does everything still feel constrained?

为什么一种明显在创造价值的技术，却比预期更难扩展？

Why does a technology that's clearly delivering value also feel harder to scale than expected?

我们以前见过这种模式。

We've seen this pattern before.

在早期的技术转型中，人们很容易认为难题已经解决了。

In early technology shifts, it was easy to assume the hard problems were solved.

基础设施曾被视为已经完成，但随后使用量激增。

Infrastructure was treated as finished, then usage surged.

为较小规模世界构建的系统开始失效。

Systems built for a smaller world began to fail.

网络承受压力，电力、物理空间和协调再次成为首要限制因素。

Networks strained, power, physical footprint, and coordination became first order constraints again.

每一个新的技术时代都迫使整个技术栈重新构建。

Each new technical epoch forced a rebuilding of the stack.

AI 正在创造如今的这一时刻。

AI is creating that moment now.

需求并非投机性的。

The demand is not speculative.

公司正在部署模型，预算正在转移。

Companies are deploying models, budgets are moving.

真实的生产力提升已经显现，但系统的几乎每一个部分都显得紧张。

Real productivity gains are already showing up, and yet nearly every part of the system feels tight.

计算资源稀缺，数据中心的审批和建设需要数年时间，电力难以保障。

Compute is scarce, data centers take years to permit and build, power is difficult to secure.

监管的进展远慢于技术本身的发展。

Regulation moves far more slowly than the technology itself.

这催生了两种主流观点。

This has led to two dominant stories.

一种认为我们正处于人工智能泡沫中。

One says we're in an AI bubble.

另一种则认为规模扩大将解决所有问题。

The other assumes scale will smooth everything out.

但这两种观点都无法完全解释当前发生的情况。

Neither fully explains what's happening.

需求持续超过供给，最大的瓶颈越来越多地出现在模型之外。

Demand continues to outpace supply, and the biggest bottlenecks increasingly sit outside the models themselves.

这一点在企业软件中尤为明显。

This is especially visible in enterprise software.

人工智能常被视为对SaaS的威胁，但SaaS的难点从来不在界面。

AI is often framed as a threat to SaaS, but SaaS was never hard because of the interface.

它的难点在于它编码了业务流程、合规要求和运营现实。

It was hard because it encodes business processes, compliance, and operational reality.

这些需求并不会消失。

Those needs do not disappear.

变化的是人类以及日益增多的代理如何与这些系统交互，以及软件的定价、购买和控制方式。

What changes is how humans and increasingly agents interact with those systems and how software is priced, bought, and controlled.

这种转变引发了一个更深层的问题。

That shift raises a deeper question.

如果代理正在编写代码、配置基础设施并选择工具，那么究竟是谁在做决策？

If agents are writing code, provisioning infrastructure, and selecting tools, who's actually making the decision?

当这个决策层变得不那么明显时，会发生什么？

And what happens when that decision making layer becomes less visible.

这场对话有助于厘清真正的瓶颈在哪里，以及为什么基础设施并没有淡出背景，反而重新回到了故事的中心。

This conversation helps clarify where the real constraints are and why infrastructure is not fading into the background but moving back to the center of the story.

这是来自65播客的音频片段，内容是十六Z资本的普通合伙人马丁·卡萨多与帕特里克·穆尔黑德和丹尼尔·纽曼的对话。

This is feed drop from the six five podcast featuring a sixteen z general partner, Martin Casado, in conversation with Patrick Moorhead and Daniel Newman.

我们私下聊聊吧。

Let's go off the record.

我知道我们不太常这么做，但当我们有机会邀请一位能真正改变我们对话方向的人，帕特，或者只是带来一些有趣想法和话题的人时，我们总是很乐意这样做。

I know we don't do these as often as we probably like, but when we have the opportunity to bring someone in that can really change the trajectory of the conversation here, Pat, or just someone that's got really interesting ideas and things to talk about, I know we love to do that.

今天我们就请到了一位，你之前在做专业建模和主持时见过他。

And we got one today that you met when you were doing your professional modeling and hosting.

那挺有趣的。

That was fun.

对我来说，坐在那里看着你真的很有趣，因为我看到你在GTC连续两天都在忙碌、流汗。

It was really fun for me to sit there and watch you because I saw you working and sweating at GTC for like two days.

来，你来介绍一下我们的嘉宾吧，我今天非常期待能请到他们来参加我们的播客。

Let's have you introduce our guests, which I'm super excited to have here today on the pod.

是的，很棒。

Yeah, it's great.

我很荣幸向大家介绍，今天加入我们的是A16Z的普通合伙人马丁·卡萨多。

So I'm proud to introduce here, Martin Casado, general partner at A16Z joining us today.

马丁，很高兴见到你。

Martin, it's great to see you.

很高兴来到这里。

Great to be here.

非常感谢你们的邀请。

Thanks so much for the invite.

是的，灯光没那么热，但观众实际上比我们在华盛顿特区GTC现场的人数还要多。

Yeah, the lights aren't as hot, but the audience is actually bigger than the in person audience that we had in Washington DC for GTC.

我真的很感激。

I really appreciated it.

不过，我们何不先确认一下我们正在和谁交谈？

But hey, why don't we start off just to make sure we know who we're talking to?

我们知道您很有名，但对于那些还不了解您的人，您主要从事什么工作？您的投资组合和关注重点是什么？

We know you're famous, but for those who don't know who you are, what do you do and what does your portfolio, what do you focus on?

是的

Yeah.

我是安德里森·霍罗威茨的普通合伙人。

So I'm a general partner at Andreessen Verowitz.

我负责基础设施基金。

I run the infrastructure fund.

我们的基础设施定义是计算机科学基础设施。

So the infrastructure fund So our definition of infrastructure, this is computer science infrastructure.

这是任何有技术购买方的产品。

This is anything with a technical buyer.

比如计算、网络、存储、数据库，当然还有底层的AI技术、开发工具、安全等领域。

So think compute network storage, databases, of course, any of the low level AI stuff, dev tools, that sort of thing, security.

我带领这个团队。

So I run the team.

我在这里已经十年了，之前我实际上是一位投资组合创始人。

I've been here for ten years, and then prior to that I was actually portfolio founder

对于A16z。

for A16z.

这太棒了。

Now that's wonderful.

是不是很有趣，五年前基础设施几乎被所有人抛弃了？

Isn't it funny how infrastructure was pretty much left for dead about five years ago?

硬件是无差别的商品，这是个梗。

Meme of hardware is an undifferentiated commodity.

但看看我们现在，我知道基础设施不仅包括硬件，也包括软件。

And look at us now, and I know infrastructure spans not just hardware, but also software.

是的。

Yeah.

是的。

Yeah.

当然。

For sure.

所以听好了，这一直是一个不断演变的故事。

So listen, it's been an evolving tale.

我的意思是，硬件历史上其实一直相当枯燥，比如网络和芯片。

I mean, hardware has historically actually been pretty boring, like networking, silicon.

现在，软件基础设施，尤其是在数据领域，已经变得非常令人兴奋。

Now software infrastructure, especially in data, has been pretty exciting.

想想Snowflake或者Databricks。

Think like Snowflake or Databricks.

所以，你知道的，已经有很多令人兴奋的进展，比如GitHub。

And so, like, there has been a lot of excite you know, GitHub.

但人工智能彻底颠覆了一切。

But AI has blown everything up.

我不记得上一次人们对一块硅芯片如此兴奋是什么时候了，NVIDIA收购了Rock，等等，它没收购Rock。

I don't remember the last time you had a lot of excitement around a silicon chip and NVIDIA just bought Rock for doesn't buy Rock.

他们聘用了Rock团队。

They hired the Rock team.

没错。

Exactly.

买了他们的

Bought their

自己的钱。

own money.

我们在购买它。

We're buying it.

你知道，我们看到网络公司再次获得融资，因为AI需要新的网络架构。

You know, we're seeing that we're seeing networking companies get funded again because AI requires new networking fabrics.

所以现在又变得非常令人兴奋了，就像早期互联网时代那样。

And so times are very exciting again, you know, kind of early Internet ask.

还有马丁。

And and Martin.

是的。

Yeah.

顺便说一下，要给帕特和我自己一点赞，你知道，帕特和我都曾为Grok提供过建议。

By the way, just a little props to Pat and maybe myself, but you know, both Pat and I actually did advise Grok.

当我们早在2021年做这个决定时，帕特，2021年、2022年，我们决定拿他们的股票作为回报，而不是像租客那样，你知道，我们不是那种很酷的家伙，但我们确实看清楚了局势。

And we decided when we did this back in '21, Pat, '21, '22, we decided to take stock from them instead So of we're not like lessees and cool, quite like, you know, we're not the cool, cool kids, but we actually saw what was going on.

因为帕特和我在2019年就经常开玩笑说，硅基产业将在未来十年吞噬世界。

Because Pat and I laughed like in 2019, we would make this joke, Silicon will eat the world in the next decade.

我们一直说，硅基产业会吞噬世界，半导体将吞噬世界。

We kept saying Silicon will eat the world, semiconductors will eat the world.

当时有些记者说，别再写关于芯片的评论文章了，我们不会报道芯片相关的内容。

And we had journalists that would say, don't write any more op eds about chips, or we aren't going to cover chips.

没错。

Exactly.

不。

No.

就在五年前，他们还说，我们根本不想听这些事。

Just five years ago, they were like, we're not even we don't even want to hear about it.

是的

Yeah.

对

Yeah.

我的意思是

I mean,

事实证明，每一个技术时代都需要你重新构建整个技术栈。

it just turns out every technical epoch requires you to redo the entire stack.

我们甚至在5G时代就看到了这一点。

We actually saw this even with, like, five g.

比如，许多英特尔处理器，像至强处理器，实际上被用于5G。

Like, a lot of the Intel processors like the Xeon was, like, actually used for five g.

我们在数据中心也看到了这种情况。

We saw this with the data center.

这就是你看到网络革命的时候。

This is when you saw network revolution.

所以，你知道，我之前从事过软件定义网络的工作。

So I, you know, I I worked in software defined networking.

那是我们重新设计网络的时候。

That was was my We redid the network.

当然，你也能看到互联网带来的变化，它催生了思科和瞻博网络。

You saw this, of course, with the internet, which gave rise to Cisco and Juniper.

所以每次技术时代更迭，你都得重做一切，而我们每次都忘了这一点。

So every time you have a technical epoch, you have to redo everything, and we forget that every time.

所以当你走到上一个时代的尽头时，我们会说‘硬件已经死了’之类的，然后下一波浪潮来临时，我们又得重新构建整个技术栈。

So when you get to the end of the last epoch, we're like, Oh, hardware's dead, or whatever, and then the next wave comes and we've got get back to building the stack.

所以我必须问你一下，因为你说的这些，我觉得我以前看过你谈过类似的观点，但我们现在是不是正处于AI泡沫中，马丁？

So I have to ask you just because, I mean, everything you're saying here, I think I've seen you speak about this a bit, but are we in this AI bubble here, Martin?

泡沫这种东西很难判断。

Well, bubbles are tough.

是的。

Yeah.

我认为人们连‘泡沫’的共同定义都没有。

I don't think people even have a common definition of a bubble.

让我这么说吧。

Here here's what I will say.

从生产力的角度来看，需求是真实的。

From a productivity standpoint, demand is real.

有真实的用户在支付真实的金钱，获得真实的价值，这一点非常明确。

You have real users paying real money, getting real value, and that's incredibly clear.

数据再清楚不过了。

The data couldn't be more clear.

那么，是否存在需求泡沫，即需求会从哪里来？

So is there a demand bubble, meaning where demand will come?

我们是否面临供给过剩，即我们在建设或期待需求出现？

Do we have a supply overhang where we're building out or hoping it'll come?

没有。

No.

答案绝对是否定的。

The answer is absolutely not.

我们并没有供应过剩。

We do not have a supply overhang.

我们有的是供应不足。

We have a supply underhang.

需求是非常真实的。

The demand is very real.

从投机的角度来看，如果逐笔交易来看，确实有些交易被高估了，但有些交易却被低估了。

Now speculatively, if you look on a deal by deal basis, sure, some deals are overvalued, but some deals are undervalued.

我在十年投资和三十年科技行业经验中学到的是，从长期和整体来看，市场实际上是非常理性的，但分布并不均匀。

So what I've learned in ten years of investing, thirty years in tech, is that markets are actually very rational in the long term and broadly, but it's uneven.

因此，取决于你如何观察以及如何眯眼去看，你会看到一些看似高估和一些看似低估的东西。

So depending on how you look at it and how you squint, you're going to see things that seem overvalued and things seem undervalued.

但我要说的是，如果从整体来看，我真正的信念是，从长期来看，所有这些都被低估了。

But I would say if you take it all in, my true belief is it's all undervalued in the long term.

这些东西极具颠覆性。

This stuff is so, so disruptive.

需求非常真实。

Demand is so real.

它的变现能力非常强。

It's monetizing so well.

它推动了如此大规模的建设，我们都应该对此感到无比兴奋

It's driving so much build out that we should all be incredibly excited for

的未来。

the future.

我本想给这家伙击个掌，因为我是

I was going give this guy a high five because I'm

我是丹尼尔，我们绝对参与其中，丹尼尔和我几个月来唯一被问到的问题就是这个。

the Daniel and I are absolutely in there and Daniel and I do a That's lot of literally the only question we got for months.

丹尼尔，你对此有什么看法？

Daniel, what are your thoughts on that?

我知道他们的立场非常一致。

I know they're very much aligned.

我觉得我这周已经做了第五十次电视访谈，每次都被人问：我们是不是处在AI泡沫中？

I think I did my fiftieth TV segment this week where I got asked the question, are we in an AI bubble?

所以我觉得，说实话，我已经厌倦了回答这个问题。

And so I think, you know, I'm so tired of answering it.

你的公关人员告诉你，当他们问你这个问题时，你不能笑，因为那样会显得他们在问一个愚蠢的问题。

You know, my comms guy said to me, you can't laugh at them when they ask you this, because you make it look like they're asking a stupid question.

那并不能说明什么。

That doesn't say what.

我知道，但我只是不断说，你看，现在整个供应链的每个环节都受到限制。

I know, but like, I just keep saying like, look, we're constrained in every part of the supply chain right now.

我们的需求远远超过供应。

Like we have so much more demand than supply.

目前还有一个关于TAM规模的变现问题尚未解决，即除了LLM订阅之外的市场空间。

Now there's a monetization question that hasn't been answered yet about the size of the TAM beyond subscriptions to LLMs.

我们现在来谈企业领域吧，因为Maeve正好为你引出了这个话题，Martine。

Like, you know, we'll get into enterprise right now, because Maeve is a perfect segue here for you, Martine.

AI对软件的颠覆。

Like the AI's disruption of software.

目前每个SaaS公司都在添加智能代理。

So right now every SaaS company is adding agents.

但与此同时，我桌上开着五个代码卡死的终端，根本没法写代码。

But at the same time, like I've got five terminals of clogged code opened on my desk and I can't program for crap.

我不是程序员，但我现在确实在IDE里构建东西、探索能做什么。

I'm not a programmer, but I am building stuff and playing inside of, you know, IDEs now and actually seeing what can be done.

我们正在构建自己的应用，能替代以前CRM和ERP的功能，还能支持项目管理。

And we're building our own applications that can do things that our CRM used to do and our ERP did, and that enables project management.

我的意思是，软件会不会才是这里最大的风险？

I mean, is software maybe the actual biggest risk here?

真正的颠覆是不是发生在企业现在能靠少数聪明人快速构建任何东西的时候？

Is that where the real disruption is happening is now that enterprises can sort of build anything really quickly with a few smart people?

这会不会就是泡沫破裂的地方呢？

Is that maybe where we're going to see the the bubble pop?

是的。

Yeah.

所以，这真是一个非常及时的对话。

So let so, you know, it's very it's a very timely conversation.

因此，我们公司，特别是我的团队，在上周对此进行了深入研究。

So we actually, as a firm, my team in particular, did a deep dive on this in the last week.

我们实际上掌握了很多数据。

We actually have a lot of data.

在开始之前，我要先说明，现在还为时过早。

Let me just predicate this entire thing by saying it's very early.

因此，我接下来要说的很多内容都只是推测，但我认为我们已经走了几年，还是可以说出一些看法的。

And so, a lot of what I'm going say is purely speculative, but I do think that we are a couple of years in and so you can say a few things.

那我就简单说几点吧。

So, me just say a few things on this.

所以，第一点是，编程几乎已经死亡，但工程却依然非常重要。

So, the first one is it's very clear that coding is pretty much dead, but engineering is very much not.

因此，很明显，门槛已经降低，人人都能成为开发者，但几乎没有任何迹象表明上限也被降低了。

And so you can clearly say the floor has been lowered, so everybody becomes a developer, there's almost no indication that the ceiling has also been lowered.

事实上，最积极使用人工智能的公司，也是招聘人数最多的公司。

In fact, the companies that are the most aggressively using AI are also hiring the most.

那么问题来了，如何调和这些看似矛盾的现象？

And so the question is, how do you reconcile these things?

即使是最先进的AI编程工具，也无法解决许多问题。

There are many things that even the best version of AI coding can't solve.

它在处理大型、复杂且稳定的软件代码库方面表现不佳。

It's not very good at solving large, complex, stable software code bases.

它对运维没有任何帮助。

It doesn't do anything with operations.

如今，写软件实际上就是运行软件，因为所有东西都是SaaS。

Let's say these days actually writing software is running software because it's all SaaS.

运维并不是一个已解决的问题，因为我们还没有掌控它，没有建立起对它的监控和根据其表现进行优化的机制。

The operations is not not a a solved problem because we haven't control we we haven't closed that control of it, like, watching what it does and then refining based on what it does.

如果你从资金投入的角度来看，进入AI编码领域的大部分资金实际上来自专业程序员，而不是业余程序员。

And if you actually look dollar weighted, the the majority of dollars that are coming in on AI coding are professional coders, not casual coders.

没错。

That's right.

因此，我相信这扩大了能编程的人群范围，这将需要更多的代码，从而意味着更多的运维工作。

And so my my belief is this is widening the aperture of the people that can code, which is gonna require more code, which means more operations.

所以，这个领域变得大得多。

And so the tent gets a lot bigger.

我认为真正的上限其实是提高了。

I think the ceiling actually goes up.

它并没有降低，因为现在的问题变得难得多。

It doesn't come down because now the problem has become much harder.

你将拥有专业的开发人员和工程师，他们会投入工作，同时还会有一批新的编程者加入进来。

You're gonna have professional developers and engineers, and they're gonna be put to work, and then you're gonna have a bunch of new coders that are coming in.

还有一个独立的问题，那就是这对SaaS意味着什么？

There's a separate question, which is what does this do for SaaS?

我想在下一个问题之前先提一点：我已经在基础设施和企业领域投资了十年。

I I just wanna submit something before the next question, which is I have been investing in infrastructure and enterprise for ten years.

我告诉你，SaaS从来就不是一个技术问题。

I will tell you, SaaS has never been a technology problem ever.

如果你要开发一个SaaS应用，这并不难。

It's just not if you build a SaaS app, it's not hard.

它从来就不难。

It's never been hard.

所以问题来了，为什么人们会购买SaaS？

And so the question is why do people buy SaaS?

答案是：你购买的是一个业务流程。

And the answer is you're buying a business process.

这是一个被其他公司理解并用来指导你如何运营业务的业务流程。

It's a business process that's been understood by another company that tells you how to run your business.

这从来就不是关于技术或软件的。

It's never been about the technology or the software.

所以我认为这也没有太大改变这种格局。

So I don't think it changes that dynamic much either.

是的，你提到这一点真的很有趣。

Yeah, it's really interesting you brought that up.

我大概两年前半前参加Google Cloud Next大会时，他们展示了一张幻灯片，上面写着营销代理、财务代理。

I was sitting in Google Cloud Next probably two and a half years ago, and they put up a slide, agents for marketing, agents for finance.

我当时在将企业级SaaS公司与这些领域的营销、人力资源等对应起来，并能看到这些领域的领军企业。

And I was aligning enterprise SaaS companies with these down the line marketing, human resources, and you could see the leaders in these businesses.

有趣的是，从市场角度来看——我知道你接触的是不同的市场，但就公开市场而言，估值表现并不好，可能除了ServiceNow这样的公司之外。

And what's interesting is we have seen, at least from a market's perspective, I know you deal with a different market, but the public markets, the valuations are not doing well, probably with the exception of folks like ServiceNow.

你已经看到Salesforce、Workday等公司出现了下滑。

You've seen folks like Salesforce and Workday and folks like that, a decline here.

但请你谈谈你的观点，再次强调，你永远不可能拥有完美的数据，对吧？

But talk me through your thesis on, again, you'll never have the perfect data, right?

你必须先做好数据管理，但我确实认为人工智能会对此有所帮助。

You have to have your data management in place, but I do think AI will help with that.

因此，你需要一个企业可访问的数据架构，这些代理可以在正确的安全性和访问权限下与该数据架构交互。

So you have a data fabric that's accessible by the enterprise, and these agents can hit that data fabric with the right security, the right access.

跟我谈谈你对这一趋势如何发展的看法。

Talk to me about your thesis of how this rolls out.

我们已经看到马克·贝尼奥夫和萨提亚·纳德拉之间发生了一些公开的争执，似乎持续了一年。

We've seen some public battles between Mark Benioff and Satya Nadella, right, that seem to go on for a year.

所以系统中确实存在紧张关系。

So there is tension in the system for sure.

我经常组织CIO圆桌会议，他们明确告诉我：我们正在寻找办法摆脱这家软件供应商。

And I do a lot of CIO roundtables and they're literally telling me, We are finding a way to get off of this software vendor.

是的。

Yeah.

所以，听我说，让我从历史的角度给你一个更清醒的视角，好吗？

So listen, let me just provide kind of maybe the historical sober view of this, right?

为什么传统公司的估值和增长较低？

So why are valuations and growth lower from traditional companies?

因为我们正经历自互联网以来最大的预算转移。

It's because we're seeing the largest movement in budget we've seen since the internet.

当预算转移时，它会流向新的领域。

And when budget moves, it goes to new places.

我们正在看到传统软件的大规模替代吗？

Are we seeing mass replacements of traditional software?

不，我们没有，对吧？

No, we're not, Right?

因此，从历史上看，这通常不是零和游戏。

And so historically, tends to get it's not zero sum.

它往往会被叠加，或者预算发生转移，或者增长放缓，但通常不会被替代。

It tends to get layered or budget will move or it tends to slow down, but it doesn't tend to get replaced.

那么，你为什么要开始替代像核心系统这样的东西呢？

Now why would you start replacing something like a system of record?

答案是它没有跟上新技术的发展。

Well, answer is it doesn't evolve with the new technology.

意思是，有一些公司没有完成向互联网的转型，它们本可以做到，只是选择不去做。

Meaning, there were a number of companies that didn't make the internet transition, and they could have, they just decided not to.

因此，作为消费者，我们会逐渐改变对如何与软件互动的看法。

So we as consumers are going to evolve our opinion on what it means to interact with software.

比如，当我使用Salesforce时，我会希望召唤它并与它对话。

When I interact with Salesforce, for example, I'm going to want to call it up and talk to it.

我会希望拥有一个大语言模型。

I'm going want to have an LLM.

但这只是使用方式的改变。

But this is a consumption layer change.

这并不是根本性的变革。

It's not a fundamental change.

业务流程依然存在。

The business process is still there.

这些保障仍然存在。

The guarantees are still there.

因此，Salesforce 完全有机会将使用层演变为符合用户期望的形式，所以我不认为会有什么端到端的根本性变化彻底颠覆整个 SaaS 行业。

So Salesforce 100% has the opportunity to evolve the consumption layer to be what the expectations of the user So I don't think there's something fundamental end to end that's going to disrupt all of SaaS.

话虽如此，SaaS 提供商仍需与时俱进，以应对不断变化的用户期望和预算。

That said, it's up to the SaaS providers to evolve to meet these moving expectations and this moving budget.

因此，当我思考未来会发生什么时，我仍然以互联网作为我的参照范例。

So again, I use the internet as my anchor example when thinking through what's going to happen.

是的，我觉得这一点也非常有意思。

Yeah, think that's really interesting though too.

我认为你所说的最具前瞻性，那就是体验必须改变。

I think what you said is the most prescient, which is the experience has to change.

我认为很多人的想法都和你一样，作为企业主，帕特，你大概也和我有同样的感受，这些业务系统本应帮助你经营企业，但实际上却并不容易使用。

And the reason that I think a lot of people, and I can say this as business owners, Pat, you share this probably sentiment with me, is these systems of records are supposed to help you run your business, but they're actually not that easy to use.

你要雇多少人，才能做到：我想对比去年和今年的业务数据，用这种切割方式、这种维度、这种角度来做报表？

How many people you have to hire to be like, oh, I want to report on last year's business versus this year's business with this cut and this slice and this angle.

你还需要专门派人坐在你桌边帮你。

And you need like a special person to like come sit at your desk with you.

或者多亏了SaaS，他们可以在另一张桌子上完成。

Or they can, thanks to SAS, they can do it in another desk.

但关键是，要生成一份报告。

But the point is, and drum up a report.

而实际上，我们现在用ChatGPT或Anthropic，只需像问你问题一样直接提问即可。

When in reality, what we can do with ChechiPT or with Anthropic now is we just ask it a question the same way I'd ask you a question.

如果它能通过API和数据源找到相关信息，就会直接输出结果。

And if it can actually go through the APIs and through the traps and find the data, it will spit out.

你可以说，我要一个饼图。

You can say, I want it in a pie chart.

是的。

Yeah.

你还可以告诉它，我要一个柱状图，或者嘿。

And you tell it, oh, I want it in a bar chart or, hey.

我希望它能以一种像埃里克·克莱普顿演唱《天堂的眼泪》那样的叙事风格呈现出来，它真的能做到。

I want it in a narrative that sounds like Eric Clapton singing singing, you know, tears in heaven, and it can do that for you.

没错。

And Right.

但让我，让我

But but me let me

我只是想提供另一个角度，我同意你所说的，但让我们看看互联网。

just provide the other the other side of so I agree with everything you're saying, But let's look at the Internet.

对吧？

Right?

那么，互联网提供了什么？

So what did the Internet provide?

它让我能够从家里连接到软件，比如。

It provided me the ability to, like, connect to software from my house, for example.

所以，它带来了使用和访问上的流畅性。

So, you know, so it it provided this fluidity with consumption with access.

人工智能做什么？

What does AI do?

它做同样的事情。

It does the same thing.

比如，我可以和它交谈。

Like, I can talk to it.

我可以使用自然语言。

I can use natural languages.

但你仍然需要应对所有的合规要求。

But you still have all of the compliance.

你仍然需要处理所有的集成。

You still have all of the integrations.

还有正式的报告流程。

There's formal reporting things.

此外，还有上面的业务流程。

There's also the business process on top.

我仍然需要进行管道审查。

I still need to do pipeline reviews.

我仍然需要进行汇总。

I still need to do roll ups.

结构化数据不会消失，我们对其进行结构化以降低复杂性。

Structured data is not going away and we structure it to limit complexity.

这种复杂性源于运营，而不是软件本身。

And the complexity is driven by the operations, it's not driven by the software.

所以我认为，正确看待这个问题的方式是，软件中存在这种复杂性是有原因的。

So I think the right way to view this is there's a reason that there is this complexity in the software.

这是因为我们有复杂的业务流程。

It's because we have complex business processes.

我们所处的环境也很复杂，还有复杂的监管框架，因此结构化数据将始终存在，但它能免除个人必须使用新的数据消费层的负担。

We've got complex environments that they sit in, we've got a complex regulatory framework, and so you're going to always have that structured data, but it frees the individual from having a new consumption layer to work with.

因此，在我看来，这些SaaS供应商应该随着用户不断变化的期望而演进，但这并不意味着可以完全省略与复杂运营系统集成的全部工作。

And so the right in my opinion, for these SaaS vendors to do is to evolve with the evolving expectations of the users, but that doesn't mean it obviates all of this work of having to integrate into a complex operational.

是的。

Yeah.

我只是不断地说，当人类用户总数减少而代理增多时，它们可能都需要将商业模式重构为基于令牌和操作的模式，而不是过度依赖其他方式。

I just keep saying that when you have less total human users and you have more agents, they all probably need to refactor their business models to some type of consumption based on tokens and actions and not so much based That's on

这是一个巨大的话题。

a huge topic.

我的意思是，你们两位可能都记得从永久许可模式向订阅模式的转变吗？

And so do I mean, you you you both probably remember the move from perpetual license on trend to the recurring?

我的意思是，那催生了一些公司，也淘汰了一些公司。

I mean, that that gave rise to companies, killed companies.

这是最具颠覆性的事情之一。

Was one of the most disruptive things.

并不是所有公司都还没达到这个阶段。

Not all companies are even there yet.

直到今天，仍然有一些公司在讨论这种转变。

There's still companies to date that are talking about this move.

现在我们正经历另一次定价变革，即从定期收费转向按使用量计费，这将带来与之前同等规模的巨大颠覆，而我们目前正目睹这一变化。

Now we're seeing another pricing change, which is from recurring to consumption basis, and that's going to be a whole massive disruption at the same level of that, and we're seeing that right now.

确实如此。

Absolutely.

是的，马丁，我想回到基础设施的话题上。

Yeah, so Martin, I want to flip back to infrastructure.

我认为你提到过，由人工智能和云计算驱动的基础设施倒置。

I think you talked about an infrastructure inversion driven by AI and cloud.

你是否持有某种与市场尚未充分认知的、关于企业基础设施的反直觉观点？

Is there a contrarian view that you have on enterprise infrastructure that the market hasn't fully internalized

企业尚未完全意识到。

enterprise yet.

这都还在路上。

That's all up and come.

我真的认为这一切都还在发展中。

I really think it's all in the come.

我的意思是，还有很多开放性问题。

Mean, there's very open questions.

举个例子，一个开放性问题是：如果由智能代理来做决策，中央采购方、平台团队和IT团队会怎样？

For example I'll tell you, one open question is, what's going to happen to central buyers and platform teams and IT teams if agents are making the decisions?

假设我现在是个开发者。

So let's say I'm a developer right now.

在AI出现之前，如果我是个开发者，想用一个数据库，IT团队会提供一套基础设施和文档。

So let's say pre AI, if I'm a developer and I want to use a database, like, the IT team provides a set of infrastructure, a set of docs.

我会接受培训，了解这些资源，并做出符合组织政策的技术决策。

I get onboarded, and I know about them, and I make these technical decisions that are in line with the policies of the organization.

对吧？

Right?

现在我每天晚上都写代码，也每天晚上用AI写代码。

Today so I code every night, and I AI code every night.

这简直是最有趣的事了。

It's the most fun thing ever.

顺便说一下，你正在编码这一事实表明这个目标市场正在扩大。

And by the way, the fact that you're coding just shows that this TAM is expanding.

我们现在都在编程，而以前我们并不编程。

We're all we're all coding now and we didn't code before.

那么，是谁在做出技术决策呢？

So who is making a technical decision?

比如，如果你在使用 Cursor 或 Cloud Code，是什么决定了你要使用的基础设施？

Like, if you're using cursor or if you're using Cloud Code, what's making the tech decision of infrastructure to use?

是人工智能在做出这个决策。

The AI is making that decision.

基础设施是一个价值数万亿美元的产业，而你基本上已经将人类从决定使用什么的环节中移除了。

And so infrastructure is a multi trillion dollar business and you've removed the human by and large from actually making the decision of what to use.

我们完全不知道这在内部意味着什么。

We have no idea what that means internally.

我们也不知道这对整个行业意味着什么。

We have no idea what that means to the industry.

因此，我认为人工智能带来的真正颠覆性变革仍在路上，我们目前只是通过个人用户逐渐普及看到了一些非常初步的迹象。

And so I think a lot of the real disruptions from AI are still on the come, and we're just seeing very, very early glimpses of that through secular adoption by individual users.

所以我们只剩下几分钟了，帕特，非常感谢你的分享。

So we've got only minute or two, and I really appreciate, Pat, we really appreciate.

你是搞基础设施的，帕特刚才把话题拉回了这里，但我一直听到大家都在说一切都很受限。

What about the, you're the infra guy, and Pat kind of shifted us back here a little bit, but I mean, I just keep hearing everything is constrained.

你知道吗，2026年将会是内存之年。

Know, is, '26 is going be the year of memory.

我们正在为那些在大多数情况下技术上还无法实现的科技签署核能合作协议，因为我们需要为这些机架供电。

You know, we're inking nuclear deals for tech that doesn't even technically work yet in most cases, because we're going to need to energize these racks.

我们现在收到的机架重量已经达到两吨了。

We've got two ton racks, you know, now being delivered.

至少有一台达到了2.6吨，对吧？

At least one up did with a 2.6 ton right?

我的意思是，詹森简直在突破物理定律，但你对计算资源的效率以及这些不同限制的判断是什么？

I mean Jensen's breaking physics laws, but like, is your sort of read on compute capital efficiency, all these different constraints?

那么，这将如何重塑我们未来扩展整个系统的模式呢？

And like, how is this going to reshape the, you know, kind of the way we're going to scale this whole thing going forward.

只有一个限制，那就是监管。

There's only one constraint, and that's regulatory.

你知道什么很有趣吗？

You know what's very interesting?

你们有没有听说过太空中的数据中心？

So do guys hear about data centers in space?

当然有。

Oh, Absolutely.

这简直是个荒谬的概念。

It's a it's a ridiculous concept.

你知道的。

You know?

你得一路跑到太空去。

Like, you have to go all the way to space.

这太蠢了。

It's so stupid.

所以事实证明，它并不蠢，只有一个原因。

So so it turns out it's not stupid for exactly one reason.

你能猜到为什么吗？

Can you guess why that is?

是的。

Yeah.

监管。

Regulation.

更少的监管。

Less regulation.

所以

So

这是100%的。

it's a 100%.

顺便说一下，这些数字之所以能对上，完全是因为监管。

By the way, the numbers pencil out just because of regulation.

在美国破土动工是如此繁琐。

It is so onerous to break ground in The United States.

与其这样，还不如把数据中心送到太空去。

It makes more sense to send a data center to space.

所以听好了，我们人类是非常非常有创新精神的。

And so listen, we are a very, very innovative species.

我们是一个非常成熟的行业。

We're a very mature industry.

如果我们需要增加带宽或芯片产能，我们知道该怎么去做。

If we need capacity on bandwidth, on chips, we know how to do it.

问题在于清除官僚障碍来实现它。

The issue is getting the bureaucracy out of the way to do it.

这说得通。

That makes sense.

不过刚才说得太快了。

That was too fast though.

这根本没给我留下几个有力的采访素材，比如我们根本建不够这一点。

It doesn't even give me a few good sound bites about like, you know, the fact that we can't build enough.

但归根结底，你的观点是我们需要建造更多晶圆厂。

Because your point though, in the end is like, we need to build more fabs.

这需要时间。

That takes time.

你需要建造更多数据中心。

You need to build more data center.

这需要时间。

That takes time.

需要

Need to

对。

Right.