为什么Stripe可能收购PayPal，智能购物课程调整，ChatGPT的语音语言障碍

欢迎各位来到Informations TITB。

Welcome everyone to the Informations TITB.

我叫阿卡什·马斯里查。

My name is Akash Masricha.

今天是2月25日，星期三。

It is Wednesday, February 25.

今天节目中，我们将首先讨论PayPal潜在的收购方可能是谁。

Today on the show, we are kicking off with a discussion around who potential acquirers could be for PayPal.

我们马上邀请Better Tomorrow Ventures的谢尔·米诺特加入。

We're bringing on Shiel Minot from Better Tomorrow Ventures in just a minute.

随后，我们的电子商务记者将上线，谈谈AI公司在电商领域的优先事项在过去几个月是如何变化的。

We then have our e commerce reporter coming on to talk about the priorities of AI companies around commerce and how they have changed over the past few months.

她今天早上发布了一篇新报道，重点介绍了该领域值得关注的主导力量。

She has a new story out this morning highlighting who the power players are to watch in that space.

接着我们将讨论以音频为重点的AI模型，以及一些大型AI实验室在将这项技术扩展到不同语言时面临的挑战。

We're then talking about audio focused AI models and the challenges that some big AI labs are having scaling that technology to different languages.

最后，我们将以关于当前人工智能广告现状的讨论结束本期节目，并邀请一位专注于该领域的创始人和投资者参与讨论。

And finally, we will end the show with a discussion on the current state of AI advertising and bring on a founder and an investor who have been focused on that space.

这将是一期有趣的节目，所以我们马上开始。

It's going to be a fun show, so let's get right on into it.

过去几年，贝宝的股价表现不佳。

PayPal shares have had a difficult run over the past few years.

其收入增长也显著放缓。

Revenue growth has also decelerated dramatically.

五年前，该公司的营收增长率约为20%。

Five years ago, the company was growing its top line around 20%.

在最近一个财年，收入增长仅为4%出头。

In the latest fiscal year, revenue grew at a little more than 4%.

本周《彭博社》还报道，Stripe正在考虑收购贝宝，或其部分业务。

Bloomberg has also reported this week that Stripe is looking into buying PayPal or potentially just parts of it.

我想邀请一位对此有深入见解的人士参与讨论。

I want to bring on someone who I know has thoughts on all of this.

谢尔·米诺是Better Tomorrow Ventures的联合创始人兼普通合伙人。

Shiel Minot is co founder and GP at Better Tomorrow Ventures.

谢尔，欢迎再次做客节目。

Shiel, welcome back to the show.

很高兴你来到这里。

It's great to have you here.

很高兴能来。

Glad to be here.

谢谢。

Thanks.

你昨天在X上发布了一篇帖子，提到一些我认为可能收购PayPal的公司。

So you published this post yesterday on X talking about, Hey, here's a couple companies that I think could acquire PayPal.

结果你发完帖子还不到一小时，彭博社就发布了报道，称Stripe正在认真考虑收购PayPal。

And lo and behold, I think it was like less than an hour after you made that post, Bloomberg comes out with this report that Stripe is looking somewhat seriously at PayPal.

首先，我想知道你当时的反应是什么。

And A, I'm wondering what your reaction was.

B，我们来谈谈你对这件事的看法。

And B, let's talk about how you're thinking about this.

是的，当然。

Yeah, absolutely.

这很有趣。

So it's funny.

对。

Yeah.

我在帖子中提到的第一家公司是Stripe，但我其实说，Stripe拥有许多对它而言极具吸引力的资产，因为Stripe没有真正的消费者品牌，而PayPal有。

What I wrote in the post, the first company I called out was Stripe, but I actually said they have a lot of desirable assets for Stripe because Stripe has no real consumer brand and PayPal has a consumer brand.

PayPal是一个现成的支付按钮。

PayPal is there as a button.

人们知道Venmo。

People know Venmo.

最重要的是，PayPal拥有数亿客户的银行账户信息。

And the most important thing is PayPal has bank account details for hundreds of millions of customers.

所以当你使用PayPal支付时，通常不是用一张会让他们支付百分之几手续费的信用卡，而是用几乎不产生成本的银行账户支付。

So instead of when you pay with PayPal, you're often paying, instead of paying with a credit card that costs them a couple percent, you're often paying with a bank account that costs them nearly nothing.

这对Stripe来说是一个巨大的优势。

And that's a huge advantage that Stripe could get.

但问题是，PayPal背负着大量的文化和技术债务。

Now, the problem is PayPal has a lot of both cultural and technical debt.

他们大约有25,000名员工。

They have about 25,000 employees.

他们的文化与Stripe完全不同。

They have, you know, just a totally different culture than Stripe.

这会是一个非常奇怪的组合。

It would be a very strange combination.

所以我说这对Stripe来说可能根本行不通，但 apparently Stripe 正在考虑它。

So that's why I said it's probably a nonstarter from Stripe, but apparently Stripe's looking at it.

你所说的文化差异指的是什么？

What do you mean by culture differences?

我的意思是，除了规模和人数之外，你指的是公司的运作方式、组织架构这些方面吗？

I mean, besides the size and the number of people that are there, I mean, is is this, you know, people you know, the company, the ways that it's working, the way the company is built out?

Doc，再多谈谈

Doc, So talk more about

Stripe 是一家以开发者为先的公司，而且在增长过程中一直非常高效。

Stripe is very much a developer first company, and they have been pretty efficient in their growth.

PayPal 存在已久，曾经长期受制于 eBay，脱离 eBay 后又经历了多次领导层更迭。

PayPal has been around a long time, you know, under the shackles of eBay, then out of eBay, a lot of different leadership changes.

它就是一个相当臃肿的组织，以决策缓慢和缺乏效率著称。

And it's just a fairly bloated organization and not known for making decisions quickly or being super efficient.

所以我在想，为什么你觉得 PayPal 现在更适合并入一家大公司，而不是独立运营？

So I'm wondering why you think PayPal would sort of fit better inside a larger company right now as opposed to standalone.

几周前，我们邀请了大卫·马库斯做客节目。

Well, we had David Marcus on the show a couple weeks ago.

当时 PayPal 宣布了首席执行官的更替，他提到 PayPal 在其发展历程中一直侧重于交易量而非产品。

This is when PayPal announced the CEO switch up, and he sort of mentioned that PayPal had emphasized payment volume instead of product throughout the course of its history.

我的意思是，我只是好奇，为什么你觉得这家公司放在更大的组织里会更好，无论是像Stripe这样的公司，还是你提到的其他选项，比如银行或信用卡公司？

And I mean, I just wonder why you think that the company would be better inside a larger org, whether it's Stripe or you called out some other options too, like banks or credit card companies?

是的。

Yeah.

好吧。

So, okay.

所以我认为，PayPal仍然是一个非常盈利的公司。

So I think, look, PayPal's still a very profitable company.

你知道，利润率下降了，但它仍然是盈利的。

You know, margins are down, but it's still a profitable company.

他们并不需要出售。

They don't need to sell.

我认为，现在股价已经大幅下跌，过去四年左右跌了85%，这为某人提供了一个机会。

I do think that now with the stock being beaten down as much as it has been, down 85% over the last four years or so, I think it's an opportunity for somebody.

所以，真正的问题是，现在是否是别人出手的最佳时机。

So it's really a question of is now the time for someone else to strike.

是的，我提到过我们讨论过Stripe。

And, yeah, I mentioned we talked about Stripe.

我认为这里还有其他几个潜在候选者。

I think there are several other candidates here.

我提到的是苹果公司，我认为这不太可能，但可以很好地补充Apple Pay。

The ones I highlighted were Apple, again, I think a non starter, but it could be a good compliment to Apple Pay.

Apple Pay在线下支付场景中做得非常好，但在线上表现得不够理想。

Apple Pay has done a really good job of getting us to use it at a checkout offline, but they have not done as good a job online.

所以我认为这是一个很好的机会。

So I think that's a good opportunity.

他们从未真正把Apple Pay的点对点支付做起来。

They never really got Apple payments peer to peer going that well.

Venmo在这里可能是个不错的搭配。

Venmo could be a good fit there.

而PayPal实际上还有一个规模不小的BNPL业务部门。

And then PayPal actually has a sizable BNPL company, so BNPL division.

所以这对苹果来说也可能很有用。

So that also could be useful within Apple.

苹果传统上并没有做过这种收购。

Think Apple has traditionally not done this kind of acquisition.

并购。

M and

对，没错。

A, right.

那信用卡公司呢？

What what about the what about the credit card companies?

是的。

Yeah.

所以维萨和万事达完全有能力承担。

So so Visa and Mastercard could both totally afford it.

有趣的是，当贝宝刚起步时，人们认为它长远来看可能会取代维萨和万事达，因为使用贝宝支付时，你根本不需要用到维萨和万事达。

It's interesting because, you know, when PayPal started, it was seen as just as potentially getting rid of Visa and Mastercard over the long run because that with that PayPal checkout, you didn't need to use Visa and Mastercard.

然后，你知道，PayPal 在发展过程中遇到了不少问题。

And then, you know, PayPal had a bunch of problems along the way.

但那个结账按钮是他们网络中极其宝贵的资源。

But that checkout button is super valuable real estate for their networks.

他们一直在努力超越单纯的交易手续费，也就是你使用卡片时他们获得的那部分收入。

And they've been trying to move beyond just interchange, which is like when you use the card, what they get.

他们还试图直接与商家建立关系。

And they've tried to build relationships directly into merchants.

因此，PayPal 在这方面可能非常有价值。

So PayPal could be super valuable there.

在这方面，我会担心反垄断问题。

There, I would worry about antitrust.

而且我觉得这个担忧是合理的，说实话。

And I I think it's valid, honestly.

我们这个圈子里的人常常讨厌反垄断，但事实上，我认为两大网络之一收购最大的独立在线结账服务商，恐怕会面临巨大挑战。

People in our world often hate antitrust, but actually, I think one of the two largest networks acquiring the largest independent online checking checkout provider would be, I think, a challenge.

让我提出另外两个选项。

Let me let me throw out two other options here.

有人在你的帖子下回复说亚马逊，实际上有人为此写了一整篇Substack文章，而且我和我们新闻团队的人聊过，他们也有这个想法。

Amazon is someone that somebody responded to your post suggesting Amazon, actually wrote an entire substack on this, and actually, I was talking to folks in our newsroom, and they had this idea as well.

谷歌有Google Pay。

I mean, Google has Google Pay.

苹果有Apple Pay。

Apple has Apple Pay.

亚马逊并没有自己的原生支付平台。

Amazon doesn't really have a native payments platform of its own.

你怎么看？

What do you think?

是的。

Yeah.

我认为亚马逊是最有趣的。

I think Amazon is the most interesting one.

这是我刚提到的那个，但我认为其实挺有意思的。

It's the one I sort of just left off, but I think it's actually pretty interesting.

而其中最有趣的一点是，他们已经保存了用户的银行账户信息。

And the main thing interesting there is the bank account information on file.

这能立即为亚马逊节省大量成本。

Like, that could immediately save Amazon itself a lot of money.

还有BNPL业务。

And and the BNPL book.

我认为亚马逊面临监管风险，但如果不是因为这个，它可能会是一个绝佳的组合。

I think with Amazon, the regulatory risk is tough, but I think it could be if not for that, I think it could be a great combo.

好的。

Okay.

最后，我就说到这里。

And last, I'll just stop.

我不再继续提其他选项了，但想想这些还挺有趣的。

I'll stop throwing out options here, but it's kinda fun to think about this.

我的意思是，你昨天就是这么做的，所以这里的游戏规则就是这样。

I mean, that's what you did yesterday, so that's the game here.

那关于一个区块呢？

What about what about a block?

或者对。

Or yeah.

区块。

Block.

对。

Yeah.

本来想说方形。

Was gonna say square.

区块。

Block.

是的。

Yeah.

方块和正方形。

Block square.

我仍然也叫它正方形。

I still call it square too.

我认为，你看，Square 和 PayPal 有非常多的自然协同效应。

I I think, look, Square and PayPal have so many natural synergies.

你知道，Square 本质上是一个线下的支付提供商。

You know, Square is effectively an offline provider of payments.

PayPal 是一个线上的支付提供商。

PayPal is an online provider of payments.

有很多理由说明它们合并是有意义的。

There's a lot of reasons why you'd want them together.

PayPal 多年来一直试图进入线下业务。

PayPal has been trying for many years to get into the offline business.

这已经十多年了。

It's been over a decade.

他们曾非常努力地与Square竞争。

They tried really hard to compete with Square.

他们甚至还推出了自己的读卡器。

They actually even had their own readers.

他们的想法是，如果你能在线下使用PayPal付款，就能加强PayPal的整体关系。

And the idea was if you could just pay with your PayPal offline, that would strengthen the whole relationship that PayPal had.

但他们始终没能将这个想法落地。

They could never get it off the ground.

现在，关于Stripe的问题，我之所以没把它算进去，是因为Stripe本身的股价大跌了。

Now, the problem with Stripe, the reason I didn't include it, is Stripe itself is trading way down.

或者抱歉。

Or sorry.

Block。

Block.

Block。

Block.

是的。

Yes.

Block 自身的股价

Block itself is

大幅下跌。

trading way down.

而且我认为，Block 的市值大约是 300 亿美元。

And Block, I think, is trading at, like, a $30,000,000,000 market cap.

所以我认为，他们要收购一家可能估值在 500 亿美元左右的公司，确实很困难。

So I I think, you know, for them to acquire a company that's probably gonna be acquired in the $50,000,000,000 range, I think, is tough.

但这两家公司的合并在战略上自然说得通。

But the combination naturally makes sense.

说到这个，我们再回过头来看看 Stripe——我们至少有一些报道称，它正在关注 PayPal。

And so on that note, let's go back to Stripe, the company that we at least have some reporting out there that is looking at PayPal.

我的意思是，Stripe 本身是一家收入规模较小的公司，那么你认为他们该如何为这样的资产支付呢？

I mean, Stripe itself, a smaller company in terms of revenue And right so how do you think they would pay for an asset like this?

如果收购的不是整个公司，你认为他们会特别感兴趣于哪些资产呢？

And then also, if it's not the whole company, which assets specifically do you think they would be interested in?

是的。

Yeah.

这是个很好的问题。

So it's a great question.

Stripe，你知道吗，昨天还有新闻说他们的估值达到了1600亿美元，所以他们并没有现成的500亿美元资金来完成这笔收购。

Stripe, you know, also yesterday there was news out that they were valued at $160,000,000,000 So, you know, they don't have $50,000,000,000 just ready to do here, ready to acquire here.

我认为最终可能发生的情况是，一群私募股权基金联合进来，与Stripe同时收购他们不打算全盘接收的资产。

I I think what ends up happening is you could have a consortium of private equity funds coming in and acquiring some of the assets concurrently with Stripe buying out the assets that they want.

我认为他们真正想要的可能是那个消费者品牌。

I think what they want is probably that consumer brand.

无论是在线结账功能还是银行账户信息，都会非常有价值。

So both at online checkout and the bank account information would be super valuable.

我认为Venmo对Stripe来说也会非常有价值。

And I think Venmo also would be super valuable to Stripe.

所以我认为他们可以获取这些资产，同时将一些其他辅助业务转让给私募股权，并在此过程中为公司注入现金。

So I think they could get those things and maybe pass off some of the other ancillary businesses to private equity at the same time, at the same time taking a cash infusion into this business.

我认为，这对Stripe来说将极大地提升价值，而且这一点很容易量化。

I think, you know, I think it would be super value accretive to Stripe, and you could easily quantify that.

而且，据我所知，科利森兄弟似乎有能力吸引无限的资本，所以我认为他们能做到。

And, you know, the Collisons seemingly have the ability to attract infinite capital, so I think they could do it.

但正如你所说，这并不容易。

But, you know, to your point, it wouldn't be easy.

事实上，今天早些时候我刚和一个人发短信讨论过，我们谈到了Stripe那惊人的估值与PayPal当前被严重低估的估值之间的对比。

Well, and I was texting with someone earlier today, we were talking about the relative valuations of the astronomic valuation Stripe is at, and then the very depressed valuation that PayPal is at.

因此，如果你认为Stripe的估值过高，或者只是正处于一个高点，那么PayPal拥有的充沛现金流恰恰可以支持另一种观点：也许Stripe的估值并没有被高估。

And so to the extent that you think that Stripe is overvalued or just really riding a high right now, I mean, you have this free cash flow that PayPal has that would sort of help make the argument that, well, maybe Stripe is, it's not overvalued at some point.

值得思考一下。

Something to think about.

是的，毫无疑问。

Yeah, for sure.

我认为，Stripe的市盈率高得离谱，而PayPal的市盈率则低得离谱。

I think, look, Stripe trades at a crazy high multiple and PayPal trades at a crazy low multiple.

所以这正是经典的案例：你能否收购这家企业，并将其估值倍数从一个水平提升到另一个水平？

So it's the classic, you know, can you acquire this business and change the multiple from one to the other?

我觉得这会非常困难，这也是我最初排除这种可能性的原因。

I think it's going to be really hard, which is why I initially counted them out.

但你知道，如果有人能做到的话，那一定是呼叫中心的人。

But you know, if anyone can do it, the call centers can.

太好了。

Great.

好了，Shiel，非常感谢你来做客。

Well, Shiel, I wanna thank you for coming on.

这位是Shiel Minot，Better Tomorrow Ventures的联合创始人兼普通合伙人，正在TI TV上。

That is Shiel Minot, cofounder and GP at Better Tomorrow Ventures here on TI TV.

过去几个月里，关于人工智能如何与购物世界互动的讨论发生了巨大变化。

The conversation around AI will inter on how AI will interact with the shopping world has changed drastically over the past few months.

我的同事、我们的电子商务记者安妮·吉安撰写了一篇关于这一趋势的文章，并更新了AI购物领域高管人才追踪名单，这些高管绝对值得关注。

My colleague, Anne Gian, our ecommerce reporter, wrote a story on this trend and also updated the information's talent tracker for the AI shopping executives that should absolutely be on your radar.

我想请安妮来为我们详细介绍一下她的报道。

I want to bring on Anne to tell us more about her reporting.

安妮，欢迎再次做客节目。

Anne, welcome back to the show.

很高兴你来到这里。

It's great to have you here.

嗨，阿卡什。

Hi, Akash.

几个月前，你在AI购物代理商业领域发布了一份值得关注的顶级高管名单，现在你重新审视了这份名单。

So you published this list of the top executives that should absolutely be on our radar a couple months ago in the AI shopping agentic commerce world, and you've revisited that list.

我稍后想聊聊名单上有哪些人。

And I want to talk about who's on the list in a second here.

但总的来说，你能跟我讲讲过去六个月里，关于AI购物的讨论发生了哪些变化吗？

But broadly speaking, I mean, me a little bit about how the conversation around AI shopping has changed over the past six months.

当然。

Sure.

我的意思是，我们最初在五月发布了这个名单的第一版，那时许多AI公司和大型科技公司刚开始思考智能代理电商，并在他们的AI产品中加入更多电子商务功能。

I mean, I think so we originally published the first version of the list in May, really kind of just when a lot of both AI companies and bigger tech companies were starting to think about agentic commerce and putting more e commerce features into their AI products.

当时，最初的愿景非常以代理为中心，即你的AI代理会接管你的浏览器，自动点击、浏览网站，为你搜索商品，无论是亚马逊还是诺德斯特龙这样的百货商店，你只需告诉代理你想要什么，它就会像人类一样浏览网页。

And at that time, the original kind of vision was very agent centered and kind of the idea that your AI agent would, you know, take over your browser and go click around and search on a website for you whether it's Amazon or a department store like Nordstrom, you would tell your agent what you're looking for and then it would go browse the web, you know, in a very similar way that you, the human would.

但过去几个月我们看到，公司推出的功能并不符合这一愿景。

And really what we've seen over the past couple of months is the features that companies are rolling out are don't match that vision.

他们推出的实际上是，包括OpenAI、谷歌以及Perplexity等公司，都在聊天窗口中直接加入了结账按钮。

What they've been launching, you know, both OpenAI and Google as well as other companies like Perplexity, they've all been bringing the checkout button directly into their chat windows.

所以，这更像是公司将购物功能叠加到聊天体验中，而不是试图开发能像人类一样浏览网页的代理。

So it's really more that companies are kind of layering that into the chat experience versus trying to come up with these agents that can browse in a very similar way to humans.

这就是主要的转变。

So that's kind of the main shift.

因此，我们更新了这份名单，以反映这一变化。

And that's why we've updated the list to kind of reflect that.

你认为这种转变为什么会发生？

Why do you think that shift has been happening?

是因为实现智能体电商在技术上真的很难，达不到人们预期的效果吗？

Is it just technically really hard to get agentic commerce to take off the way that people thought?

是因为钱的问题吗？

Is this a money thing?

那根本原因到底是什么？

Or what's the root cause here?

是的。

Yeah.

我的意思是，我也写过这方面的东西，你知道，网页仍然是为人类使用而设计的。

I mean, I think, and I've written about this as well, you know, the web is still designed for humans to use.

尤其是对于电商网站来说，当你访问一个零售商网站时，可能会弹出一个窗口，要求你提供邮箱以换取折扣码。

So especially you know for an e commerce site when you go to a retailer site there might be a pop up asking for your email in exchange for a discount code.

很多网站都有反机器人保护和反欺诈技术，旨在识别浏览网站的用户是否为真人。

A lot of websites you know have bot protection and fraud technology that's designed to you know identify when someone browsing the site isn't human.

所以，从用户体验的角度来看，网站的结构方式基本上仍然是为人类设计的。

So just kind of the way that websites are structured, know, at least from kind of the user experience perspective, it's really still kind of built for humans.

因此，许多公司在这方面一直难以取得突破。

And so a lot of companies have really struggled to build something around that.

而点击结账按钮这种方式对他们来说更容易、更快地实现。

And kind of the checkout button route has been a lot easier and faster for them to get going.

尽管这当然也带来了自己的挑战。

Although it certainly, you know, has its own challenges.

那么，这如何反映在我们网站上你对人才追踪器所做的改动呢？

So how does this then reflect the change in the talent tracker that you made on our website?

在这个故事中，谁如今拥有更多权力，谁又拥有更少权力？

Who has more power today in this story and who has less power?

是的。

Yeah.

我的意思是，我们更新并反映在列表中的变化是，现在越来越多的工作落到了工程团队身上。

I mean, we so what we've updated and kind of reflected in the list is that now a lot more work is kind of falling to people on the engineering side.

像OpenAI、谷歌这样的大公司一直在开发这些协议或规则，以构想这些聊天机器人交互如何整合。

A lot of big companies like OpenAI, Google have been developing these protocols or sets of rules for how they envision, you know, these chatbot transactions coming together.

因此，显然看到越来越多的人在清单上从事细致的工程工作，以及商业合作。

And so certainly seeing a lot more people kind of doing the in the weeds engineering work on the list, as well as commercial partnerships.

在过去几个月里，这些合作变得越来越重要。

Those have become more important over the past couple of months.

所以你开始看到更多人出现在销售和合作这一侧。

So you're starting to see more people on kind of the sales and partnership side.

你也会看到许多公司招聘担任这些角色的人。

You're seeing a lot of companies get people in those kinds of roles as well.

给我们讲讲最让你着迷的几个人的名字。

Give us a few of the names of the people that fascinate you most.

我想知道，如果你和这些人都聊过，你现在最想问他们什么？

I'm wondering if you had a conversation with any of these people, what would you want to ask them right now?

是的。

Yeah.

我的意思是，有两个人我一直在密切关注，他们都在名单上。

I mean, two people that I am watching pretty closely who are on the list.

一个是Shopify的产品高管瓦妮莎·李。

One is Vanessa Lee, who is a product executive at Shopify.

我们上次就把她列在名单上了，而她之后在Shopify承担了更多责任。

And we actually had her on the list last time and she has since taken on more responsibility at Shopify.

因此，跟踪她的晋升轨迹非常有意思。

So it's been interesting to kind of track her rise.

我认为这其中一个重要原因是，AI电商对Shopify来说正变得越来越重要。

And I think a big part of that is just that, you know, AI commerce is becoming really important to Shopify.

所以，看到他们显然一直在与AI公司合作开发所有这些结账功能，同时也在为商家开发大量功能，让他们能轻松将自己的产品接入这些全新的结账功能，这很有意思。

And so it's been interesting to see they obviously have been working with the AI companies on all of these checkout features, but also at the same time developing a lot of features for their merchants to make it really easy for them to get their products in all of these new checkout features.

因此，我不仅密切关注她，也密切关注OpenAI新加入的员工阿萨德·阿万。

So I'm definitely watching her very closely as well as a relatively new hire, Assad Awan at OpenAI.

他是在过去几个月才加入的，此前在Meta工作了十多年。

He just joined the past couple of months, after more than a decade at meta.

他负责广告业务，而这一块目前还没有明确与他们的电商业务挂钩。

And he's working on the ad side, which hasn't been explicitly linked to their ecommerce efforts just yet.

但我认为，整个电商和广告领域的人们都期待着这一整合最终会发生。

But I think everyone in kind of the commerce and ads world is expecting that to happen eventually.

就在过去几周，随着OpenAI推出广告测试，我们看到一些零售合作伙伴，比如Shopify和Target，已经开始向OpenAI购买广告位，然后将这些广告链接到自己的广告业务中。

And something interesting just in the past couple of weeks as open AI has been rolling out their ads test, we've seen some of their retail partners, like Shopify and Target, they're actually buying ad space from open AI and then turning around and linking that to their own ad businesses.

因此，他们展示的是使用Shopify搭建网站的品牌广告，并引导用户前往这些品牌的App或网站进行购买。

So they're showing ads for you know, brands that use Shopify to host their site and driving them to, you know, their own apps and sites to make purchases.

这很有趣，因为尽管OpenAI方面还没有正式建立这种关联，但零售商们已经迅速朝这个方向行动了。

So it's interesting, because even though there's not like a formal link there yet, from open AI side, you already see retailers very quickly moving in that direction.

我必须说，让我印象最深的是，当我们七到八个月前，去年夏天刚启动这个节目时，每个人都热议‘智能体电商’，那是当时大家谈论的焦点。

I have to say the thing that really stands out to me here is when we started this show about seven to eight months ago back in the summer, everyone was raving about agentic commerce, and that was the thing that everyone talked about.

那时AI所处的阶段略有不同。

It was a slightly different moment for AI.

我认为，当时对展示成果、尽快实现自由现金流的压力，没有现在这么大。

I think the pressure on showing results and getting closer to someday generating free cash flow, there was less pressure in that moment than there is now.

所以现在我非常觉得，你所描述的这种转变正在回归现实，也就是说：好吧，我们并不会征服整个世界，杰西。

So now I very much feel like this shift that you're describing is then coming back to reality a bit and saying, Okay, well, we're not going to conquer the whole world, Jesse.

我们也不会把整个世界翻个底朝天。

We're not going to flip the whole world on its head.

我们只是会以我们熟悉的方式去做事情。

We're actually just going to sort of do things the way that we know how to do them.

说到底，我甚至都不希望有个智能代理替我购物。

I don't even know that I want an agent to do my shopping for me at the end of the day.

没错，这是个非常好的观点，我认为现在很多公司都在思考这个问题：在哪些情况下使用智能代理或ChatGPT代购是合理的，而在哪些情况下人们更愿意自己动手？

No, mean, that's a really great point, and I think that's a question that a lot of these companies are working through right now is where does it make sense to have an agent or chat GPT shop for you and where do people want to do it themselves?

很好。

Great.

安妮，非常感谢你来参加节目。

Well, Anne, I want to thank you for coming on.

这位是安妮·吉安，我们《信息》杂志的电子商务记者。

That is Anne Gian, our e commerce reporter here at The Information.

AI音频模型正日益成为AI实验室的关注重点，但一个阻碍这些AI音频模型成功的关键障碍是，即使是最大的公司也难以将这项技术扩展到多种语言。

AI audio models are increasingly becoming a focus for AI labs, but one logistical hurdle that is challenging the success of these AI audio models is the fact that even the biggest companies are having trouble scaling this technology to different languages.

这是我的同事斯蒂芬妮·帕拉佐洛本周发表的一篇文章中提到的，我想请她来详细谈谈这个问题。

That is according to a column that my colleague Stephanie Palazzolo published this week, and I want to bring her on to talk all about it.

斯蒂芬妮，欢迎再次做客我们的节目。

Stephanie, welcome back to the show.

很高兴你能来。

It's great to have you here.

谢谢。

Thanks.

很高兴能来这里。

Great to be here.

你是怎么发现AI音频模型在不同语言中的有效性存在问题的呢？

How did you come across this issue of how effective AI audio models not just are, but are in different languages?

确实如此。

Definitely.

最近几个月，我写了大量关于音频AI模型兴起的文章，指出AI正朝着能够像人类之间那样倾听和对话的方向发展。

So in recent months, I've been writing a lot about the rise of audio AI models and how there's really been the shift towards AI that's able to listen and talk to humans the way that we do with each other.

尤其是上周关于OpenAI设备努力的一篇报道，这些设备将采用音频优先的设计，意味着人们将与设备对话，设备也会回应他们。

And especially with a recent story from last week about OpenAI's device efforts, which are going to be audio first, meaning that people will be talking to the device and it will be talking back to them.

你可能会认为，这些公司希望这些设备能在全球范围内被使用，这意味着它们必须理解和使用各种不同的方言和语言。

You would imagine that these companies want these sorts of devices to be used by people around the world, meaning that they have to understand and speak all sorts of different dialects and languages.

随着我与研究人员的交流越多，他们越告诉我，文本模型在西方语言和非西方语言之间已经存在差距，而音频模型的差距甚至更大。

And then, you know, the more I talked with researchers, the more that they told me, you know, there's already this gap between Western and kind of non Western languages with text based models, but that gap is even bigger with audio models.

那么，为什么音频模型在非英语语言中的表现不够好呢？核心原因是什么？

And what are the core reasons here as to why the audio models are not as good in non English languages?

是的，归根结底，就是因为可用的训练数据太少，尤其是音频数据。

Yeah, I mean, it really just boils down to the fact that there just isn't that much training data there, especially audio data.

为了以最佳方式训练这些模型，你需要大量不同年龄、性别的人们谈论各种主题的音频数据。

So in order to train these models, you know, in the best way possible, you basically need data of people of different ages, genders, speaking about a variety of topics.

从客户服务到医疗领域，方方面面都需要。

So everything from, you know, customer support to medicine.

而且你需要所有这些语言的数据。

And you need that in all sorts of languages.

而这类数据并不会自然产生。

And so that type of data doesn't really occur naturally.

因此，公司必须费尽心思去收集这类数据，这显然对他们来说非常困难。

And so companies would kind of have to go out of their way to collect that sort of data, which is obviously very difficult for them to do.

你在专栏中聚焦了OpenAI。

You focused on OpenAI in your column.

我确信还有其他公司也在做完全相同的事情，并面临同样的问题。

I'm sure there are other companies that are working on the exact same thing and have the same issues.

但我们就先以OpenAI为例，谈谈他们在北美以外的扩张计划，这些国际市场对他们有多重要。

But I mean, if we just stick with OpenAI for the moment, tell us about their expansion plans outside of North America, how important those international markets are for them.

是的。

Yeah.

正如你所说，专栏主要聚焦于OpenAI，但我们可以想象，这对任何希望实现全球覆盖的AI公司来说都将是一个重大问题。

So as you mentioned, the column kind of focused on OpenAI, but we can imagine that this is going to be a big deal for any sort of AI company that wants to have a global reach.

像谷歌和Meta这样的公司。

So companies like Google and Meta for instance.

但就OpenAI而言，你知道，上个月萨姆·阿尔特曼表示，仅在印度，就有1亿人每周使用ChatGPT。

But with OpenAI specifically, you know, we saw Sam Altman say in the last month that the company has 100,000,000 weekly active users of ChatGPT in India alone.

这意味着全球每周使用ChatGPT的人中，有十分之一以上位于印度。

And so that's more than a tenth of all the people that are using ChatGPT on a weekly basis around the world that's located in India.

而在印度，许多交易是通过电话进行的，或者人们大声交谈完成的，这意味着为了适应当地人自然的工作方式，OpenAI必须让其模型和ChatGPT满足这一需求，并教会ChatGPT如何大声说话。

And many transactions there are being handled over the phone or by people speaking out loud to each other, which means that in order to kind of meet that kind of natural way that people there do their work, OpenAI is going to have to meet that with their models and ChatGPT, and it's going to have to teach ChatGPT how to speak out loud.

因此，拓展这些地区对OpenAI至关重要，因为正如我的同事谢丽和我在上周的一篇报道中所指出的，自GPT-5发布以来，OpenAI的ChatGPT使用量已出现放缓。

And so expanding to these regions are very important for OpenAI because as my colleague Shree and I reported in a story last week, you know, in recent months, basically since the release of GBT5, OpenAI has seen a slowdown in chat GBT usage.

因此，他们在美国内部能做的事情是有限的。

And so, you know, there's only so much that they can do in The U.

美国。

S.

因此，他们必须大力拓展全球其他地区，才能维持ChatGPT的增长势头。

So they're really going to need to expand to other areas around the world in order to keep that CHATGBT growth going.

所以，我理解你的意思是，这不仅仅关乎增长这一层面。

So, basically what I hear you saying is that it's not just important for the growth element of the story.

我的意思是，能够适应不同的行为模式和人们使用AI的方式也很重要，因为并不是每个人都在任何地方发短信。

I mean, it's also important to be able to accommodate different behaviors and the way different people use AI because not everyone is texting anywhere.

听起来人们在亚洲和其他大洲更多地使用语音。

Sounds like people are using voice a lot more in Asia and other continents.

是的。

Yeah.

我的意思是，确实存在一个关键点，就是要在用户所在的地方满足他们，匹配全球各地不同的文化偏好，以及人们希望如何工作和交流——无论是通过文字还是语音。

I mean, there definitely is this element of meeting the users where they are and matching just these different cultural preferences around the world and the ways that people want to work and communicate with each other, whether that's through you know, text or speaking out loud.

你现在听起来真像个产品经理，斯蒂芬妮。

Now you're starting to sound like a product person, Stephanie.

你真的深入研究了用户旅程，他们这么称呼它。

You've really been studying up on the user journey, as they call it.

喜欢。

Like it.

好的，在你离开之前，你曾在专栏中提到一些正在尝试解决这个问题的初创公司，以及其中蕴含的机会。

Okay, before we let you go, you talked a little bit in the column about the startups that are trying to tackle this issue and the opportunity that stands there.

我们应该关注哪些公司？你正在和哪些人交流？

Who should we be watching and who are you talking to?

所以我想象，这些数据标注公司正开始扩展到音频数据领域。

So I'd imagine that, you know, all these data labeling firms are starting to expand into audio data.

但有一个我在这篇文章中关注的、尚处于早期阶段且正在崛起的初创公司，名叫Poseidon AI。

But, you know, one startup that's kind of early and up and coming that I focus on for the piece is called Poseidon AI.

这家公司开发了一款应用，世界各地的用户都可以上传自己朗读不同主题（如客户服务或法律）文本的音频文件。

So this company basically has an app where, you know, any sort of user around the world can upload audio files of them, you know, basically reading out loud different transcripts on different topics like customer service or law.

但对于Poseidon和其他试图做类似事情的初创公司来说，要做好这一点非常困难，因为你可以想象，他们需要特殊的技术来确保用户确实按照脚本朗读，而不是偏离脚本或使用与要求不符的语言。

But this is very difficult for Poseidon and other startups trying to do this to do very well because you can imagine that they need special technology to make sure that people are, you know, actually following the script correctly, not going off script or speaking a different language than what they're supposed to.

因此，这些初创公司必须开发自己的专有技术，以确保音频数据达到尽可能高的质量。

So that requires these startups to build their own kind of proprietary technology in order to make sure that the audio data is of the highest quality possible.

好了，Stephanie，非常感谢你来参加这次对话。

Well, Stephanie, I want to thank you for coming on.

这位是斯蒂芬妮·帕拉佐洛，我们《The Information》的AI记者。

That is Stephanie Palazzolo, our AI reporter here at The Information.

随着广告和AI聊天机器人热度上升，投资于那些旨在抓住这一浪潮的初创公司也在增加。

As advertising and AI chatbots starts to heat up, so too are investments in startups that are offering products to ride that wave.

Koa就是其中一家公司。

Koa is one of those companies.

这家初创公司本周完成了由Theory Ventures领投的2050万美元A轮融资。

The startup raised $20,500,000 this week in a series A funding round led by Theory Ventures.

我想邀请Koa的联合创始人尼克·伯德和Theory Ventures的普通合伙人托马斯·杜恩古斯，帮我们深入解析这笔交易以及接下来的发展。

I want to bring on Koa co founder Nick Baird and Theory Ventures GP, Tomas Dunguz, to help us unpack the deal and what's coming next.

尼克和托马斯，欢迎来到TI TV。

Nick and Tomas, welcome to TI TV.

很高兴你们两位能来。

It's great to have you both here.

能来到这里我们也很高兴。

It's great to be here.

谢谢你们邀请我们。

Thanks for hosting us.

不客气，托马斯，托马斯，这次你带了朋友一起来。

My pleasure, Tomas, Tomas, you you brought a friend with you this time.

通常通常，就你一个人。

Normally normally, it's just you.

人多更热闹。

The more the merrier.

我们非常高兴能参与其中，Theory 对能与尼克合作感到兴奋，正如你所知，我曾在 AdSense 团队担任产品经理，而人工智能领域正在全面接管，其中很大一部分变现和未来发展方向将由协同合作推动。

I'm thrilled to be part we're thrilled at Theory to be partnering with Nick, and as you know, I was a product manager on the AdSense team, and the world of AI is absolutely taking over, and a big part of that monetization, and the future of it will be driven by co op.

好吧。

Well, okay.

那么，托马斯，我稍后再找你。

So, Tomas, I'm gonna come back to you.

我先问尼克。

I'm gonna I'm gonna go to Nick first.

我们先跟他聊一会儿。

Let's let's talk to him for a while here.

所以，尼克，跟我们说说Koa是什么，公司成立多久了？

So, Nick, tell us what Koa is, and how old is the company even?

是的。

Yeah.

公司非常年轻。

The company is very young.

我们在这一领域进展迅速。

We've moved quickly in this space.

我狗在背景里。

I had my dog in the background.

所以我们成立刚过一年。

And we so we're a little over a year old.

产品已经上线十三个月了。

The product's been live for thirteen months.

我们基本上做的就是帮助AI的应用层通过赞助来实现变现，非常简单。

And what we do basically is we we help the application layer of AI monetize via sponsorships very simply.

所以我们安装SDK。

So we install SDKs.

通过这些SDK，使用AI、生成式界面或任何交互式动态功能的应用程序，可以投放一种为它们量身定制的原生应用格式，帮助它们维持增长、实现收入增长，并覆盖部分推理成本。

And from those SDKs, the the applications that use AI, that use generative interfaces, that use kind of anything that's interactive and dynamic, can serve basically a native app format that works for them, that can help them sustain their growth, their revenue growth, and and, you know, cover some of those inference costs.

好的。

Okay.

那我们再简单地分解一下。

So let's break it down a little more simply here.

一方面，你有我们都熟悉的聊天机器人。

So I mean, on one hand you have the chatbots that we're all used to.

另一方面，你也有一些出版商，我想。

You also have some of the publishers, I guess.

我想到的是像Uber Eats这样的打车平台，对吧？那里有广告。

I'm thinking about the ride hailing Uber Eats type of platforms, right, and there's advertising there.

然后还有品牌方。

And then you have the brands.

所以帮我理清一下，你们在其中扮演什么角色，你们的实际产品是什么，以及你们的客户是谁。

So just help me understand where you fit in there, what the actual product is, and who you're selling to.

是的。

Yeah.

我们采用的是平台模式，类似于Uber作为司机和乘客之间的中介。

So we're marketplace model similar to, you know, Uber being the intermediary between the driver and the rider.

对吧？

Right?

我们则是广告位所有者（拥有广告空间的应用）和广告主之间的中介。

And so we are the intermediary between the inventory, the apps that have space to advertise on, and the advertisers themselves.

明白了。

Okay.

那么，你们向双方提供的实际产品到底是什么？

And so what what is the what is the actual product that you're selling to to both parties?

是的。

Yeah.

所以对于广告主来说，这类似于他们在Google AdSense上的横幅广告、Google AdWords上的搜索广告，或者实际上非常类似于ChatGPT正在做的事情，对吧？

So basically, for the advertisers, similar to how they might show up in a banner ad on Google AdSense or a search ad on Google's AdWords or or, you know, really actually very similar to what ChatGPT is doing, right?

广告主可以进来购买广告位。

Advertisers are able to come in and buy slots.

我们销售的基本上是一个HTML代码片段，你可以这么理解，这是我们从发布者或AI应用的广告位中划分出来的一部分，用作赞助帖子或赞助内容的标识。

And what we're selling is basically an HTML snippet, if you will, a part of the publisher or the AI applications real estate that we're carving out as a separate entity to, you know, mark this as a sponsored post or a sponsored piece of content.

因此，广告主可以确保他们的内容出现在模型回复旁边。

And so the advertiser gets to make sure that they show up, you know, alongside the model response.

所以，尼克，在OpenAI正在推进他们自己的广告业务的世界里，为什么这不是他们自己就能做出来、自己开发这些产品的事情呢？

So in a world, Nick, where OpenAI is pursuing their own ads, I mean, why is this not this not something that they could just do themselves and and create these products themselves?

是的。

Yeah.

我的意思是，这是个很好的问题。

I mean, it's a great question.

我认为这个问题的简单答案是，看看移动应用市场。

And I think the easy answer to that is that look at the mobile marketplace.

对吧？

Right?

在移动领域，我们喜欢使用这些免费的应用程序，它们由广告驱动，而且没有任何一家公司占据超过50%的移动广告市场。

So in mobile, you know, we love having all these apps that are free and they're powered by ads, and there is no player that owns more than 50% of the mobile ad market.

然而，AppLovin这家公司却凭借40%的移动市场份额，估值达到了2250亿美元。

And yet AppLovin exists as a company that's worth $225,000,000,000 on 40% of the mobile market share.

对吧？

Right?

所以，每个人都有可能做到这一点。

So it is possible that everyone is able to do this.

顺便说一句，这甚至超越了谷歌的AdMob产品，而谷歌本身已经是巨头了。

And by the way, that's, you know, beating Google's AdMob product, and that's a huge company in its own.

对吧？

Right?