Atlassian CEO谈数据大战、Klarna IPO、甲骨文股价飙升与微软Anthropic交易 | 2025年9月10日

欢迎各位收看Information's TITV节目。我是Akash Pashricha。今天是9月10日星期三，我们将关注Klarna公司终于上市后的股票表现。我们对此话题进行了大量报道，同时还将讨论这次IPO中的大赢家。请不要走开，因为大约25分钟后，我们将为您带来与Atlassian首席执行官精彩对话。

Welcome everyone to the Information's TITV. My name is Akash Pashricha. It is Wednesday, September 10, and we are watching Klarna shares today as the company finally goes public. We've got a lot of coverage on that topic, but we're also gonna talk about who the big winners are in this IPO. And don't go anywhere because in about twenty five minutes, we have got a great conversation for you with the CEO of Atlassian.

我们将讨论他们收购大型浏览器公司、企业数据大战，以及该公司如何内部应用人工智能。最后，我们还将独家解析昨日曝光的消息：微软正转向Anthropic以减少对OpenAI的依赖。今天内容非常丰富，我甚至还没提到甲骨文。实际上我们也会讨论甲骨文的情况。

We're talking about their big browser company acquisition, about the corporate data wars, and about how the company is using AI internally. And finally, we've also got the inside analysis of a scoop that we published yesterday that Microsoft is now turning to anthropic as it seeks to become less reliant on OpenAI. There is a lot going on. I didn't even mention Oracle. We're actually gonna talk about Oracle as well.

开场时没有提到。我们将讨论甲骨文股票——今日涨幅近40%，但让我们先从今早Klarna即将上市的大新闻开始。这家公司走过了漫长道路，近二十年历程。这次IPO最有趣的一点在于，它实际上是对'先买后付'整个行业的全民公投。

It wasn't in the intro. We're gonna talk about Oracle shares, which are up nearly 40% today, but let's start with the big news this morning that Klarna is set to debut. Well, it has been a long road for the company. It has been almost twenty years. And really, Klarna I mean, one of the interesting parts of this IPO is that it is really a referendum on buy now, pay later as an industry as a whole.

因此我想邀请一位能分享行业前景见解的嘉宾。Mercedes Bent是新创投公司Premise的创始人，这是她首次做客本节目。Mercedes，欢迎来到TI TV，非常高兴您能来。

And so I want to bring on someone who can share with us her perspective on where she's seeing this industry go. Mercedes Bent is the founder of a new venture firm called Premise. It is her first time on the show. Mercedes, welcome to TI TV. It's great to have you.

非常感谢邀请，Akash。

Thanks so much for having me, Akash.

我们尚未在本节目中讨论过'先买后付'模式。所以我想暂时跳出Klarna，请您谈谈当前这个行业及整个领域的发展阶段。

So look, we have not talked about buy now, pay later on this show yet. And so I want to take a step back from Klarna for just a minute. I mean, talk to us about where we are in the buy now, pay later story and the sector at large.

这个行业几年前作为年轻一代获取信贷和进行线上消费的新方式变得非常流行。作为风投，我关注的核心问题是：这个品类能否成为用户的首选？即成为他们主要使用的交易工具。说'掏出钱包'这个词时就能看出我的倾向性了。

So the industry is one that I think a couple of years ago became very popular as a new way for the younger generations to access credit and to really spend their transaction money online on the Internet. The big question with the category as a venture capitalist, my question is, is this category going to become one that has primacy with users? Meaning, they're top of wallet. They are the main provider that is being pulled out at transactions. Pulled out is actually terminology still where obviously you can see my bias.

我指的是信用卡。我认为我们需要审视这个行业是否已经发展到能够触及那种首选钱包卡片或支付方式的成熟阶段。这正是我在观察像Klarna这样的公司时所关注的。

I'm referring to credit cards. And I think this is something where we need to look at if the industry has matured to the point of reaching that first kind of top of wallet card or payment method. And that's really what I'm looking for when I look at companies like Klarna.

那么明确一下，你的意思是押注这些公司将完全取代信用卡吗？

So just to be clear here, mean, is the bet here with these companies they will replace credit cards entirely?

要知道，我认为现在下结论还为时过早。从先买后付（BNPL）的角度看，目前还没有完全替代信用卡的实体方案。显然，这些公司正在附加和交叉销售多种产品。作为多年参与消费者信贷历程的人——我在一家私营后期信用卡公司担任董事——

You know, I think that would be too early to say. I mean, there isn't exactly a physical replacement for credit cards from a BNPL perspective yet. Obviously, these companies are tacking on and cross selling a lot of different products. As someone who has been, you know, involved in the consumer credit journey for a number of years, I'm on the board of a private late stage credit card company.

没错。

Right.

我密切关注这些在线交易的发生场景，无论是卡在场还是不在场。我不认为先买后付会取代信用卡，但是

I really closely watch where these transactions are happening online, card present, card not present. And I don't know that BNPL is going to replace cards But

但但但这里的理念是，年轻一代可能会转向这些先买后付平台。希望在于它们能某种程度上替代信用卡，对吧？

but but but the idea here is that it kinda it I mean, the younger generations I mean, the the hope here is that they sort of turn to these buy now, pay later platforms. The hope is a that credit card. Right?

我认为这确实是先买后付公司的期望，这对传统信用卡公司构成了实质威胁。比如观察Carta的购买频率与美国平均信用卡的对比——这是他们在S-1文件中披露的——其月均交易频率仍仅为美国信用卡平均水平的50%左右。这表明他们仍有发展空间，但提升使用频率显然是他们的重点战略方向。

I think that's the hope certainly of the buy now pay later companies that is a real threat to the only credit card companies. If you look at, for example, Carta's purchase frequency compared to an average US credit card, this is something put they put in their s one. They're still at about 50% of the monthly credit card frequency purchases for an average US credit card. So that shows they still have a ways to go. But I think that is something they're highly focused on.

那么让我们来谈谈这里的商业模式。我的意思是，我相信你已经看过Klarna的财务状况了。从商业模式的角度来看，'先买后付'是一门好生意吗？

And so let's get down to the business model here. I mean, you've looked at Klarna's financials, I'm sure. Is buy now, pay later a good business from a business model perspective?

是的。我通常关注的是，这家公司能否实现盈利性增长？在信贷业务中，以50美分的成本放出一美元是很容易的。所以问题在于，他们的毛利率是多少？他们的队列单位经济状况如何？

Yes. I mean, what I look at always is, what is this company able to grow profitably? It's really easy as a credit business to give away a dollar for 50¢. And so the question is, what are their gross profit margins? What are their cohorted unit economics looking like?

净息差是多少？我会深入研究许多信贷特有的指标。但宏观来看，仅看毛利率的话，Klarna在2024年实现了44%的稳健毛利率。这强烈表明了他们的市场定位——要知道信用卡业务的毛利率通常不会像软件SaaS产品那样达到70%或80%。所以这个表现很出色。

What's the net interest margin? There's a lot of credit specific metrics I would dive into. But on a very high level, if you just look at the gross profit, you know, Klarna had solid gross profit last year, 44% in the year of 2024. And that is a strong indication of where they're at relative to, you know, credit card gross profits are not often in the seventies or eighties like a software SaaS product. So it it's strong.

如果你再观察信贷损失占贷款账簿的比例，他们大约是同行水平的50%。这也体现了他们在信贷账簿管理上的优势。

If you also look at credit losses as a percentage of their loan book, they are at about 50% of their peers. So that also shows help in their credit book.

那我们来谈谈同行竞争者。有Klarna，有Affirm，有属于Block的Afterpay，甚至本周PayPal也加入了。

So let's talk about the peers. You've got Klarna. You've got Affirm. You've got Afterpay, which is part of Block. You even had PayPal this week.

PayPal推出了'先买后付'服务，其CEO表示这将成为他们增长战略的一部分。能否帮我们区分这些平台？它们是完全同质化的，还是各有差异化？

PayPal has a buy now pay later offering, the CEO says that he sees BNPL being part of their growth strategy. Can you just help us differentiate between all these platforms? Do all these platforms literally do the same thing, or how do they differentiate themselves?

我比较这类平台的一个维度是：它们主要用于什么类型的消费？比如Affirm以支持较高客单价著称，像他们与Peloton的合作就以融资数百甚至上千美元的大额消费闻名；而Klarna更专注于日常小额高频消费。在信贷领域，我们常问关于'钱包首位度'的问题——回到我先前的分析框架——即是否被用于食品、汽油等日常消费？当然这通常指线下刷卡交易，而我现在讨论的是无卡支付，其实甚至不该用'卡'这个词。

One of the vectors that I tend to look at for comparing this is what are the type of purchases they're being used for? Some of the players for example, Affirm is more well known for being larger average order values, larger purchases. You know, Affirm was famous for their Peloton partnership and being used to finance really large, let's call it, on the order of high hundreds of dollars, even thousands of dollars purchases versus Klarna was more well known for smaller, more frequent purchases your everyday. In the credit world, one of the questions we always ask about primacy of wallet, going back to one of my earlier kind of frameworks, is that are we being used for food, gas, and kind of the every at average everyday purchases? Now, obviously, that's offline transactions, card present, and I'm talking about card not present, and really, I shouldn't even be using the word card.

但我认为Klarna令人兴奋之处在于，他们成功触及了更普遍的日常消费行为。

And so but what I do think is exciting about Klarna is that they have tapped into the more average everyday purchase behavior. And

我想就此插一句，这正是我一直在思考的问题——我们听过这些'先买后付'的故事。如今购买金额正变得越来越小，小到甚至可以用分期付款买个三明治。我在想，当我们试图进一步渗透这个市场，或者说扩大支付领域的市场份额时，如果人们已经将其用于如此小额消费（而非信用卡），是否意味着市场已接近饱和？

I just want to jump in on that because that is something that I've been thinking about is that we've heard these stories about buy now what was it? Eat now, pay later. The idea that the size of the purchase is getting smaller and smaller and smaller. At this point, you can even buy a sandwich and you can pay later. And one of the things I've been thinking about is, as we think about further, I guess, penetrating this market, or rather increasing market share in the payments landscape, has it kind of reached a point of saturation a little bit if people are already using it for these really small purchases that, you know, I don't just use a credit card?

这个...我也不确定。

I'd like I don't know.

你提出了个价值十亿美元的问题，Akash。我认为美国消费者信贷已达顶峰——今年早些时候我们刚见证市场创下历史最高信贷规模，虽然上季度略有回落。但核心问题始终是：美国民众真的能负担日常生活吗？

You're asking the billion dollar question, Akash. I think that consumer credit in The US has reached a height. Earlier this year, we reached and we constantly are on this journey, but we reached the highest amount of consumer credit that the market has ever seen. It came down a little bit in the last quarter, but there's constantly a question of, are Americans okay? Are they able to afford their average daily life?

这其实是个更宏观的社会议题。先买后付平台和信用卡公司本质是杠杆工具，为人们提供资金流动性。虽然不能单方面归咎于它们，但确实需要全社会共同审视这个现象。

And I think that's a larger societal question that, you know, players like BNPL and credit card companies are facilitators of leverage and enabling people to have liquidity and how they move in and out of their daily finances. I don't think you can, you know, speak standpoint to them. But I do think it's something that more holistically as a society we have to consider.

说得好。Mercedes非常感谢你参与节目。未来几周随着股价波动，我们肯定还会继续讨论Klarna。你提到的消费者支出趋势这个大问题，我们完全可以再聊上几个小时——下次务必请你再来深入探讨。

Great. Well, Mercedes, thank you so much for coming on the show. We're going to be talking about Klarna more in the weeks to come, I'm sure, as we see shares move. And I think that broader question that your points to here, which is where are we in consumer sending? I mean, we could talk for many more So hours about that we'll have you back on to talk about that.

这位是Mercedes Bent，新兴风投公司Premise的合伙人。另外，佛罗里达IPO将成为众多风投的盛宴，而最大赢家非红杉资本莫属——考虑到多年来围绕红杉董事会的戏剧性事件，此刻堪称圆满收官。

That is Mercedes Bent. She is partner at a new venture firm, Premise. Okay. The Florida IPO is also set to be a big winner for many venture capitalists, but one of the biggest winners is none other than Sequoia Capital. It is certainly a full circle moment because of all the board of directors drama over the years that very much centered around longtime Sequoia leaders.

我想请我们的风险投资副主编凯蒂·鲁夫来为我们解释她在报道中的发现。凯蒂，这是你第一次上节目。欢迎你。很高兴你能来。

I want to bring on Katie Roof, our deputy bureau chief of venture capital, to explain to us what she found in her reporting. And Katie, it is your first time on the show. Welcome. It's great to have you.

很高兴来到这里。是的。所以，毫不意外，红杉资本在那里赚了一些钱。他们通过长期持有Klarna的股份赚了近30亿美元。他们仍然保留了超过20%的股份，尽管他们已经是股东十五年了，但并没有出售股份。

Great to be here. Yeah. So, you know, no surprise Sequoia made some money there. They made almost $3,000,000,000 off of holding Klarna long term. They still retained a more than 20% stake because they didn't sell shares even though they've been a shareholder for fifteen years.

好的。那么红杉是唯一的赢家吗？还是说还有其他大赢家？

Okay. And is Sequoia the only winner or who are the other big winners?

这对IPO来说非常不寻常。其他大股东并不是典型的金融投资者。第二大股东是一家家族办公室，丹麦的Heartland家族办公室。他们拥有公司约10%的股份。许多金融股东不在股东名单上的部分原因是，与红杉不同，他们中的许多人，比如General Atlantic，已经出售了股份。

So this is very unusual for an IPO. The other large shareholders are not typical financial investors. The next largest shareholder is a family office, a Danish family office called Heartland. They own about 10% of the company. Part of the reason that a lot of the financial shareholders aren't on the cap table is unlike Sequoia, many of them, such as General Atlantic, sold shares.

好的。我是说，让我们回到红杉。红杉与这家公司有一段有趣的历史。我们在这里报道过，所有的董事会戏剧性事件等等。我的意思是，天哪，跟我一起回顾一下吧。

Okay. And I mean, let's go back to Sequoia. Sequoia has this kind of interesting history with the company. We've covered it here at the information, all of the the boardroom drama and stuff like that. I mean, gosh, like, just reflect with me a bit.

这有点像是它通往公开市场的曲折道路，现在红杉，你知道的，坐拥30亿美元的收益。这真是相当了不起。

This is kind of it's had a winding road to the public markets, now Sequoia, you know, it's sitting on a $3,000,000,000 gain. This is kind of outstanding, really.

当然。我是说，几年前当科技行业出现调整时，这对金融科技的影响尤为严重，这种紧张局势导致了Klarna董事会内部的戏剧性事件，包括红杉公司内部。所以最终，作为某种和平协议，安德鲁·里德代替马修·米勒加入了董事会。而迈克·莫里茨虽然仍在董事会，但不再拥有红杉的董事会席位。所以，关于如何管理公司以及最终IPO的计划，确实有一些戏剧性的事件。

Sure. I mean, a couple of years ago when there was the tech correction, which disproportionately impacted fintech, that that tension led to drama within the board of directors at Klarna and including within Sequoia the firm. And so ultimately, as some sort of peace agreements, Andrew Reed joined in lieu of Matthew Miller. And he's in Mike Moritz, who is still on the board, but no longer has Sequoia's board seat. So there was a bit of, you know, a bit of bit of drama there over how to govern the company and plans on the eventual IPO.

但大家都在赚钱，所以没关系。对吧。

But everyone's making money, so it's fine. Right.

最终他们确实赚了钱。没错。是的。

In the end, they made money. Exactly. Right.

对。没错。实际上是30亿美元。嗯，我想从广义上谈谈这次IPO。你知道，你在《信息》杂志上报道了所有风投和IPO相关的内容，我们在节目中也一直在探讨这个问题：什么时候才是上市的最佳时机？

Right. Yeah. Dollars 3,000,000,000, actually. Well, look, I I wanna talk a little bit about the IPO broadly speaking. You know, you cover everything venture capital, everything IPOs for the information, and, you know, we've been sort of assessing this question on the show about when is the right time to go public?

现在是上市的好时机吗？你认为Klarna选择现在上市是个好决定吗？你怎么看？

Is this the moment to go public? Do you think this was a good moment for Klarna to go public? What's your sense of that?

当时股市正处于历史高点，所以他们可能选了个完美时机。但总体而言，今年IPO表现都非常好。科技股IPO，比如最近的Figma就大幅上涨。显然Klarna也预期如此——他们将发行价定在区间上限再高出3美元，IPO行家都知道这预示着强劲的市场需求。

The stock market was at a record high, so maybe they timed it perfectly. But in general, IPOs this year have been performing very well. Tech IPOs, you look at you know, recently, Figma just popped significantly. And clearly, Klarna is predicting the same. They priced $3 above the range after raising the range, and IPO nerds know that that means that they're expecting very strong demand.

基本上可以预见今天股价会大幅飙升。

And it's it's basically assumed that it's gonna pop significantly today.

你目前还关注哪些其他公司？或者你名单上正在向人们打听的名字有哪些？

And who are the other names that that you're watching or that you're asking people about on your list right now?

实际上，本周还有另外三家风投支持的IPO。首先是Gemini，那个与加密货币相关、由Winklevoss双胞胎兄弟创立的项目。

Well, actually, there's three other venture backed IPOs just this week. There's Gemini, the crypto related thing connected to the Winklevoss twins.

没错。

That's right.

其次是Via，一家交通服务公司，还有Figure，同样属于金融科技领域。

There's Via, the transportation service, and then there's there's Figure, also a fintech.

我很喜欢你用'加密货币相关项目'这种说法，因为这些本质上都是如此。到头来它们终究只是些概念性产物。不过我想问的是——这也是我们几乎对每位上节目的投资人都会提的问题——现在算是IPO窗口期开启了吗？根据你的交流情况，你觉得人们是否正争先恐后地涌向这个机会？

I I like how you said crypto related thing because that's it's it's kind of what all these things are. They just end up being things at the end of the day. No. But I mean, this is kind of a you you know, this is a question that we ask just about every investor that comes on the show is, is this the opening of the window? And from your conversations, I wonder, are people rushing to the floodgates now?

你观察到这种趋势了吗？

Do you see that happening?

没有想象中那么明显。应该说这个窗口期已经开放超过一年了，因为自从科技股反弹后，市场对科技公司上市的兴趣就一直存在。但考虑到现存独角兽企业超过一千家，真正上市的占比非常小。越来越多公司开始意识到这点，不过私募市场仍有大量可用资金，这反而降低了企业IPO的紧迫性。

Not as much as you would think. I would say the window has actually been open for over a year because once the public tech stocks rebounded, the appetite for new tech listings also was there. But considering how many unicorns are there's over a thousand, we've really seen a very small percentage going public. I think that more and more are starting to get the memo, but there's also a lot of available capital on the private markets, which hasn't really motivated people to go for the IPO.

明白了。太棒了。凯蒂，这么说吧，更多IPO意味着能和你多聊聊。作为我们新上任的风险投资副主编，我相信你将会站在许多这类事件的最前沿。感谢你参加本期节目。这位是来自The Information的凯蒂·鲁夫。

Right. Great. Well, Katie, I'll tell you what. I mean, more IPOs means more talk time with you, and I'm excited to talk with you more because as our newest deputy bureau chief of venture capital, I think you're going to be front and center for a lot of these Thank you for coming on the show. That is Katie Roof from The Information.

好的。我们已经在节目中讨论过微软与OpenAI的关系近年来如何变得越来越复杂。最新的迹象是，本周《信息报》率先报道微软现在将在其部分工具中使用Anthropic的技术。众所周知，Anthropic是OpenAI最大的竞争对手之一。我想请报道此事的微软记者Aaron Holmes加入我们，分享他了解到的更多信息。

Okay. We've talked on the show already about how Microsoft and OpenAI's relationship has grown increasingly complicated over the years. In the latest sign of that, the information was first to report this week that Microsoft will now use Anthropic's technology for some of its tools. Anthropic, of course, is one of OpenAI's biggest rivals. I want to bring on our Microsoft reporter, Aaron Holmes, broke that story to share more about what he's learned.

Aaron，欢迎回到节目。很高兴你能来。

Aaron, welcome back to the show. It's great to have you.

嗨，Akash。

Hi, Akash.

好的。那么告诉我们你了解到了什么。

Okay. So tell us what we learned.

是的。我的意思是，我们在这里看到的是，长期以来微软完全或几乎完全依赖OpenAI的模型来驱动其办公软件中的AI功能，这些功能理论上可以创建PowerPoint演示文稿或编辑Excel电子表格。但我现在听说微软正准备推出新版本的Office Copilot，这些版本将使用Anthropic模型来支持一些更高级的功能。而且，有很多猜测认为这与OpenAI和微软之间正在进行的谈判有关。但据我所知，实际上更多是因为模型本身的质量，以及微软发现Anthropic模型在自动化某些任务方面比OpenAI的模型更出色。

Yeah. So, I mean, what we're seeing here is for the longest time, Microsoft has relied entirely or almost entirely on OpenAI's models to power the AI features in its office software that, you know, can theoretically create PowerPoint presentations or edit Excel spreadsheets. But what I'm hearing now is that Microsoft is getting ready to launch new versions of Office Copilot that are using Anthropic models to to power some of those more advanced features. And, you know, a lot of there's a lot of speculation that this has to do with the negotiations that are going on between OpenAI and Microsoft. But, you know, what I'm hearing is it's actually just more about the quality of the models themselves and the fact that, you know, Microsoft found Anthropix models are simply better at automating some of these tasks than OpenAI's models were.

所以我认为，根据我所听到的，这实际上更多是一个基于产品的决定。

So I think really what this comes down to is a product decision more than anything based on what I'm hearing.

所以这实际上并不太反映双方关系，而是反映了模型的质量。

So this actually isn't really so much a reflection of the relationship. This is actually a reflection of the of the quality.

没错。而且，你知道，微软一直非常坚定地表示仍将OpenAI视为其默认的AI提供商。总的来说，我认为微软在绝大多数AI工具上仍会依赖OpenAI。但这确实说明了，尽管我们不断听到关于AI市场日益商品化的讨论，但在某些企业任务上，仍存在明显的领先者。目前这个领先者就是Anthropic。

That's right. And, you know, Microsoft has been pretty adamant that it still considers OpenAI its default AI provider. And, you know, in general, I think we're we're still going to see Microsoft relying on OpenAI for the vast majority of its AI tools. But I think this really kind of speaks to the fact that, you know, for all of the talk we've heard about the AI market becoming more and more commoditized, there is still, in some cases, a really clear front runner for certain enterprise tasks. And right now that front runner is anthropic.

哇。

Wow.

我一直在思考其中的复杂性，因为就Anthropic而言，它获得了谷歌的投资，亚马逊也是其大股东。而亚马逊和谷歌显然是微软的主要竞争对手。微软愿意在此选择Anthropic，虽然这在大型企业间并不罕见，但足以说明这些模型肯定有显著优势。

One of the things I've been thinking about is sort of the complexities here, because as it relates to Anthropic, I mean, Anthropic has got an investment from Google, and also Amazon is a big investor in Anthropic. Amazon and Google are obviously big rivals with Microsoft. So the fact that it's willing to go with Anthropic here, you know, I mean, look, these are big businesses. I mean, it's not like this is uncommon. But gosh, the models must be quite a bit better than.

是的。更值得注意的是，微软将通过托管Anthropic的亚马逊云服务（AWS）支付这些模型费用——而亚马逊正是微软在云计算领域的最大对手。此外，微软因向OpenAI投资超130亿美元并达成深度协议，已拥有自主使用其模型的权利。这进一步表明，微软当前优先考虑的是优化Office 365 Copilot功能、取悦客户，且不惧尝试多种模型来实现目标。

Yeah. Mean, it's also pretty remarkable that we're going to see Microsoft pay for these models via Amazon Web Services, which hosts Anthropic. Amazon is, of course, Microsoft's biggest rival in the cloud. And, you know, Microsoft also already has the rights to use OpenAI's models on its own without paying OpenAI for them because it's invested more than 13,000,000,000 into the startup and has this, you know, really deep deal. So I think that just speaks even more to how Microsoft right now is prioritizing, trying to get Office three sixty five Copilot to work as well as possible and to really charm customers and that it's not afraid to branch out and use multiple different models to do that.

能否谈谈客户对Copilot产品或Office 365的反馈？微软在推广这些产品时是否顺利，还是遇到了一些问题？

Can you talk a little bit about how customers have been finding the Copilot products or the Office three sixty five? Have have has Microsoft been very successful at sort of getting acceptance among customers with them, or have there been some hiccups?

目前反响参半。我了解到客户非常喜爱Copilot的某些功能，比如参与Teams会议并发送摘要、生成邮件、总结公司文档等。但在更高级的代理功能上存在挑战，例如处理Excel数据或生成完整PPT演示文稿时仍有漏洞，微软正在努力解决这些问题。

You know what? There's been a mixed response so far. I I've heard that customers really love certain features in Copilot such as, you know, the ability for Copilot to sit in on your Teams meetings and send you summaries, the ability to generate emails, summarize company documents. And I think where they're struggling is with these more advanced agentic features, things like, you know, manipulating data in an Excel spreadsheet or generating an entire PowerPoint presentation is where we've heard customers say that there's still some bugs that I think Microsoft is now trying to work out.

明白了。最后一个问题：虽然我们强调这关乎产品而非合作关系，但就微软与OpenAI关于治理结构等谈判而言，你认为这会否影响双方谈判筹码，还是说两者互不干涉？

Got it. Got it. And last question for you. I mean, you know, again, we've we've made it clear this is about the product, not about the relationship. But as it relates to Microsoft's negotiations with OpenAI about their governance structure and stuff like that, Do you see this having any impact or giving one party more or less leverage in those conversations at all, or are they sort of separate?

要知道，我认为微软和OpenAI都有充分的理由达成一项协议，允许OpenAI改变其结构、继续融资并最终上市。所以我最终不认为这会阻碍他们关于该协议的谈判。没错。但更广泛地说，这确实传递了一个信息，即归根结底，微软希望其产品成为业界顶尖，如果这意味着能从竞争对手那里获得更优性能，它不会必然依赖已达成协议的初创公司来实现这一目标。

You know, I think that both Microsoft and OpenAI have a good reason to, you know, reach a deal that would let OpenAI change its its structure and continue to raise money and eventually go public. So I ultimately don't think that this is going to hamper their their talks on that deal. Right. But I think more broadly, it does, you know, send a message that at the end of the day, Microsoft wants its products to be best in class, and it's not going to necessarily rely on the startup that it already made a deal with to do that if it means that it can get, you know, better performance from a competing startup.

太棒了。这真是个引人入胜且重大的故事。感谢亚伦的到来。这位是报道微软资讯的亚伦·霍姆斯。好的。

Great. Well, it was a fascinating story and a big one too. Thank you, Aaron, for coming on. That is Aaron Holmes who covers Microsoft for the information. Okay.

甲骨文今天在市场上表现惊人。今早股价上涨40%，源于公司大幅上调收入预期。现在最大的问题是：这能否被称为甲骨文的‘英伟达时刻’？我想请来KeyBank分析师杰克逊·阿德为我们解读。杰克逊，很高兴再次邀请你上节目。

Well, Oracle is having a blockbuster day on the markets. The stock was up 40% this morning after the company's big revenue projections. The big question now is could you call this Oracle's NVIDIA moment even? I want to bring on Jackson Ader, who covers the company at KeyBank, to help us make sense of all this. Jackson, it's great to have you back on the show.

谢谢。是的，阿卡什，很高兴来到这里。

Thank you. Yeah, it's good to be here, Akash.

好的，请为我们详细说明。昨晚的具体预测是什么让投资者今早如此狂热？

Okay, so lay it out for us. What exactly was the projection last night that has investors in a frenzy this morning?

预测涉及OCI未来四到五年的收入目标。这些数字远超市场对OCI的普遍预期，尤其是进入2028、2029财年及之后——我的意思是，届时OCI贡献值可能不会达到共识预期的两倍，但也相差不远。你绝对...

The projection would be the OCI revenue targets for the next four to five years. They are so far above a street or consensus expectations for OCI, particularly once we get out into fiscal twenty eight, '29, and beyond that, I mean, you you're talking about maybe not doubling consensus estimates for OCI contribution in those out years, but close to it. You you're definitely

你似乎无言以对。你几乎像是说不出话了。

you seem speechless. You seem speechless almost.

我是说，昨晚我并非完全无言以对。我们在笔记中提到过，当时在办公室看到数据时，我们忍不住发出惊叹，甚至不得不下楼向家人解释为何如此喧闹。但就数据本身而言——不是预测值——那个RPO数字确实令人震惊。他们公布的预订量超过3300亿美元。要知道，过去四个财年他们的总预订量加起来都没达到3300亿美元。

I mean, I wasn't necessarily speechless last night. I think we said in the note, you know, it's like we were, you know, audibly reacting to the print here in our office and had to go downstairs and explain to the family, you know, why we were so loud. But, you know, as far as the print, not the projection, it was that RPO number that was truly stunning. They they put up bookings of over $330,000,000,000. They didn't do $330,000,000,000 in total bookings in the last four fiscal years combined.

给大家解释一下，这基本上预示着他们未来将确认为收入的业务规模，对吧？

And just to translate for people, I mean, this is basically an indication of the business that plan to incur as revenue essentially in the future. Right?

没错。这是已签约的未来收入。虽然尚未体现在资产负债表上，但作为未完成订单属于表外项目。

That's right. This is contracted future revenue. It's not yet on the balance sheet, but it's off balance sheet in terms of backlog.

明白了。他们把这归功于几个大客户，但没透露具体是谁？

Okay. And they chalked this up to a couple of big customers, right? But they didn't tell us who.

是的。他们提到四笔交易，四笔非常庞大、长达多年、涉及三家客户的数十亿美元级交易。但没透露

No. I think they said four deals, four very large and substantial, call it multibillion dollar, multiyear deals spread across three customers. Didn't tell

可能是OpenAI。

us OpenAI. Maybe OpenAI.

对。虽然没明说，但我知道至少有一家对此消息灵通的媒体可能掌握内情。OpenAI确实有可能。关键是公司强调过——其中一笔交易规模特别庞大（他们在8-K文件里披露的300亿美元年合约收入），但其他交易也绝非等闲之辈。若非被这笔超大订单掩盖，这些交易本身都值得单独宣传，都是非常重大的OCI胜利。

Right. Yeah. They didn't tell us who, but, you know, I know at least one media outlet that's been pretty good on this that might have some insight as to who it is. OpenAI. And and, you know, the important thing is, and the company stressed this, that one of these deals is certainly outsized, the one that they released in an eight ks, dollars 30,000,000,000 in the annual contract revenue, but that the others are nothing to sneeze at and would have been worthy of callouts and very substantial OCI wins if it weren't if they weren't kind of dwarfed by this one

关于OCI，为不了解这个缩写的观众解释一下——抱歉，全称是Oracle云基础设施。明白了吧。

And OCI, just for people who don't know the acronym. Apologies. Oracle Cloud Infrastructure. Got it.

这是与谷歌云或微软Azure竞争的云服务业务

It is the business that competes with Google Cloud or Microsoft Azure

或者AWS。对，对。现在我想退一步谈谈这个预测可能伴随的风险，因为这一切都建立在几个大客户在AI领域大量投入资金的前提上。看涨的理由很明确——这可能意味着甲骨文云业务将迎来重大机遇。

or AWS. Right, right. So now I want to take a step back here and talk a little bit about, you know, some of the risks that could come with this projection, because all of this is banking on a couple of big customers spending heavily and investing heavily in AI. And so I think the bull case is pretty clear. You know, this could mean big things for Oracle's cloud business.

那么看跌的观点是否存在呢？比如这是否依赖于这些客户能否筹集大量资金？或者依赖于AI热潮持续？你如何看待这些潜在风险？

Is there a bear case here to be made in that, you know, maybe this relies on those customers being able to raise a lot of money, or this relies on the AI boom continuing? Or how are you thinking about the the risks that come with this?

没错。昨晚我们讨论时说过，这就像庆典上的零星雨点。我们不想彻底浇灭今天甲骨文的庆祝氛围，但确实存在一些隐忧。

Yeah. Yeah. We so last night, we we said this is this is the sprinkling on the parade. Right? We don't wanna fully rain on the parade that's happening today in Oracle, but there are some sprinkles for sure.

先说说融资问题。这涉及两个层面：首先，根据客户身份不同，他们可能需要外部融资来履行这些合同，这就形成了潜在障碍。

So let's talk about financing. It's it's financing on two levels. One is yeah. I mean, depending on who these customers are, they are going to have to raise outside financing to pay for these contracts. That creates a possible hurdle.

尽管目前资本对于需要推理和训练能力的AI公司并不吝啬。另一个融资问题是甲骨文自身的资金需求——我们预计他们未来四年将新增约1000亿美元债务或融资，用于支持建设云基础设施所需的资本支出。这是资金层面的挑战。此外，从甲骨文损益表来看，OCI业务的毛利率远低于传统软件业务。

Although it doesn't seem like capital is shy when it comes to funding some of these AI companies that need the inference and the training. The other financing that you need to be aware of is Oracle's financing. We now assume that they will raise about a $100,000,000,000 in incremental debt, or incremental financing over the next four years in order to fund the capital expenditure investment required to build out this cloud infrastructure. So that's on the that's on the financing side. And then, you know, the other thing is, at least in terms of of Oracle's income statement and Oracle's financials, Oracle Cloud Infrastructure, the OCI business, comes in at a much lower gross margin than your traditional software.

要知道，SaaS软件的毛利率是85%。支持和许可软件的毛利率则高达95%以上。而我们估计OCI（Oracle云基础设施）的毛利率可能在35%到40%之间。为什么？首先，他们尚未享受到规模效益。

You know, SaaS software is 85% gross margin. Support and license software is is 95 plus percent gross margin. And we estimate that OCI is maybe mid to high thirties. Why is that? Because, well, number one, they don't yet have the benefits of scale.

对吧？规模经济效应还未充分体现在这项业务中。其次，这本质上就是不同的商业模式。为客户提供服务器和硬件并按使用量收费，与开发一次软件就能永久分发的情况截然不同。明白吗？

Right? The economies of scale that will flow into this business are not yet there. The other is it's just fundamentally a different business. I mean, you are providing servers and hardware to your customers that will pay you on a usage based model, whereas software, you write it and distribute it forever. Right?

软件确实是绝佳的商业模式。因此短期内，业务结构向Oracle云倾斜会拉低毛利率和运营利润率。但如果营收能实现如此大幅增长，投资者包括我们在内显然愿意暂时忽略这点。

I mean, that it it's a it's a wonderful business model software. Right. So that mix shift more toward Oracle Cloud will weigh on, we think, gross margin and operating margin here in the short run. But if numbers on the top line are gonna be going up this substantially, clearly investors and and ourselves included are willing to look past that.

好的，最后总结性问题。您既分析了风险，也提到诸多需要满足的前提条件，比如双方的融资安排。您认为事态会按当前投资者预期发展，还是未来可能有所偏差？

Okay. And so last question just to wrap things up here. I mean, you know, you've talked about the risks here. You've talked about a lot of the if statements, I guess, that need to fall into place, the financing on both sides. Do you think this plays out the way that investors today are thinking it might, or do you think that the future is looking slightly different?

让我欣慰的是，这些收入预测都有未完成履约义务的积压订单支撑。听着，合同总可能修订——就像您说的，若AI热潮未能兑现，OpenAI完全可能要求重谈合约。

I am heartened by the fact that these are not revenue projections without backlog sitting in remaining performance obligation. You know, a lot of these a lot of the no. Look. Contracts can always be amended. Like you said, if if the AI boom doesn't actually come to fruition, certainly an open AI is gonna say, hey, we need to rework that contract.

他们或许会致歉并表示无力支付费用，这确实是风险。但Oracle的预测令我安心之处在于，有大量已签约订单等待确认为收入，这让我对这些预测多了几分信心。

We're sorry about that, but we just don't have, you know, we we don't have the the money to pay for it. So certainly that is a risk. But I would say in terms of the projections from Oracle, I am heartened by the fact that they have so much bookings waiting to be recognized into revenue. And so that does make me feel a little bit better about these projections.

太棒了。杰克逊，感谢您再次做客节目解读局势。随着事态发展愈发复杂，我们定会多次邀您回来——正如您所言，其中几个关键条件必须顺利达成。这位是KeyBank资本市场软件行业股票研究主管杰克逊·艾德。接下来这场对话同样令人期待。

Great. Well, Jackson, thanks for coming on the show and helping us make sense of it. It's your second time on the show, and I'm sure that we will have you on more and more to come out as this story gets even more complicated, because like you said, there's a couple if statements in there that need to really go the right way. That is Jackson Ader, who leads equity research for the software sector at KeyBank Capital Markets. Okay, well, I am really excited for this next conversation.

Atlassian是一家软件公司，其广受欢迎的协作工具如Jira和Trello已成为科技界许多人的依赖。上周，该公司通过收购浏览器公司下了一大赌注，几乎宣告将全力投身AI浏览器大战。昨晚，我与Atlassian首席执行官Mike Cannon Brookes坐下来进行了广泛讨论，不仅涉及此事，还包括他对企业数据战的看法，以及他对Marc Benioff声称AI能替代4000人工作的大胆言论的评价。我们昨晚交谈了约二十分钟。以下是我与Atlassian CEO的对话。Mike Cannon Brookes，欢迎来到TI TV。

Atlassian is a software company that many people in the tech community have become reliant on with its popular collaboration tools like Jira and Trello, and the company made a big bet last week in buying the browser company, all but declaring that it would basically be jumping with two feet into the AI browser wars. Now last night, I sat down with the CEO of Atlassian, Mike Cannon Brookes, for a wide ranging discussion, not just about that, but also about his take on corporate data wars and what he thinks of Marc Benioff's bold claims that AI can do the work of 4,000 people. We spoke for about twenty minutes last night. Here is my conversation with the CEO of Atlassian. Mike Cannon Brookes, welcome to TI TV.

很高兴你能来。

It's great to have you.

谢谢邀请，伙计。

Thanks for having me, man.

看，Atlassian最近真是激动人心。你们这次收购浏览器公司的交易正被所有人瞩目。上周节目里我们请到你们的产品主管时简单聊过，他大致解释了背后的逻辑。但最让我着迷的是，如果你去看推特和一些博客，有些人对此简直一头雾水。

So look, it's been an exciting time for Atlassian. You guys are the deal makers that everyone's watching now with this recent acquisition of the browser company. You know, we talked about this a little bit on the show last week when we had your head of product on. He kind of explained a little bit of the rationale. But the thing that I've been so fascinated by is that if you go on Twitter and on some of these blogs and stuff like that, I mean, some people are just confused by it.

你知道，人们会在网上随心所欲地发表意见。但有个推文我觉得很典型：有人说Arc卖给Atlassian是今年极客圈最失望的事件。类似观点层出不穷。所以我想请教你，Atlassian收购浏览器公司的真正原因是什么？

And, you know, people are they'll write what they want on the internet. But the tweet that I think sort of encapsulates this is that one person said, Arc selling off to Atlassian has been the biggest letdown for nerds this year. And there are all sorts of variations like that. So I just want to ask you, I mean, talk to me about why Atlassian bought the browser company.

当然。我们对浏览器未来发展方向有着高度共识。首先，我认为从未有人真正为知识工作者——那些整天抱着笔记本、使用SaaS应用的人——打造过浏览器。而ARC是目前最接近这个目标的。

Sure. We have a very shared vision about where we think browsers are going. Firstly, I don't think anyone has ever really built a browser for knowledge workers, people who live at their laptop, who use SaaS apps all day every day. Right. ARC is the closest that has come.

我是Arc的日常用户，多年来一直如此。Atlassian作为浏览器公司的投资人已有相当年头，我们也是SaaS生态的忠实拥趸。市面上有无数SaaS应用存在。

I'm a I'm a daily Arc user, have been for many years. We've been an investor in Arc, an investor in the browser company, I should say. Atlassian has for a fair number of years now. And we're big fans of the SaaS ecosystem. There are many, many SaaS apps out there.

Atlassian始终注重集成能力并与各类工具良好协作，我们计划继续保持这一优势。另一方面，当前AI技术正在颠覆行业格局，我们有机会为终日与笔记本电脑为伴、日常工作离不开SaaS应用的知识工作者重塑浏览器体验。两家公司强强联合后，我们完全具备打造革命性产品的所有要素，这正是我们决心实现的目标。

Atlassian has always had integrations and played well with all of them, and we intend to continue to do so. And the second part of it is obviously the AI disruption that's happening and the ability to change browsers and build a better browser for knowledge workers who live at their laptops, who use SaaS apps all day every day as part of their job. And I think between the two companies, we have all of the ingredients to give that a really red hot go, and that's what we intend to do.

当人们评价这笔交易时说'Atlassian要转向消费者市场了'，您会如何回应？

So when when people look at this deal and they say, oh, Atlassian is pivoting to consumer, what do you say to that?

我们已多次明确否认这种说法。旗下Trello、Loom等专业级工具拥有数千万用户，这些被我们称为'专业消费者工具'的产品既被个人知识工作者使用，也服务于全球各规模企业。我们始终聚焦知识工作者领域，现有30万客户覆盖从5人到50万人的各类企业。

I think we've been pretty clear that that is not the case. We have a lot of prosumer tools, you know, tens of millions of people use Trello, use Loom. We have lots of what we call prosumer tools. I used by knowledge workers, in an individual capacity and also used inside of companies small, medium, and large across the world. We are squarely focused on those knowledge worker companies, and we have 300,000 customers from five person companies up to 500,000 person companies.

没错。我们一直宣称服务对象是'财富500,000强'，突然转向服务家庭用户、旅行者或购物者会显得非常奇怪——这显然不是我们的方向。浏览器公司的创立初衷也是为知识工作者开发产品，这正是我们双方合作的共同目标。

Right. Always said that we're targeting Fortune 500,000, and so it would be strange for us to go out and target, you know, moms and dads or travelers or shoppers or anything else. So, obviously, that's not our focus. Nor would I say is that the founding focus of the browser company. They set out to build a browser for knowledge workers, which is which is what we're all working together to do.

在探讨未来协作方式时，您提到知识工作者这个概念——此前产品负责人做客节目时曾向我们解释过，我们新闻间还开玩笑说'原来我也是知识工作者'。考虑到这些用户群体（包括部分专业消费者），在AI技术蓬勃发展的背景下，您认为3-4年后协作工具会呈现怎样的图景？

So as you think about sort of the future of collaboration, I mean, you've talked about these knowledge workers, we had your head of product on the show. He kind of explained to us what a knowledge worker is, And we were sort of joking about it in the newsroom that, Hey, I didn't know this, but I am a knowledge worker. And so as you think about these people using these products, and maybe some of them being prosumers, I guess, what do you think the future of collaboration looks like three to four years from now with the advent of AI?

变革将持续发生。AI技术已展现出对各类知识形态的消化、整合、创作及重构能力，这已成为共识。

Okay. I think it's gonna keep changing. Right? It's obvious that what AI can do is consume, synthesize, create, rewrite knowledge of all forms. Right?

目前文本处理已很成熟，图像视频领域也日益普及——任何数字信息都能被AI转化利用。所谓知识工作者，其处理的数字信息通常与无需实体资源（如钢材、混凝土、木材等）的脑力劳动密切相关。

And we're used to it with text. We're seeing it increasingly with with images, with video. Any form of digital information can be used in that way. When we talk about knowledge workers, usually that digital information is related to some sort of a job where there are no resources involved. There's often very little, steel, concrete, wood, grass, whatever the resources are.

没错。而AI正以巨大的方式增强这种知识获取能力。因此，通过SaaS应用和浏览器利用AI实现更快速、更高效的协作，显然是主要途径——我认为这是当前大多数人访问应用程序的主要操作系统，无论是电子邮件、日历、CRM系统、项目管理系统、知识管理系统，任何你使用的系统，你很可能大部分时间都是通过网页浏览器访问的。

Right. And that knowledge is enhanced by AI in massive ways. So the ability to collaborate faster, more efficiently using AI through through SaaS apps and browsers is obviously a major way, the major operating system, I would argue, that most people get to their applications today, whether that's your email, your calendar, your CRM system, your project management system, your knowledge management system, whatever system you're using, you probably spend the majority of time accessing that system through a web browser.

是的。我想转向一个稍有不同的主题，即企业数据战争这个概念，这是我们《The Information》一直非常着迷的话题。我们报道过像Salesforce和Atlassian这样的公司如何使某些AI平台更难获取数据。我想听听你的看法，因为我们曾就此撰写报道并引用了已发布的内容。在我读到的报道中，Atlassian在一份解释限流措施的文件中表示，他们注意到API使用量异常增长，正在实施速率限制以保障对客户和合作伙伴的服务可靠性。

Right. Well, look, I want to pivot to talking about a slightly different topic, which is this idea of the corporate data wars, which is something that we've been so fascinated by here at The Information. You know, we've written about companies like Salesforce and like Atlassian making it harder for some AI platforms to access data. And I want to get your take on it because this was a story that we wrote and took pull from one of the stories that we published. The story I read, In a document explaining the throttling moves, Atlassian said it had noticed an unusual increase in API usage and was enforcing rate limits to maintain reliable services for both Atlassian customers and partners.

这份文件曾出现在你们官网上。这是我们报道过的新闻。所以，我想问你，虽然你已为此报道提供过声明，但你认为这对你们的业务构成风险吗？或者说我们该如何看待这个问题？

So this was a document that was on your website. This was a story that we wrote. And so, I mean, I just want to ask you, you provided a statement for the story, but I mean, do you see this as a risk to your business, or how should we think about that?

我本身不认为这对我们业务构成风险。我们始终以客户为中心，这是我们的高层理念。我们一直设有速率限制，有时这些限制虽公布但未强制执行。

I I don't see it as a risk to our business per se. We are always trying to be customer focused. We'd say that at a high level. We've always had rate limits. Sometimes they have been published, but not enforced and other things.

当遇到可能突破这些限制并损害所有客户集体利益的新技术时，我们会与合作伙伴共同确保限制合理。需要指出的是，尽管那篇文章有所暗示，但我们的限制比许多竞争对手宽松得多。其次，有些企业的限制实质上为零——这算不上真正的限制。

And you run into new technologies that can hurt those rate limits and hurt all of your customers collectively. We try to work with all our partners to make sure those limits are sensible. I will point out that our limits are more than quite a lot of our competitors despite what that article seems to imply. And secondly, other people's limits are basically zero. So if you have a limit that's basically zero, it's not really a limit.

而我们的限制非常宽裕且完全可用，多年来一直如此。我们始终是极度开放的公司，并坚信开放数据与数据交换对互联网、对所有SaaS应用都至关重要。我们提供REST API、命令行界面、MCP服务器等多种方式实现Elysium平台的数据进出，还与数百家商业伙伴保持定期数据往来。

Whereas our limits are very generous and perfectly usable and have been for many, many years. We have always been an extremely open company and continue to believe that open data and data exchange is incredibly important to the web, to all SaaS applications. We have REST APIs, command line interfaces, MCP servers. We have all sorts of ways to get information in and out of the Elysium platform. And we have hundreds of commercial partners that, we move data to and from on a regular basis.

确实。因此我认为这并不符合我们公司二十年来秉持的核心理念。

Right. So I don't think that's necessarily backed in in actually the reality of of the company philosophies for for two decades now.

跟我聊聊Robo吧。我是说，你们正在开发Robo。这是个非常棒的工具，而且它正与AI生态系统中一些热门的搜索产品竞争。你们是如何构建Robo，使其能与像Glean这样已经获得大量关注的公司竞争的？

And and talk to me about Robo. Mean, you're building Robo. It's it's a really neat tool, and it's it's going up against some of these other search products that have become popular in the AI ecosystem. How are you building Robo to be competitive against a company like Glean, for example, that has gotten a lot of traction?

我们可以讨论实际关注度有多少，但我们正在打造最优秀的

We can argue how much traction is actually there, but we're building the best on the

等等，你是说你不认为他们有真正的关注度吗？

Wait. Do do you don't think there's traction?

哦，关注度肯定是有的。好吧。只是具体有多少，这一点我需要澄清。

Oh, there's definitely traction. Okay. Just the amount of it is is is what I would contend with to be clear.

所以你认为Robo获得的关注度比Glean更高，可以这么说吗？

So you think you think Robo is getting more traction than than Glean, so to speak?

如果不是这样，而且差距还很大，那我会非常震惊。好吧。我们

I would be shocked if that is not the case by quite a long distance. Okay. We are

毕竟你是这个领域的核心人物，我不确定。好吧。

I mean, you're at the center of this. I don't know. Sure.

绝对如此。让我明确一点。我们已有数十万家企业正在使用我们的搜索引擎。没错。我们构建了自认为最优秀的企业搜索引擎，用于挖掘您公司所拥有的知识资产。

Absolutely. Let me be clear. We have hundreds of thousands of companies that already use our search engine. Right. We have built what we believe is the best enterprise search engine out there for the knowledge that your company has.

部分知识存在于Atlassian产品中，部分则以各种形式分散在第三方产品里。我们通过多种方式获取这些知识，并将其整合到Teamwork Graph中。多年来，我们持续优化这个知识图谱，使其逐年精进，让您能通过聊天代理（无论是我们的还是第三方的）、MCP服务器或REST API，或构建访问该数据的Forge应用，高效获取最佳知识及其关联——无论是文档、会议记录、团队信息等。我们在搜索领域的相关性、准确性等所有指标都堪称卓越，且正获得极佳的市场反响，这才是最有力的证明。

Some of it is with Atlassian products. Some of it is with third party products in all sorts of different manners. We have many ways to access that knowledge, which we put into the Teamwork Graph. We have, for many years, built a graph that's getting better and better every single year to enable you to access the best knowledge and the connectivity between those knowledge objects, documents, meetings, teams, whatever they are, in chat via agents, either our agents or other people's agents to get to it through an MCP server or a REST API or to build forge applications that access that data. Our relevance, accuracy, everything else that we have built in search, we would we would pause it as very good, and it's getting extremely good customer reaction, which is the ultimate proof.

是的。正如我们上次财报电话会议所述，我们的AI产品月活用户已突破230万，季度环比增长约50%，增势持续强劲。七八月份表现尤为亮眼，因此在知识管理领域，我们确实做得非常出色。这正是我们数千万用户每日醒来后——没错——就会使用的平台。没错。

Right. As we've said, we now have in our last earnings call, we passed 2,300,000 monthly active users of our AI products, and that's going up about 50% quarter on quarter and continues to grow really strongly. We we've seen a great July and a great August, so I think we're doing very well when it comes to knowledge. That's what our tens of millions of people wake up every day Right. And use Right.

通过Atlassian云平台获取知识。这些知识还持续链接到各类SaaS应用。我们能以极其精妙的方式处理这些知识并回馈用户，这完美契合AI的需求特性。

Atlassian cloud platform to get to their knowledge. And they have links to all sorts of other SaaS applications on a continual basis. We can process that knowledge and give it back to them in really, really fantastical ways, which is perfect for what AI needs.

明白了。我想请教关于AI编程工具的机遇。毕竟Bitbucket已在贵司产品矩阵中，同时大量开发者都在使用你们的工具。那么，Atlassian是否会进军AI编程领域？

Right. I wanna ask you about the opportunity that exists with AI coding tools. I mean, you have coding tools within your umbrella with Bitbucket. You've also got a lot of developers using all of your tools. And so I wondered, I mean, can we expect Atlassian to make a play into the AI coding space?

我们现有RoboDev，目前在全套基准测试中位居编程工具榜首。比如领先Claude Code约三四分。众多客户正使用RoboDev进行软件开发。我们在Jira和Bitbucket中部署了多种形式的RoboDev代理，既能制定编码计划，也能直接为客户编写代码。

So we have RoboDev, which is currently sitting on top of the full suite bench rankings for writing, code. It's about three or four points ahead of, Claude code, for example. Okay. And many, many customers using rovodev to write software. We have, RoboDev agents that run-in various forms inside of Jira and Bitbucket to do coding plans, to actually write code for customers.

了解了。

Got it.

拉取请求、为拉取请求添加安全性，以及各种不同的活动。

Pull requests, adding security to pull requests, all sorts of different activities.

没错。

Right.

不过，你也看到了，我们与许多现有工具建立了合作关系。比如我们的开发者使用Cursor，使用GitHub Copilot。在一个庞大的工程团队中，我们运行着多种不同的编程工具，它们各有优劣。我们也大量使用Cloud Code。

You also, though, partner like, if you look at our developers, what's more important is we partner with many existing tools. So our developers use Cursor. Our developers use GitHub Copilot. We have many different coding tools that we run-in a very large engineering team, and they all have different strengths and weaknesses. We use a fair bit of Cloud Code as well.

所以

So

我真正想问的是，我们是否能在该领域期待任何形式的收购或并购活动。

And and more more what I was getting at is if we could expect any kind of acquisitions or M and A in that space.

听着，这很有可能。软件团队是我们客户群体的重要组成部分。他们与服务团队、广泛协作团队和战略领导团队一样，是我们长期重点关注的群体之一。

Look, it's quite possible. Software teams are a significant part of our customer audience. Right? They're one of the teams alongside service teams, broad collaboration teams, and strategy leadership teams that we spend a lot of time thinking about. Right.

这部分客户占比很大，因此我们始终在该领域进行创新和开拓。

It's a good portion of our customer base. So we're always innovating and creating in that area for sure.

好的。迈克，最后一个问题。你知道，现在有种说法是AI要来吞噬SaaS公司，你处在一个有趣的交叉点，因为正如你刚才所说，你拥有大量开发者用户，他们开始使用各种AI编程工具。但你本质上是从传统SaaS时代成长起来的。所以我很好奇你如何看待AI可能吞噬或蚕食SaaS业务这个问题。

Right. Last question for you, Mike. So, you know, this narrative that AI is coming to eat SaaS companies, I mean, you kind of sit at an interesting intersection here because you have a lot of developers, like you just said, I mean, you know, you have a lot of these AI coding tools that your customer base is starting to use. But you kind of came up in sort of the old school SaaS era. And so I wondered how you think about that question that AI could be coming to eat SaaS, or in some cases, cannibalize the businesses.

说实话，我认为对大多数人来说这是个偷懒的观点。你必须理解原因。SaaS本质上是软件的交付机制。我们为客户所做的是让他们能够存储业务流程、工作流、各类知识，并与其他流程和知识建立连接，从而通过我们云平台上的弹性系统获得商业优势和协作效率。

I think it's a lazy take for most people, to be honest. And and you have to understand why. SaaS is a is a delivery mechanism for software. What we do on behalf of our customers is allow them to store business processes, workflows, knowledge of various types, connected to lots of other processes and knowledge, and hopefully gain some sort of business advantage, some collaborative speed through through the elastic system at work in our our cloud platform. Right.

AI正在为这个系统注入超强动力。AI技术已广泛普及，各种API、大语言模型——无论是开源的、闭源的还是基于API的——现在对所有供应商开放。这就要求每个供应商思考：如何继续完成现有使命？要知道，我们每天支撑着数百万个业务流程的运行。

AI is incredibly supercharging to that. AI is broadly available. These APIs are available, large language models, open source ones, closed source ones, API driven ones to every vendor out there. Now that will require every vendor to think, how can I continue to do the same jobs that I'm doing? You know, we power millions of business processes on a daily basis.

这些业务流程完全可以通过AI在各个环节获得增强：提速、提高准确性和质量。我们的责任是通过明智的研发投入，组建优秀的设计师和工程师团队来改进这些应用。AI本身并不意味着SaaS会消亡。事实上，我认为更可能相反，但最终结果还需时间验证。

Those business processes can probably be enhanced, sped up, made more accurate, higher quality with AI in various places. It's up to us to innovate, to spend our r and d dollars wisely, to put talented teams of designers and engineers on making those applications better. Right. There's nothing inherently to do with AI that should mean that SaaS ceases to exist or ceases stops existing. In fact, I think it is most likely to be the opposite, but we'll have to all wait and see how it plays out.

以开发者为例，同样流传着AI将摧毁软件开发者的论调，说未来开发者数量会锐减。但换个角度想：同样的开发者数量将创造出更多技术和软件。我个人倾向于后一种观点。

If you look at developers, for example, there's a long similar meme that that AI is gonna destroy software developers. There's gonna be far less software developers in the world. Right. Take the same thought pattern that said there's gonna be far more technology and software created in the world by the same number of developers. I I tend to think it's gonna be in the latter camp.

看看那些清醒的科技公司CEO的声明，他们正在招募越来越多的工程师。AI正帮助企业拓展业务边界，推动行业前进。我们需要时间观察事态发展，这绝不会是非此即彼的局面。

Right? If you look at most of the sensible statements from live CEOs, they are hiring more and more engineers. AI is allowing their businesses to do more, and that is empowering a drive forward. We're gonna have to wait and see how this all plays out. And it's not gonna be all of one or all of the other.

有些SaaS企业会陷入困境，有些则会蓬勃发展。我们正在AI领域大力投资，拥有顶尖团队引领模型应用落地。

There are gonna be some SaaS companies that struggle. There are gonna some SaaS companies that profit. We are heavily investing in AI. We have a huge team, world class leading model adoption. Right.

从Rovo到Rovo Dev，乃至收购像The Browser Company这样的企业，以及其间的一切，都是为了帮助我们的客户比以往更快、更好地完成他们二十年来一直依赖我们完成的工作。这将为客户创造巨大的经济价值。

And everything you can see from from rovo to rovo dev to even acquisitions like the browser company and everything in between Right. To help our customers continue to do those jobs that they have turned to us for two decades to do better, faster, and quicker than they ever have before. Right. Right. I think that's gonna deliver amazing economic value to our customers.

关于这一点，我想快速请教Atlassian如何应用AI的问题。过去几周有消息称公司裁减了客服人员，外界猜测这是否与AI有关。您能否说明这些裁员是否由AI导致？或者说，我们该如何理解这一情况？

And and on that note, very quickly, before I let you go, of how Atlassian is using AI. I mean, you know, we've seen news that the company has let go of customer service workers that has sort of trickled out over the past few weeks. And there's been a question around whether or not they are being let go because of AI or for some other reason. Can you comment on that at all? Is AI responsible for these people being let go, or what do we Again, think of

我们已明确表示裁员与AI无关。媒体总爱渲染这类标题，最近相关报道层出不穷。

we've been very clear they are not being let go because of AI. Right. Media loves to write that. It's been a whole series of headlines.

正因如此我才想直接向您求证。我的意思是...好吧，那么...

Well, and that that that's why I wanted to ask you that's why I wanted to ask you directly. I mean, you know? Okay. So so so Let

容我说明，'开放公司，拒绝空话'是我们二十多年来的核心价值观。关于这次裁员，我们已坦诚说明：这是基于多方面因素做出的艰难决定。

me just say, one of our core values is open company, no bullshit. That's a core value of the company has been for more than twenty years. We were very clear about that. That's a very difficult set of decisions. We are delivering our customer service for many, many reasons.

漏洞修复速度加快，系统可靠性已提升至99.99%以上，客户咨询量大幅减少。我们必须根据未来需求调整人员规模，导致客服部门出现人员冗余，迫使我们做出痛苦的人事调整。

Bug fixes are faster. Our reliability has gone up to to well over sort of four nines over the last few years and various other things that mean we get far less customers questions per customer. We have to forecast for where we're going to be. So that's led to the unfortunate situation that we have far more people in customer support and customer service, great people, than we needed. And hence we had to make some hard decisions.

我们很早前就放缓了招聘节奏。虽然可以说AI间接加速了软件开发，进而提升系统可靠性，但这种因果链条很难精确归因。

We slowed down hiring quite a long time ago. Various other things happened that we were very clear on why and what the reasons were. Okay. Now you might argue that AI led to software development being slightly faster, which has led to the bugs being fixed faster, which has led to higher reliability. It becomes a little hard to case ingredient.

但要说这些是因为AI导致的，那绝非事实。

But to say that they are because of AI is is is just not true.

那么你如何看待像马克·贝尼奥夫这样的人公开表示，比如AI已经替代了4000人的工作？在像你们这样的企业背景下，我们该如何理解这种说法？

And so then how do you think about something like someone like Mark Benioff coming out and saying, well, we have been able to let go of or AI has now done the work of 4,000 people, for example. I mean, do you well, what do we make of that for in the context of of a business like yours?

听着，马克喜欢发表些夸张的声明。我让他为自己的言论辩护。毫无疑问，AI将以多种方式提升企业生产力和效率。可以把它看作更优质的电力，是企业的基础要素。

Look, Mark likes to make some some grandiose statements. I'll let him defend his own statements. I think there's no doubt that AI is going to improve the productivity and speed that business has done in lots of different ways. Think of it like better electricity. It's a fundamental ingredient to a business.

它为工厂供电，为数据中心供电，为电脑、照明、空调等一切设备提供动力。AI将有点像电力，因此很难说它具体完成了X或Y。

It powers your factories. It powers your data centers. It powers your computers, your lights, your air conditioning, everything else. AI is gonna sort of be a little bit like electricity. So it's gonna be very hard to say it has done x or it has done y.

无疑，它应使企业获得更好成果、更高产出质量、更多产量、更高效产出，某种程度上还能创造更多技术产出，进而成为其他一切的输入源。我认为这很可能是事实。我会观望其他企业的财务数据，看其员工数量是否大幅减少。可以告诉你，我们尚未看到任何影响。作为拥有数十万客户且用户多为知识型员工的企业——

There's no doubt it should lead to businesses getting better results, higher quality of output, more quantity of output, more efficient output, to to some extent better volume of of technology created, which then becomes another input source to everything else. So I think that's likely to be true. I'll wait to see other people's financial results if their employee counts go down massively. I can tell you we haven't seen any impact. As a company that has hundreds of thousands of customers and their users are knowledge workers.

正如我所说，我们每天有数千万人在工作。如果有影响，我们会在客户扩展率中看到迹象。比如一个平均100人的客户，一年后通常增长到110或115人，两年后到130人。我们拥有如此庞大的数据集，但并未发现多年来历史扩展率有任何与AI导致的所谓大撤退相关的变化。

And we have, as I said, tens of millions of people at job every single day. We would see some impact in our expansion rates. So a customer with a 100 people on average ends up as, you know, a 110 or a 115 after one year and a 130 after two years. We have such a large data set. What we are not seeing is any change in those historical expansion rates that we've had for many, many years that is anything to do with some sort of great, you know, pullback due to AI.

所以我理解你的意思是，人们可以随意发表观点。但你认为很难证明这些裁员或劳动力变化直接归因于AI，这种关联性很难确立，这是我的理解。

So what I hear you saying is that it's people can come out and say what they want. In your view, it's very hard to correlate that, hey, this headcount reduction or this change in workforce, that was directly attributed to AI. It's very hard to draw that correlation, what I'm hearing you say.

我认为在某些领域会有人得出这种关联性。就个人而言，我认为在整个行业内建立这种关联性很困难。但毫无疑问，这将使企业能够以更少的资源做更多事情。至于他们如何利用这一点，则取决于他们自己。短期内，这通常会导致利润或利润率提高。

I think there are people who are gonna draw that correlation in some areas. I think it's hard to have an industry wide correlation for that personally. But there's no doubt it will lead to businesses being able to do more with less. Now what they choose to do with that is up to them. Often what that leads to is higher profits or margin for a short period of time.

没错。当人们追逐这些利润时，竞争会更加激烈。然后就会催生新技术、新产品、新服务来填补空白，或是为现有产品和服务增添更好的功能。对吧。

Right. More competition when people go and chase those margins. Right. And then creating new technologies, new products, new services to to fill in that or adding better features to their existing products and services. Right.

所以我认为随着这项伟大技术的普及，你将看到所有这些结果。从Atlassian的角度来看，我们正在以前所未有的速度招聘工程师，试图构建能带来客户价值并为Atlassian创造价值的优秀AI技术。

So I I think you're gonna see all of these outcomes coming as this great technology flows through. Right? Atlassian's point of view, we are we are hiring engineers as fast as we ever have. To try to build great AI technologies that will deliver customer value and accrue value to Atlassian.

这让我回到了我们开始对话时的那个问题：AI时代的协作未来。我的意思是，在这个有AI代理相互对话的未来，比如它们在不同平台和工具之间交流，企业员工或你所说的知识工作者的角色是什么？如果代理完成所有对话，那么人在这种协作中的角色又是什么？

And that kind of brings me full circle to the question that we sort of started this conversation with, which is the future of collaboration in the AI era. I mean, in this future where we have AI agents, for example, doing the talking to each other across different platforms and across different tools, what is the role of the corporate employee, or as you call it, the knowledge worker? What is the role of the people in that collaboration then if the agents are doing all the talking?

我认为我们误解了代理的作用。人们在定义待完成工作方面将发挥巨大作用。我们现在有很多软件程序。只需将AI重命名为软件程序或代理。代理非常强大。

I I think we mistake what agents are going to do. The the people are gonna have an incredible role in defining what the work is to be done. We have a lot of software programs today. Just rename AI to a software program, an agent. Agents are very powerful.

我并没有贬低AI的力量。我们正代表客户大力投资于此。但我认为他们不会放任代理自主运行并完成大量工作。你仍然会有项目，仍然会有团队。

I'm not diminishing the power of what AI is. We're investing very heavily on behalf of of our customers in doing so. What I don't see is them just letting agents run wild autonomously and going to do a bunch of work. You are still gonna have projects. You're still gonna have teams of people.

人们需要与这项技术结合来实现某些目标。他们能否以更高质量或更多数量实现这些目标？当然可以，成本绝对更低。但如果你只把它们视为软件程序，就像手机，我们并不认为它们能自主运行取代整个人类工种，但我们可以随时随地使用它们来填补时间，提高各种工作的效率。对吧。

They're gonna need to come together with this technology to achieve some sort of outcome. Can they do that outcome with better quality or more quantity or Yes, a lower absolutely. But if you just treat them as software programs, you know, mobile phones, we didn't think replaced whole categories of people by autonomously going to do things, but we can use them on the go to fill in time and and achieve a better efficiency for all sorts of different work. Right? Right.

飞机上的WiFi，我们能够越来越频繁地使用。可能有很多很多例子可以说明这一点。我不认为代理与代理之间的交流会神奇地提出所有这些解决方案。他们仍然需要人。那些代理仍然会产生结果。

Wi Fi on planes, we can do more and more times. Like there's probably many, many examples of how this works. I don't think the agents talking to the agents will sort of magically come up with all of these solutions. They're still gonna need people. Those agents still have a result.

他们会输出一份文档、一个流程、一个对象。人类需要查看那个对象，也许再把它发送给另一个代理。我们今天在做什么呢？我收到通过计算机系统发来的邮件。我可能会思考一会儿那封邮件，然后转到另一个计算机系统。

They will output a document, a process, an object. A human will have to look at that object, maybe send it off to another agent. Well, what do we do today? I get email that comes through a computerized system. I take that email and I may think about it for a while and I go to another computer system.

我可能会查找一些客户详情，做一些其他零碎的事情。我会给客户打电话。所有这些工作仍然需要完成。公司仍然会有团队。

I may look up some customer details. I may do some other bits and pieces. I call the customer. There is gonna be all of this work that still needs to happen. Companies are still gonna have teams.

他们仍然会有项目。他们仍然会有目标。他们需要实现这些目标。这只是一项伟大的技术，能帮助他们实现更多的目标。对吧。

They're still gonna have projects. They're still gonna have goals. They're gonna need to achieve them. This is just a great piece of technology to allow them to achieve more of their goals. And Right.

我非常相信人类的创造力。对吧？那些公司会想出更多他们想要构建的东西。他们会想出更多取悦客户的方式，并且能够更快地实现这些。

I'm a big believer in human creativity. Right? Right. Those companies are gonna come up with more things that they wanna build. They're gonna come up with more ways that they wanna delight their customers, and they're gonna be able to do that faster.

这将导致竞争的加速。所以，如果你在采用这些技术方面行动不够快，未来五到十年可能会相当痛苦。但对于那些行动迅速的人来说，我认为他们将创造出的技术、产品和服务，我们都会成为受益者。

It's gonna lead to a speeding of competition. So if you're not moving fast enough to adopt these technologies, it's probably gonna be a pretty painful world over the next five to ten years. But for those that do, I think they will deliver just we'll all be the beneficiaries of the technologies and products and services that get created.

太棒了。迈克，非常感谢你参加TI TV的节目。很高兴你能来。这位是Atlassian的首席执行官迈克·坎农-布鲁克斯，我们期待在你开发更多产品时能再次邀请你。感谢你的到来。

Great. Well, Mike, thank you so much for joining us on TI TV. It is great to have you. That is Mike Cannon Brooks, the CEO of Atlassian, and we look forward to having you on more as you look to develop more of these products to come. Thanks for being here.

谢谢邀请我参加，老兄。

Thanks for having me on, man.

刚才发言的是Atlassian公司的首席执行官迈克·坎农·布鲁克斯。好的，今天的节目就到这里。提醒大家，我们每周一至周五太平洋时间上午10点、东部时间下午1点在此直播。我要感谢亚马逊网络服务对本节目的冠名赞助。

That was Mike Cannon Brooks, the CEO of Atlassian. Okay. Well, that does it for today's show. A reminder that we are live on this stream Monday through Friday at 10AM Pacific, 1PM Eastern. I want to thank Amazon Web Services, who is our presenting sponsor for this production.

同时也感谢各位的收看，我们非常珍视每一位观众。我已经开始期待明天的节目了。那么，我们下次再见，拜拜。

And I want to thank you for tuning in. We really do appreciate your viewership. I'm already excited for our next show tomorrow. And so until then, bye bye for now.