健康科技的未来：Oura CEO专访，Brex CEO谈金融科技增长，AI助力科学发现 | 2025年10月14日

欢迎大家来到Information的TITV。我是Akash Basricha。今天是10月14日星期二。今天我们为大家准备了科技领域的全面报道。我们将与Aura的首席执行官讨论健康科技。

Welcome everyone to the Information's TITV. My name is Akash Basricha. It is Tuesday, October 14. We have got another full tour of the tech sector today for you folks. We are talking about health tech with the CEO of Aura.

我们还将与Brex的首席执行官讨论金融科技。接着我们会讨论租赁NVIDIA芯片的竞争。最后，我们为大家安排了一场关于AI如何实际影响物理世界的精彩讨论。但在开始之前，我想重点介绍一下我们昨晚发布的一则重大新闻。Information率先报道，OpenAI正在与主要由软银持有的芯片公司ARM合作开发CPU。

We're also talking fintech with the CEO of Brex. We're then going to talk about the race to rent out NVIDIA chips. And finally, we have a great discussion planned for you about how AI can actually have an impact on the physical world. But before we get going, I want to highlight for you a big story that we published last night. The information is first to report that OpenAI is working with ARM, the chip company owned in large part by SoftBank, to develop CPUs.

提醒一下，CPU与席卷全球的AI芯片GPU不同。但关键在于，所有AI芯片都需要CPU才能工作。因此这则新闻很重要。这一消息紧随OpenAI宣布与博通合作共同开发自家芯片之后。所以这是一条大新闻。

A reminder that CPUs are different from GPUs, the AI chips that have taken the world by storm. But the key thing here is that all AI chips need CPUs to work. And so that is why this story is important. The news came on the heels of OpenAI's announcement that it is working with Broadcom to jointly jointly develop their own chip. And so this is a big story.

这则新闻推动了ARM股价的变动——准确地说，是昨晚盘后交易中推动了ARM股价变动。今早股价继续波动。我们会在节目说明中附上链接。说到这里，让我们请出第一位嘉宾。Aura今天发布了重大新闻，宣布以110亿美元的估值融资了9亿美元。

It is moving arm shares after it moved arm shares after hours last night, I should say. They continue to move this morning. We will link it in the show notes. And with that, let's get to our first guest. Aura has made big news today, announcing it has raised $900,000,000 at an $11,000,000,000 valuation.

本轮投资由富达管理与研究公司领投，Iconic参与投资。公司今年有望突破10亿美元的年销售额，我想请公司首席执行官Tom Hale上台，分享他计划在下一阶段将Aura带向何方。Tom，欢迎来到TI TV。很高兴再次见到你。

The investment was led by Fidelity Management and Research Company with participation from Iconic. The company is on track to pass $1,000,000,000 in annual sales this year, and I want to bring on Tom Hale, the CEO of the company, to share more about where he plans to take Aura in this next phase of growth. Tom, welcome to TI TV. It's great to see you again.

很高兴见到你，Akash。谢谢邀请。

Great to see you, Akash. Thanks for having me.

对Aura来说这是重要的一天。我想从公司上次的情况开始——几个月前你和主编Jessica Lesten一起上节目时，你向她提到公司是盈利的，但你们仍然筹集了这么多资金，我的问题是，既然盈利了，为什么还需要融资？

So a big day for Aura. The place I want to start with the company is last time you were on the show a couple months ago with editor in chief Jessica Lesten, you mentioned to her that the company's profitable, and yet you've still raised all this money, and my question is, why do you need to raise the money if you're profitable?

嗯，有消息。大家都知道人工智能非常昂贵。我认为人工智能与医疗保健的未来密不可分。因此，我们希望确保有能力投资人工智能，特别是在我们考虑吸纳人才、进行投资，并可能最终开发自己的模型时。这无疑是我们首先要投资的方向。

Well, news. Everyone knows that AI is incredibly expensive. And I think AI and the future of healthcare are intertwined. So I think we want to make sure that we have that ability to invest in AI, particularly as we think about acquiring talent and making investments and maybe ultimately developing our own models. So this is certainly the first thing that we wanna invest in.

第二点当然是，我们还处于这项业务的起步阶段。Aura发展迅猛。我们的业务一直在翻倍增长，年复一年，即将达到十亿美元的销售额。而现实是，我们才刚刚开始，因为我们几乎只在美国市场。

The second thing of course is that, you know, we're still at the very beginning of this business. Aura has been on a tear. We've been doubling the business, you know, year over year, we're about to reach a billion dollars in sales. And the reality is, is we're just getting started because we're barely just in The U. S.

如今。因此，我们在全球范围内还有很大的增长和扩张空间。我认为这需要我们投入相当多的资金，以确保能够覆盖全球。最后一点当然是医疗保健。我们的愿景一直是关注整体健康的方方面面。

Today. So we've got a lot of headroom and expansion around the globe. And I think that's gonna take us, I think a fair amount of investment to make sure that we can cover the planet. And the last piece of course is healthcare. Our vision always has been to be aware of all that is across the entire surface of holistic health.

医疗保健正在转型。我们没有足够的医生，也没有足够的预防医学。我们所做的预测，一些基于科学的实践仍然是数据驱动的。我们拥有佩戴Aura设备人群的最新实时数据。

And healthcare is in transformation right now. We don't have enough doctors. We don't have enough preventative medicine. The prediction that we do, some of the practices that are grounded in science are still, they're data data. We have up to date current data on the people who are wearing the Aurang.

我们希望利用这些数据，将其应用于健康领域。我认为这将使公司变得与现在截然不同。因此，我们需要投资于人才、人员、流程、系统和科学，以真正推动Aura在医疗保健旅程中迈出下一步。

Want to take that data and make it present in health. And I think that's a very different kind of company than what we've been today. And so we need to invest in the talent, in the people, in the process, in the systems and the science to really make a healthcare journey for Aura take its next stride.

你们目前在使用哪些AI模型？用它们来做什么？

Whose AI models are you using right now? What are you using them for?

我们几乎使用所有可用的模型。我认为我们的应用范围很广。首先当然是系统使用算法进行的预测，这些算法基本上是用来告诉你我们从你身体采集的信号中得出的数值。我们在这方面已经有十年的经验，可能是在信号处理这类算法上全球最擅长的。

We use almost every model that's available. And I think we have a range of things that we do. It starts of course with the predictions that the system makes using algorithms that are basically there to tell you what are the values that we're doing from the signals that we're taking off of your body. And we've been doing that for ten years. We're probably maybe the best in the world at these kinds of algorithms for in signal processing.

所以我们讨论的是开源模型和闭源模型的结合。我的意思是，你知道，像那些大型模型，

And so we're talking a combination of open source models, closed source models. I mean, you know, the Well, big models like

对于这些模型，它们几乎完全由我们开发，尽管我们使用了所有工具。第二点是这种想法，你可能认为是口袋里的医生，对吧？你已经有了超级计算机，有了可穿戴设备。有一些智能正在审视这些数据。它知道你与它分享的许多关于你的信息。

for those models, those are almost entirely developed by us, although we're using all the tools. The second place is this kind of idea of what you might think of as the doctor in your pocket, right? You've already got a super computer, you've got a wearable device. There's some intelligence that's looking over that data. Knows lots of things about you that you've shared with it.

你的背景、健康史、标签、生物特征，并理解这些。这实际上是AI的另一种形式。而这主要是我们开发的模型。然后是洞察的解释和呈现。所以基本上你可能想询问你的数据并说，发生了什么？

Your context, your health history, your tags, your biometrics, and making sense of that. That is actually really another form of this AI. And that's largely models we develop. Then there's interpretation and presentation of insights. So basically you might wanna interrogate your data and say, what's going on?

而一个LLM类型的应用程序会回应你说，嗯，这就是正在发生的事情。或者你可能会寻求建议。而那些LLM，我认为，你知道，是你所期望的经典LLM以及可用的开源和闭源模型。所以在这三个层面中，你都有一个AI的基础。现在，我认为这一切的发展方向是，你考虑像Oura Ring这样的设备所提供的数据质量，意思是它是一致的、准确的、连续的、是夜间测量的。

And an LLM type application will respond to you and say, well, this is what's going on. Or you might ask for advice. And those LLMs are, I think, know, what you would expect to be the classic LLMs and the open source and closed source models that are available. And so across all three of those surfaces, you've got a substrate of AI. Now, where I think this all goes is you think about the quality of the data that something like the Oura Ring has, meaning it's consistent, it's accurate, it's continuous, it's measured overnight.

用这些数据喂养一个，称之为大型生理学模型，将能够创建一个有效地针对你并围绕你定向的AI，因为你的健康和我的健康并不相同。那么你如何创建一个智能来支持全球健康条件、人群、民族、种族的巨大多样性？我认为这是我们希望投资应对的挑战。

Feeding a, call it a large physiology model with that data is gonna be able to create an AI that is effectively targeted and oriented around you because your health and my health are not the same. And so how do you create an intelligence that supports the vast diversity of both health conditions, people, ethnicities, races around the planet? I think that's the challenge that we wanna invest behind.

不过我想问你关于开发这些AI工具的盈利能力，因为如你所提到的，公司现在盈利了，但我们已经看到很多这些应用层公司，尤其是那些想要构建自己模型或利用现有模型的公司。我的意思是，这很昂贵。这是毫无疑问的。所以你担心这会降低你的利润率吗，当你在AI领域进行开发时？你如何看待这一点？

I do wanna ask you though about the profitability of building out these AI tools, because as you mentioned, the company is profitable now, but we've seen with a lot of these application layer companies, certainly those companies who are looking to build their own models or utilizing models that exist. I mean, it's expensive. There's no two ways about it. And so are you concerned at all that this is going to lower your profitability margin as you look to build out in AI at all? How do you think about that?

嗯，我有一个非常有趣的观点。我认为部分原因与隐私有关。人们希望对他们的健康数据有非常严格的控制。我认为这是我们作为首要原则的事情，确保我们不分享你的数据是Aura从成立以来就坚信的理念。现在，关键是在一个AI解释甚至通过LLM与年轻人沟通你的健康信息的世界里，你如何做到这一点？

Well, I have a really interesting kind of thesis on this. And I think part of it goes to privacy. People want to have very strict controls over their health data. And I think that's something that we take as a first principle and making sure that we don't share your data is like something that Aura believes in from our foundation. Now, the key is, is that how do you do that in a world where you have AI interpreting or even communicating to youth via an LLM about your health?

我们的愿景是将人工智能部署在边缘，也就是在你面前的设备上。如果我们能够尽可能多地将处理和人工智能能力放在离你更近、由你控制、经过加密并可能通过生物识别锁定的设备上，这就是我们所看到的愿景。这样做的一个附带效应是，你还能利用所有这些设备未被充分利用的计算能力，它们的处理能力正在加速提升。因此，我们愿景的一部分是拥有私密的边缘人工智能，最终还能降低成本。这就是我们对你提到的成本问题的思考方式。

Our vision for this is to put that AI at the edge, meaning that it is on the device that's in front of you. And if to the degree that we can put as much of that processing, as much of that AI on a device that's close to you and under your control, and it's encrypted and may be locked by your biometrics, that's the vision that we see. Now, the side note effect of that is, is that if you do that, you're also taking advantage of the latent processing power of all these devices, which are underutilized in terms of their computing capability and are accelerating in terms of their power. So part of our vision is to have AI at the edge that is private, also ultimately lower cost. And so that's how we think about that cost problem that you highlight.

我理解你的意思，这是一个艰难的挑战。你知道，很多可穿戴设备公司都在思考这个问题。我想问的另一个问题是，今天早上有人问到你关于是否进行IPO的计划，我的意思是，你如何看待将这家公司打造成一家独立企业，而不是——说实话——可能非常适合融入一家更大的科技公司。当你谈到边缘计算时，你知道，将大型语言模型部署在设备本身的能力，这是我们听说像苹果这样的公司甚至也在考虑的事情。

And I hear you, it's a tough challenge. You know, it's something that a lot of these wearables companies are thinking about. You know, the other question I wanted to ask is, and you were asked this morning about your, your plans of whether or not you IPO or not, I mean, how do you think about building this as a standalone company versus something that, you know, honestly, could fit very nicely into a bigger tech company. And as you talk about edge computing, you know, the ability to put LLMs on the devices themselves, this is something that, you know, we've heard companies like Apple is even considering.

我认为事情将会这样发展。我绝对这么认为。特别是对于这类应用，我认为实际上在边缘拥有隐私将变得非常必要。所以我认为这是可行的。我认为这就是我们正在走的道路。

I think this is the way it's going to go. I think absolutely. And particularly for these kinds of applications, I think actually it's gonna be really mandatory to have that privacy at the edge. So I think that is viable. I think that's the path that we're on.

而且我认为这是一个——

And I think that's a-

你认为你能独自完成吗？你能独自做到吗？

And you think you can do it alone? You can do it alone?

我认为，阿卡什，科技领域没有人是单打独斗的。这一点非常清楚。但如果你真正问的是我们作为一家独立公司，你知道，我的意思是，我们已经融了资，并且基于我们的财务状况，我们是以相当不错的估值融资的。我们的财务数据支持这个估值。这并不是一个离谱的估值。

I think, I don't think anybody in tech does anything alone, Akash. This is really clear about that. But if you're asking really about us as a standalone company, you know, I mean, we've raised money and we've raised money at a pretty good valuation based on the back of our financials. Our financials support that valuation. This is not an outrageous valuation.

这不是某种预测价值。这是基于我们去年实现的收入以及我们将要实现的收入。所以这相对来说是令人安心的。然而，这里有一个看涨期权。这个看涨期权是关于人工智能和医疗保健的未来，我认为这个领域正在发生变革，其速度可能比任何人预测的都要快。而且我认为这似乎是不可避免的。

This is not some projected value. This is based on the dollars that we delivered last year and the dollars that we're gonna deliver, So that's relatively comfortable. However, there's a call option here. And that call option is on that future of AI and healthcare, which I think is being transformed, is happening faster than maybe anyone predicted. And I think that it seems inevitable.

所以这里有一个看涨期权，我认为这实际上足以支持像我们这样规模的公司从100亿美元估值增长到200亿、300亿，甚至4050亿美元估值。因为医疗保健的未来以及健康管理和交付的方式正在转变，人们成为自己健康的主导者，掌控自己的健康旅程。我认为这种变革是世代性的转变。因此，我不认为我们只是一家小型的可穿戴设备公司，而是将我们视为一个平台，一个您健康的操作系统。想想这能在市场上创造多大的价值。

So there's call option on that, which I think is actually honestly enough to support a company of our size increasing to, you know, from a $10,000,000,000 valuation to a 20, a 30, a $4,050,000,000,000 valuation because the future of healthcare and the way health is managed and delivered where people are the agents of their own health they are in charge of their own health journey. That change is a transformation I think is a generational transformation. So I don't think of us as like, oh, well, we're a little wearable company. I think of us as like a platform, an operating system for your health. Think about the value that that could create in the market.

还有两个快速问题想问您。我想了解一下设备本身的情况。您今天也推出了一些新设计。价格点仍然在几百美元左右，对吧？我看到那些新设计了。

Two more quick questions for you. I wanted to ask about the devices themselves. You introduced some new designs too today. The price point is still around a couple $100, right? I see the new designs there.

你们通过销售设备本身能赚多少钱？

How much money do you make on selling the device itself?

令人惊叹的是，我们的产品能产生贡献利润。所以在第一天，从单位经济角度来看我们就已经盈利了。这一点非常强大。其强大之处在于Aura还有一个会员模式，这种会员模式是一种持续的收入流。当我们销售戒指时，我们就能收回获取成本和制造成本，并实现一些利润。

So what's amazing is that we make both a contribution margin when our we products. So on day one, we're profitable on a unit economics basis. That's really powerful. And the reason why it's powerful is because Aura also has a membership model and that membership model is sort of an ongoing revenue stream. When we sell a ring, we're making back the cost of acquisition and the cost of manufacturing and a little bit of profit.

然后这就成为了订阅业务的入场券。

And then that is a ticket to a subscription business.

没错。

Right.

那么我们的利润率是多少？这个贡献利润具体是多少？您的意思是设备上的贡献利润率是多少？

So our margin How much is that contribution? What do mean? How much is the contribution margin on the devices?

哦，这个利润率相当高。我是说，你知道这类硬件利润率的情况吧？硬件利润率如果表现良好是40%。我们的利润率更高。

Oh, it's quite high. Mean, you know that sort of hardware margins, right? Hardware margins, if they're doing great are 40%. We're north.

高于40%，好的。

North of 40%, okay.

顺便说一下，这实际上是因为我们是一家规模化的优秀硬件公司，能够实现这样的利润率。但这相当于为零客户获取成本的订阅业务打开了大门，从而进一步提升了我们的利润率。所以当你查看Aura的财务状况时，我们看起来不像一家硬件公司，而更像是一家带有硬件组件的软件公司。

And by the way, like this is a function of like being a really good scaled hardware company delivering that. But then that is a zero CAC ticket to a subscription business. And that elevates our margins even higher. So when you look at Aura's financials, we don't look like a hardware company. We look much more like a software company with a hardware component.

我认为这是一个关键点，再次回到独立公司的估值问题。这是让Aura能够处于估值金字塔顶端的关键因素之一。

I think that's a key, again, going back to the valuation of a standalone company. That's one of the keys that allows Aura to be like, you know, at the top of the pile thumbstone valuation.

没错。我应该提醒大家，这是你们连续第二年融资了。我不知道你们是否连续三年都在融资，但我知道去年你们也完成了一轮大额融资。在你离开之前，我想问的最后一个问题是：我知道你们与多家拥有FDA批准设备的公司建立了合作伙伴关系。我们也看到像Eight Sleep这样的公司宣布未来将为其产品线寻求FDA批准。

Right. And I should remind people, mean, this is the second year in a row you're raising money. I don't know if you You may have raised money three years in a row, don't know, but I know you raised another big round last year as well. The last thing I want to ask you before you go is I know you have a number of partnerships with other companies that have devices that are FDA approved. And we've seen companies like Eight Sleep, for example, coming out and declaring we are going to seek FDA approval for our line of products in the future.

我们能否期待Aura在未来一两年内为其任何产品寻求FDA批准？

Can we expect Aura to seek FDA approval for any of its products in the coming year or two?

是的，我们的生育窗口功能已经获得了一类医疗器械认证。基本上这是一个软件功能，用于预测女性排卵时间。对于正在尝试怀孕的夫妇来说，这个功能是我们经过FDA审批的。我们还有其他产品正在筹备中。我认为这是我们投资于越来越像一家医疗保健公司的组成部分。

Yes, we already have a class one medical device in our fertile window. Basically that's the feature, the software feature that predicts when a woman is ovulating. So if you're a couple who are trying to conceive, that feature is one that we've worked through the FDA on. And we have others in the pipeline. I think that's part and parcel of kind of our investment towards becoming more and more like a healthcare company.

获得FDA批准不仅对安全性和有效性至关重要，同时也是技术准确性、科学验证性及临床可用性的严肃性和信心的标志。令我惊讶的是，有如此多的人将Aura纳入他们的健康管理方案。我们的客户中有10%到11%是各种形式的临床医生。因此，我们已经以多种方式进入医疗保健领域，并且随着与FDA合作的深入，我们预计将进一步扩展这一领域。

Having FDA approvals is both important for safe and effectiveness, but it's also a mark of seriousness and confidence that the technology is accurate and that it's validated by science and that it can be used by clinicians. It's stunning to me the number of people who use Aura as part of their health regime. Ten percent, 11% of our customers are clinicians of one form or another. So we're already in the healthcare space being used in a variety of ways, and we expect to actually extend that further as we work deeper and deeper with the FDA.

太好了。汤姆，非常感谢你再次来到节目。每次对话都很精彩。恭喜你们获得新一轮融资，也很期待看到今天发布的新设计。这里是TI TV，Aura的首席执行官汤姆·黑尔。

Great. Tom, well thank you so much for coming on the show again. It's always a great conversation. Congrats on the new funding ground excited to see the new designs as well that are being released today. That is Tom Hale, the CEO of Aura here on TI TV.

好的。本月早些时候，Brex的首席执行官向员工发送了一封邮件，大胆宣布转型期已经结束，公司将回归无拘无束的雄心壮志。我想请来撰写这份备忘录的首席执行官佩德罗·弗朗西斯基，告诉我们他如何将思维转向Brex的这一新增长阶段。佩德罗，欢迎来到节目。很高兴你能来。

Okay. Earlier this month, the CEO of Brex sent an email out to his employees boldly declaring that the turnaround is over and the company will now return to unconstrained ambition. I wanna bring on Pedro Franceschi, the CEO who wrote that memo, to tell us about how he's shifting his thinking to this new phase of growth for Brex. Pedro, welcome to the show. It's great to have you.

谢谢邀请。很高兴来到这里。

Thanks for having me. Excited to be here.

那么我们来谈谈转型期结束和这段无拘无束的雄心时期。我想把讨论分成两部分：一部分是转型，一部分是无拘无束的雄心。告诉我们你具体指的是什么样的转型。

So let's talk about the turnaround being over in this period of unconstrained ambition. I kinda wanna segment the discussion to two halves. One is the turnaround, one is the unconstrained ambition. Tell us about what exactly the turnaround is that you're referring to.

好的。也许让我先从我们现在的处境开始。如果你回顾过去二十个月的业务，我们彻底重建了公司的运营方式。所以，当你考虑我们构建产品的方式、市场进入策略、运营模式、领导团队，以及公司领导层的 talent density（人才密度）时，它们与二十个月前截然不同。我写了几份备忘录谈到这一点，关于让领导者在各个层级运作，改变我们构建产品的方式。

Yeah. So maybe let me just start with where we are today. So if you just look at the business over the past, you know, twenty months, we rebuild entirely how we operate as a company. So when you think about, you know, the way we build products, the way we build our go to market, the way we operate, the leadership team, and just the sort of talent density on leadership levels in the company, they are very different from where they were twenty months ago. And I wrote a few memos about this on, you know, having leaders that operate at all levels, changing the way we build a product.

而且，作为创始人和首席执行官，我更加深入地参与产品路线图和我们发布产品的质量。结果，在八月份，Brex实现了近50%的同比增长，并且自成立以来首次实现运营现金流为正。

And, you know, me as a founder and CEO being much more involved in a product roadmap and the quality of what we ship. And as a result, you know, in August, Brexit grew, you know, just shy of 50% year on year and was operating cash flow positive for the first time since his inception.

没错。

Right.

所以这与二十个月前的情况截然不同。而且，你知道，我们对此感到非常自豪。

So it's a very different reality from where it were twenty months ago. And, you know, we're very proud of that.

我很好奇你用了‘不受约束的雄心’这个词，因为你提到的一点是公司历史上首次实现了正向运营现金流。我的理解是，我们现在正重新聚焦于增长。而不惜一切代价的增长有时可能会带来，你知道，有风险的结果，就像我们在硅谷许多公司看到的那样。所以我的问题是，你计划如何激发这一新的增长阶段，同时实现盈利性增长？我的意思是，你知道，这正是随之而来的约束。绝对如此。

And I'm curious that you use the word unconstrained ambition because one of the things that you talked about is the company has been operating cash flow positive for the first time in its history. And my read on it was, we are returning to focusing on growth now. And growth at all costs sometimes can yield, you know, dicey results, you know, as we've seen with many different companies in Silicon Valley. And so my question for you is, how do you plan to spark this new phase of growth, I guess, while growing profitably? I mean, you know, that is the constraint that comes with this Absolutely.

所以当我们说‘不受约束的雄心’时，我们真正的意思是，如果我们只看当前所处的市场，我觉得大家未必充分认识到这一点。美国的企业卡支出达到2万亿美元。而Brex目前只占其中的1%多一点。如果我们把所有新玩家加起来，可能也只占市场的2.5%。所以这是一个由老牌企业深度主导的市场。

So so so when we say unconstrained ambition, really what we mean by that is if we just look at the market that we're in today, and I think folks don't necessarily appreciate that. There's $2,000,000,000,000 of corporate cards spent in The US. And Brexit today is like, you know, a little bit over 1% of that. And if we look at all the new players combined, we're maybe 2.5% of the market. So this is a market that's very deeply dominated by incumbents.

所以你有这些存在了几十年的银行，它们以非常糟糕的方式服务客户，没有软件，没有真正改善他们运营业务的方式。而当我们看客户如何使用Brex运营时，你会看到，你知道，像ARM、Cursor、Intel、Anthropic、Pounds这样的公司，对吧？它们都因为与Brex合作而在实质上改变了业务运营方式。这才是我们真正在谈论的。我们如何将这种精神和我们为市场上这微小部分——这1%的市场——运作如此良好的产品，带入那98%由老牌企业主导的市场。

So you have banks that have been around for decades serving customers in a very poor way with no software, with no way of actually improving the way they run their business. And when we look at the way customers run their operation with Brex and you see, you know, companies like ARM, Cursor, Intel, Anthropic, Pounds here, right? All sort of materially changing how their business operates because they partnered with Brex. That is really what we're talking about. How do we bring this spirit and this product that we have working so well for this tiniest liver of the market, this 1% of the market, into the, you know, this 98% that's co dominated by income.

而这正是我们所说的‘不受约束的雄心’的真正含义。

And that's really what we mean by unconstrained ambition.

那么你们打算怎么做？具体要怎么做？

We So how do how do you do it? How do you do it?

这仅仅是持续执行同样严格的标准，每天不断提高要求。当你观察市场时，你会意识到大多数人其实根本不知道Brex是什么，甚至不知道有一种更好的方式来思考公司如何管理财务。这确实是故事中非常重要的一部分。尤其是现在，如果你看到人工智能的发展，我们认为存在一个更大的机会，不仅仅是构建金融服务（这是我们早期所做的）和软件，还包括所有的基础设施，以帮助这些公司取代传统上由人工完成、无法自动化的工作。我们对这个机会感到非常兴奋。

It's just continue to execute the same level of rigor, raising the bar every day. And and just really, when you look at the market and you just realize that most people actually have no idea what Brex is, or that there's even a better way to think about how you manage your finances as a company. And that's really a very big part of the story. And especially now, if you see what's happening with AI, you know, we think there's this bigger opportunity to not just build the financial services, which is what we did early on, and the software, but all, you know, effectively all the infrastructure that you build to help these companies replace all the labor that you have traditionally doing work that cannot be automated. And we're very excited about the opportunity.

那么，你认为金融科技的下一阶段会是什么样子？广义上讲，金融科技经历了零利率时代，然后是某种清算时期，现在人们正在试图弄清楚未来几年会是什么样。在你看来，让我们做一些预测。

And what do you see as the next phase of FinTech, broadly speaking, because FinTech is it's had a period where we had the ZERP era, we sort of had the reckoning era, and now people are sort of trying to figure out, okay, what does the next few years look like? In your mind, let's make some predictions here. What are

我们会看到什么？是的。我内部和外部都有一个说法，美国的金融科技实际上还没有真正发生。当你看看市场上使用像Brex这样的现代解决方案的比例时，它仍然非常小。所以很多工作只是让客户到达那里。

we going to see? Yeah. So I have a saying that internally, and I guess externally as well, that FinTech in The US hasn't really happened yet. So when you look into the percentage of the market that is using a modern solution like Brex, it's so tiny still. So a lot of the work is just getting customers there.

这实际上就是瞄准美国的每一家企业，使用现代解决方案并向他们解释，嘿，你不应该再使用美国运通卡了。你不应该再使用你的银行卡了。你不应该再使用传统的基于账单的解决方案了。这里是将所有这些整合到Brex的好处，以及它如何帮助你的业务更快地发展、更聪明地花钱，对吧？以及我们看到客户在他们的团队中部署像Brex这样的解决方案时所获得的所有好处。

It's just literally going after every single business in America, using a modern solution and explaining them, hey, you shouldn't be using American Express. You shouldn't be using your bank card anymore. You shouldn't be using your traditional bill based solution. And here are the benefits of bringing this all under Brex and how it helps your business move faster, spend smarter, right? And all the benefits that we see customers as they deploy a solution like Brex in their team.

所以，我认为这是第一点，即当你只看非常主流的用例时，金融科技实际上还没有真正发生。第二点是当你看看我们所构建的东西时，对吧？我们做出了一些我认为在我们的金融科技公司中相当不同的决定。金融服务方面的一切都是内部构建的。所以所有的卡、支付通道、资本市场、承销、信贷，我们把整个银行技术栈都放在了内部。

So, think that's one, which is, you know, FinTech hasn't really happened when you just look at sort of a very mainstream use case. The second one is when you look into what we've built, right? And, you know, we made a few decisions that I think were pretty different in our FinTechs. You know, everything that you have on the financial services side was built in house. So all the card, payment rails, you know, capital markets, underwriting, credit, we put the entire sort of bank stack in house.

然后我们建立了一个围绕客户的合作伙伴生态系统，尤其是较大的客户。他们需要的每一个解决方案，都与Brex有深度集成。

Then we built this partner ecosystem that surrounds our customers, especially larger ones. Every single solution they need, they have deep integrations of Brex.

而我们做到了

And we did

慢慢地来。当你审视需要服务这些复杂用例所需的广度和深度时

it slowly. When you look into sort of the breadth and the depth that you have to have to serve these very sophisticated use cases

没错。

Right.

我认为这将成为一种趋势，而不仅仅是在金融科技领域。

That's something that I think will be more of a trend than having that in fintech as well.

我想问问你。Ramp融了很多钱。你们肯定也在融资吧？对吗？

I do wanna ask you. Ramp raised a ton of money. You guys must be raising too. Right?

所以，我的意思是，我们这个领域得到了非常好的验证，这里可以诞生大公司，我们对此感到非常兴奋。

So, I mean, it's a there's a very very good validation in our space that there's huge companies to be built here, and we're very excited about it.

你们在融资吗？你们正在外面筹集资金吗？

Are are you raising? Are you are you out there fundraising?

我们没有在融资。

We're not raising.

不融资。好的。英国脱欧不是重点。所以如果你不融资，你知道，在实现运营现金流为正的这一盈利基础上，今天就不多谈了。你们已经实现了运营现金流为正。

Not raising. Okay. Brexit is not raising. So if you're not raising, you know, on the on the heels of of this profit operating can't speak today. Operating cash flow positivity that you you guys have achieved.

这里的策略是什么，来与像Ramp这样的公司竞争？

What is the strategy here to compete with companies like Ramp?

首先，当你审视市场的规模和庞大程度时，真正的竞争对象是美国运通和大银行，我们在90%的交易中看到的是它们。这才是我们竞争的对象。我们努力做的很多事情，可以说是教育市场，就是当你真正去看我们参与的交易和实际的竞争对手时，我们看到的是大银行，它们以同样的方式对待客户。而这正是我们试图战胜的对象。这对我们来说才是真正的竞争。

So first, when you look at the size and sheer scale of the market, right, really, the competition is American Express, the big banks, that is who we see 90% of the deals. That is what we're competing with. And when a lot of what we try to do and sort of educate the market is when you go actually look at the deals that we're in and actually the competition, like who we see is big banks that are treating customers the same way. And that's really who we're trying to win against. That's really the competition for us.

我们看到该领域有很多新玩家，实际上它们几乎起到了一种推动作用，帮助行业变革并意识到有更好的做事方式。所以对我们来说，那才是真正的竞争。只是那些现有银行的数量，它们为客户提供的解决方案远远落后于当前水平，甚至可能还不如四五年前。然后当你加入人工智能，以及缺乏对如何构建一个从根本上改变财务团队运作方式产品的认识和理解时，门槛就异常低。而这确实是我们竞争的对象。

And we see a lot of new players in their space actually almost as a forcing function to help the industry change and realize that there's a better way of doing things. So for us, that's really the competition. It's just the amount of incumbent banks that treat customers with a solution that is so subpar from where they are, from where they were maybe four or five years ago. And then when you add an AI and just the lack of awareness and understanding of how to build a product that actually fundamentally changes how finance team operates, the the bar is exceptionally low. And that's really like who we compete with.

对。很好。佩德罗，感谢你来到节目。我们非常感激。我很期待看到你们为我们准备了哪些新产品，因为你在备忘录中提到的一件事是你们有很多储备，这意味着我很期待看到它们推出。

Right. Great. Well, Pedro, I want to thank you for for coming on the show. We really appreciate it. I am excited to see what new products you you have in store for us because one of the things you talked about in your memo was that you've got a lot of stuff under the hood, which means I'm excited to see it rolling out.

这位是Brex的首席执行官佩德罗·弗兰切斯基。非常感谢你加入我们TI TV的节目。嗯，OpenAI正在积极推进其自研芯片的计划。不过，它很大程度上依赖于租赁AI芯片的业务，而该业务催生了一个名为Neo Clouds的新公司类别。我的同事斯蒂芬妮·帕拉佐罗在这里发表了一篇关于该主题的报道。

That is Pedro Francheski, the CEO of Brex. Thank you so much for joining us here on TI TV. Well, OpenAI is charging ahead with its plans to make its own chips. Largely, though, it relies on the business of renting AI chips, and that business has given rise to a new segment of companies called Neo Clouds. My colleague Stephanie Palazzolo published a story on that topic here.

我意识到那个介绍不太合理。我们在节目中讨论的是Neo Clouds。我的同事斯蒂芬妮发表了一篇关于该主题的精彩报道，她有一些关于Together AI的报道，我想谈谈。斯蒂芬妮，欢迎回到节目。很高兴你能来。

I realize that that intro doesn't quite make sense. We're talking about Neo Clouds here on the show. My colleague Stephanie published a great story on the topic, and she has some reporting on Together AI, which I want to get to. Stephanie, welcome back to the show. It is great to have you.

让我们来全面聊聊Neo Clouds这家出色的企业。你发现的关于together.ai有什么新闻？

Let's talk all about the wonderful business of Neo Clouds. What is the news with together.ai that you found?

我们就从这里开始吧。

Let's start there.

是的，当然。together.ai这家公司到目前为止，实际上一直被认为是一个推理服务提供商，或者有些人称他们为GPU转售商。基本上这意味着Together从其他云提供商那里租用计算资源，然后转售给AI开发者。他们还有另一块业务是托管像Llama这样的开源模型，让它们运行得超级快且成本极低，然后让开发者通过API访问这些开源模型。这至今是他们的业务，但现在他们基本上想要不再从其他公司租用计算资源转售，而是打算自己购买NVIDIA GPU，放在自己的数据中心里，然后运营这些数据中心。

Yeah, definitely. So together.ai is a company that's actually up until this point kind of been known as either an inference provider or some people call them a GPU reseller. And basically what that means is Together goes out and rents compute from other cloud providers and then resells that compute to AI developers. They also have another segment of their business where they host open source models like Lama and get them to run super fast and for super cheap and then basically allow developers to access those open source models through an API. That's been in the business so far, but now it's essentially looking to instead of renting out compute from other companies and reselling it, it wants to actually buy its own NVIDIA GPUs, put them in its own data centers, and then operate those data centers.

所以本质上，它想成为自己的云提供商。Neo Cloud基本上只是云提供商的另一种说法，特指那些较小的云提供商，比如CoreWeave、Crusoe、Lambda，together现在也算是加入了这个行列，它们试图与亚马逊、谷歌和微软这样的巨头竞争。

So essentially it wants to become a cloud provider of its own. Neo Cloud is kind of just another word for cloud provider that is specifically talking about companies, I guess, smaller cloud providers like CoreWeave, Crusoe, Lambda that, you know, together is now kind of joining that are looking to, you know, compete against the likes of Amazon, Google, and Microsoft.

那么，让我在这里澄清几点。所以我们有这群neo clouds，对吧？这些公司是较新的云公司，我想。它们传统的业务基本上是租用计算资源，然后转售给其他客户。

So so let me just clarify a couple points here. So we've got this this group of neo clouds. Right? These are companies that they are the newer cloud companies, I guess. And their business traditionally has been renting compute essentially and then selling them out to other customers.

我们在这里看到的是，纠正我，我是在提问。我们在这里看到的是，together作为这些neo cloud公司之一，在说，嘿，我有点厌倦做Neo Cloud了，我只想成为一个常规的云公司。所以他们出去购买这些芯片，有点像AWS或Microsoft Azure的方式，然后他们将把计算资源卖给他们的客户。我说得对吗？

What we're seeing here, correct me, I'm asking here. I'm not doing what we're seeing here is together one of these neo cloud companies saying, hey, I'm kind of tired of being a Neo Cloud. I just want to be a regular cloud company. So they're going out, they're buying these chips, sort of similar to the way AWS or Microsoft Azure has, and they're going to sell that compute then to its customers. Do I have that right?

是的。所以Neo Clouds和云提供商本质上是一回事。更多的是，together现在想通过购买自己的芯片成为云提供商。但Neo Cloud只是云提供商群体中的一个子集，特指这些较新的初创云公司，比如CoreWeave，它们试图与更大的云提供商竞争。

Yeah. So Neo Clouds and Cloud Providers are essentially the same thing. So it's more that, together, wants to become a Cloud provider now by its own chips. But Neo Cloud is just like a subsection of the Cloud provider group that's specifically referring to these newer kind of upstart clouds like CoreWeave that are trying to compete against the bigger cloud providers.

所以实际上，我们基本上就可以称它们为云提供商，真的。

So effectively, we can kinda just call them cloud providers, really.

是的，是的。基本上，我工作中很大一部分就是给事物起新名字，帮助保留旧名称并让它变得更简单。所以我们基本上可以称它们为...是的。

Yeah. Yeah. Basically, a big a big part of my job is just giving new names to things that help add old names and making it even easy. So we can basically call them. Yeah.

好吧。嗯，我们应该说你做的远不止这些，Stephanie。好了，别低估自己。你想谈谈...你提到了新云领域这里最大的参与者。

Okay. Well, that that we should say you do a lot more than that, Stephanie. Okay. So don't sell yourself short. Do want to talk about so you talked about the biggest players here in the in the neo cloud space.

我们有CoreWeaves。我们有Together AI。但这里很重要的一部分是租用这些计算资源相关的利润率。这就是你提到Together正在寻求购买这些GPU的故事的一个重要原因。是否有证据表明这实际上会帮助他们的利润率？

We've the CoreWeaves. We've got we've got the together AI. You know, a big part of this, though, is the the margins associated with renting out this compute. That's a big reason that you talked about the story that Together is looking to buy these GPUs. Is there evidence that this actually will help their margins?

因为芯片仍然相当昂贵，真的。

Because the chips are still quite expensive, really.

是的，绝对是。所以这里的思路是，如果你从另一个云提供商那里租用计算资源，显然会损失一部分利润给他们。所以如果你自己建造这些芯片，就相当于去掉了那个中间云提供商，理论上这应该有助于提高利润率。在Together的案例中，看起来确实如此。我从消息来源听说，利润率一直在上升，即使在他们开始购买自家芯片后的相对短时间内，现在已经达到45%左右。

Yeah, definitely. So the the kind of thinking here is, if you're renting out compute from another cloud provider, you are obviously losing some margin to to them. So if you were to build those chips by yourself, you're kind of taking out that middleman cloud provider, which which should help margins theoretically. And it does seem like it it has in the case of of together. I've heard from sources that the margins have been, you know, going up even in the relatively short amount of time so far that it's been buying its own chips and now set at around 45%.

对。这仍然远低于传统软件70%以上的毛利率。

Right. That's still a lot lower than, you know, your traditional kind of software 70% plus gross margins.

对。

Right.

但是，你知道，这比过去那种必须先从云提供商那里租用芯片的情况要好得多。

But, you know, it is it is better than kind of what together was experiencing in the past from having to rent out chips from a cloud provider first.

帮我理解一下，为什么客户会寻求从Together获取计算资源或云服务，而不是选择大型超大规模云公司？我的意思是，这些新兴云或小型云提供商的实际价值主张是什么？

And and help me understand why would a customer seek to get compute essentially or cloud services from together as opposed to one of the big hyperscaler cloud companies? I mean, what is the value proposition that these neo cloud or smaller cloud providers actually have?

是的。开发者可能这样做有几个原因。首先，对于许多大型云提供商，有时你必须承诺签订多年合同，比如三年或五年。很多时候，如果你只是一家小型初创公司，你根本不知道三到五年后你的计算需求会是什么样子，甚至不确定公司是否还能存在。新兴云允许客户承诺更小或更短的合同，并且更擅长在需求激增时满足需求。

Yeah. So there's a couple of reasons why a developer might want to do that. So I think first, with a lot of the bigger cloud providers, sometimes you have to commit to, you know, multiyear contracts, three or five years. A lot of the times if you're just a small startup, you really have no idea what your compute needs are going to look like three to five years from now. Sometimes if you're even going to be around as a company, Neo Clouds let customers commit to smaller or shorter contracts and are better at kind of meeting demand whenever it spikes.

嗯。Together声称的另一件事是，他们拥有技术专长，能够以某种方式连接GPU，使其运行更快或成本更低。因此，他们也可以声称自己可能比微软、谷歌或亚马逊等一些大型云提供商便宜得多。而且

Mhmm. Another thing that together claims is that they have the kind of technical expertise to connect the GPUs in a way that make them run faster or that make them cheaper. And so they can also argue that maybe they're a lot cheaper than some of the bigger cloud providers like Microsoft, Google, or Amazon. And

最后一个问题，Together所追求的策略除了必须实际出去购买这些非常昂贵的芯片之外，还有其他风险吗？我的意思是，风险是什么？如果客户不来，那我就囤积了这些芯片，不知道如何处理，还要为所有这些库存付费。

the last question for you, is there a risk at all with this strategy that Together is pursuing other than the fact that they have to literally go out and buy these chips, which are very expensive? I mean, I guess the risk is what? That if the customers don't come, then I'm sitting on these chips and I don't know what to do with them. I'm paying for all this inventory.

是的，是的。我的意思是，这是一个巨大的风险。你还必须应对GPU的风险。我们看到英伟达每年或每两年就推出新的GPU。

Yeah. Yeah. I mean, that's that's a huge risk. And you also have to deal with the risk of GPUs. We were seeing NVIDIA coming out with a new GPU every year or every other year.

因此存在这样一种风险：如果你用H100或Blackwell芯片填满数据中心，到了第二年，就会有新芯片问世，人们就不想再用那些旧芯片了。但我确实认为这些新兴云服务商（Neo Clouds）的一个优势是，英伟达确实在给予他们特殊待遇，提供特殊的芯片分配，因为他们不像微软、亚马逊和谷歌这样的大型云提供商那样试图开发自己的竞争性芯片。

And so there's kind of the risk that if you fill your data center with H100s or Blackwell chips that by the next year, there'll be a new chip out and people won't want to use those older chips. But I do think the one pro that some of these Neo Clouds do have is that NVIDIA is, know, kind of giving them special treatment, giving them special allocations of chips because they aren't trying to create their own competing chips in the way that some of the bigger cloud providers like Microsoft, Amazon, and Google are.

没错。斯蒂芬妮，这真是个引人入胜的故事，我很高兴我们至少稍微厘清了新兴云服务商之间的异同。它们基本上就是较新的云提供商，我现在明白了，我同意你的观点。

Right. Well, Stephanie, it's a fascinating story, and I'm happy that we demystified at least a little bit what the differences and similarities are between Neo Clouds. They're basically just the newer cloud providers. I get it now. I'm with you.

非常感谢你参加节目。我相信我们很快会再见到你，因为新闻进展非常迅速。这位是The Information的AI记者斯蒂芬妮·帕拉佐洛。好了，AI最有趣的应用之一就是它如何在物理世界中带来变革，而不仅仅是人们如何使用互联网。为此，Radical AI这家公司最近筹集了大量资金，它利用AI革新材料科学，这可能影响从太空探索到清洁能源、机器人甚至生物技术的方方面面。

Thank you so much for coming on the show. We I'm sure we'll see you very soon because the news is moving very quickly. That is Stephanie Palazzolo, our AI reporter here at the information. Okay, one of the most interesting applications of AI is how it can make changes in the physical world beyond just how people use the Internet. And to that end, Radical AI is a company that has been raising a ton of funding lately for the way in which it uses AI to innovate material science, which can impact everything from space exploration to clean energy to robots and possibly even biotech.

我想请来该公司的CEO约瑟夫·克劳斯，为我们详细介绍他对这一领域的看法。约瑟夫，欢迎来到TI TV，很高兴你能来。

I want to bring on the CEO of the company, Joseph Krausz, to tell us more about his approach to this sector. Joseph, welcome to TI TV. It's great to have you here.

是的，非常感谢邀请，很高兴加入。

Yeah, thanks so much for having me. Happy to join.

那么，坦率地说，这是一项有点复杂的业务。所以先告诉我们你们做什么，然后我们再谈整体战略。

So look, it's a bit of a complicated business, okay, just to put it bluntly. So tell us what you do, then we'll get to the strategy at large.

是的，完全正确。不仅复杂，而且非常传统。我认为如今许多大型材料公司可能已经存在了五十年、七十五年，甚至有些已达百年。所以它不仅复杂，而且深深植根于传统的商业模式。我们在Radical AI专注于从新材料发现开始，将该材料一直扩大到全面生产规模，而我们目前关注的材料是我们所谓的赋能型基础技术。

Yeah, absolutely. Complicated and incredibly old as well. I think a lot of big material companies today probably going on fifty, seventy five, a hundred years in some cases. So not only is it complex, but deeply rooted in kind of old ways of doing business. What we're focused on here at Radical AI is going from a novel materials discovery and scaling that material all the way up into full scale production, and the materials that we focus on today are what we call enabling based technologies.

这些材料能够解锁我们目前技术上尚未具备的未来能力。

The materials that unlock future capability that we technologically do not have today.

比如什么样的材料？我们具体在讨论哪些种类？

Materials like, what kind of materials are we talking about here?

是的，范围很广。高熵合金是我们重点研究的第一个领域。这些结构金属含有五六种不同元素，具有非常有趣的特性，

Yeah, so it can be a range of things. High entropy alloys is a big first area for us. These are different structural metals that have five or six different elements in them that have really interesting properties,

那些

that

能承受极端环境的材料。比如太空旅行——您在开场白中提到的——会面临极高温度、极低温度、不同压力、各类氧化或腐蚀元素。火箭需要保持结构稳定，电子或半导体特性也要维持，同时还得耐受这些极端环境。这类材料正是我们公司重点研发的方向。

have really extreme conditions. So think about space travel, you mentioned that in the introduction. You get really high temperature, really low temperature, different pressures, different type of oxidation or corrosive elements. You need to be able to maintain structure in a rocket, or different electronic or semiconducting properties, while also being able to withstand those extreme environments. So materials like that are the things that we look into as a company.

所以你们是用AI让材料研发更快更好？具体怎么实现？AI在这其中扮演什么角色？

And so you're using AI to make materials better, faster? How are you how's the what's the AI angle to this?

是的，分两部分。AI引擎主要专注于生成新材料，对吧？这本质上是将人类如今采用的串行研发流程用AI加速。我当科学家时，可能需要花几周、几个月甚至一年研究新领域，阅读文献，提出新假设，然后嗯。

Yeah. So two parts. So the AI engine is really focused on generating new materials, right? And this is really a serial based process that humans do today, turbocharged with AI. You know, when I was a scientist, I would spend weeks, months, in some cases a year, researching a new field, reading publications, making new hypotheses, and then Mhmm.

进行测试。在我们进入测试阶段之前的所有前期工作，如今AI都能瞬时完成，并且准确率极高。过去需要数月才能完成的所谓发现或探索工作，我们现在仅需几分钟或几小时就能完成。而我们在这里做的第二部分工作是一个全自动化的机器人自主实验室。材料科学界有句名言：除非你在实验室里把那个东西制造出来，否则你实际上并未发现新材料。

Out and testing them. All of that first part, before we get to testing, AI can do instantaneously today, and it can do it with very high accuracy. So what used to take months to do what we call discovery or exploration, we can now do in mere minutes or hours. And the second part of what we do here is a fully robotic self driving lab. And we have a big saying in material science that until you make that thing in the lab, you actually have not discovered a new material.

因此，我这边实际上是将这些材料送入全自动的机器人自主实验室，这样我们就能制造这些材料并实际测试其性能。

And so this is my side actually sends these materials into a fully robotic self driving lab so that we can make these materials and then actually test their performance.

所以实验室是...是机器人设备在制造材料吗？这就是你们正在构建的理念？

So so the is the the labs, is it is robotic devices that is that is making the the materials? That's the idea that you're building?

里面有机器人技术。你会看到很多机械臂和不同类型的自动化工具，还有材料加工基础设施。这些都是行业中现有的工具，用于进行表征（也就是搞清楚我制造了什么）以及性能测试。

It's robotics in there. You get a lot of robotic arms and different type of automation tools, and then material tooling infrastructure as well. So these are a lot of tools that exist in the industry to do both characterization, which means, hey, what did I make? And then equity testing. Hey.

我在实验室里制造的东西的性能如何？它们的特性是什么？

What are the performance? What are the properties of the things that I made in that lab as well?

那你们怎么赚钱呢？

And how do you make money?

我们想要大规模销售材料。如果你看看当今大多数大型材料公司（如果不是全部的话），它们都是通过大规模销售材料并从中赚取利润。这正是我们采取的方法。

We wanna sell materials at scale. So if you look at most of the large material companies today, if not all of them, they sell materials at scale and make a margin on that. That's the exact approach that we're taking as well.

然后得到了

And the Got

原因在于这个领域的许多公司已经转向信息学，销售软件或试图授权材料——这在电池领域很常见，但由于诸多原因，这在材料科学中是一种非常糟糕的商业模式。我们实际上是想制造并销售材料。

reason why is a lot of companies in this space have moved into informatics and selling software or trying to license out materials you see that's big in the battery space, and that is a real poor business model in material science for a lot of reasons. We want to actually manufacture and sell materials.

所以你们的目标不是把技术卖给公司，帮助它们更快或更好地制造材料。你们想把技术保留在公司内部，基本上就是扩大你们的制造业务，对吧。

So your goal here is not to sell the technology to companies to help them make their materials faster or better type of thing. You want to keep the technology in house and basically scale your manufacturing operations, essentially.

完全正确。这一点极其重要。这主要是因为当今材料的现状。如果你去看电动汽车、太空、核聚变或航空航天领域的许多终端公司，它们并不是材料科学公司。它们造火箭。

Absolutely. Incredibly important. And that's really because the way materials exist today. If you go a lot of these end companies in EVs, or space, or nuclear fusion, or aerospace, they are not material science companies. Build rockets.

它们造汽车。它们造电池。它们造核反应堆。材料不是它们的核心重点，但真正专注于新发现的主要是学术界或国家实验室的研究人员，这对推动科学理解极其重要，但通常缺乏商业化视角。因此，在这两者之间存在着一个广阔的空缺地带，可以进行这种基础性的新发现，但完全专注于商业化，专注于那些可能受其影响的市场。

They build cars. They build batteries. They build nuclear reactors. Materials are not their core focus, But the only people that really focus on novel discovery is just a lot in academia or national lab research, which is incredibly important to driving your understanding of science, but it's typically missing that commercialization perspective. So in the middle of both of these is a wide open white space to do this fundamental novel discovery, but entirely focused on commercialization in the markets that can be impacted by them.

这些市场会推动发展，所以我们可以填补这个空白。

Those are gonna do it, and so we can just kind of fill the gap there.

那么所有这些会怎么样呢？我相信有大量的博士研究人员，他们在材料科学某个非常具体的领域完成了博士学位，然后被材料制造公司雇佣，他们拥有发现这些新化合物的专业知识。你的意思是，你们基本上要取代这些博士级别的研究人员吗？如果是这样，那些研究人员会何去何从？他们之后做什么呢？

So what's gonna happen to all of these, I'm sure there are a ton of PhD researchers, I guess, who they finish their PhD in something very specific in material science. They get hired by a materials manufacturing company, and they have the expertise to discover these new compounds. I mean, are you essentially saying that you're going to replace these PhD level researchers? I mean, if you are, then what becomes of those researchers? I mean, what do they do then?

是的。实际上，更像是给它们赋能。很多博士毕业生，我自己也曾一度攻读博士学位，正如你所说，他们花费数年时间在某个领域发展专业技能，结果到了研究中心却只是进行移液和称量不同样品的工作。你并不需要博士学位来做移液和称量样品的事情，这相当简单，高中水平就能完成。

Yep. Actually, rather supercharge them. A lot of PhDs that go through, and I myself was going through a PhD program at one point, really spend years developing expertise, as you mentioned, in an area, only to get to a research center and pipette and weigh out different samples. You don't need a PhD to pipette and weigh out different It's quite simple. You can do it at the high school level.

因此，通过构建自动驾驶实验室和我描述的这个平台，实际上可以让那些博士们把全部脑力从称量样品转移到真正思考如何实施新颖的材料系统，以及探索能解锁更优性能的新设计系统的来源。现在，对于在特定领域获得博士学位的科学家来说，有机器人和自动化技术为他们运行所有实验，他们可以持续关注实验结果、所从事的应用领域，并在新材料发现与——

And so actually, by building self driving labs and building this platform that I described, you can have those PhDs spend all of that brainpower on weighing samples to actually thinking about implementing novel material systems and where new design systems that they can build to unlock better performance come from. And now for the scientist who gets a PhD in a specific area, has robots and autonomy running all their experiments for them, and can constantly be looking at the results of those experiments, their application that they're working in, and tying this bridge between novel materials discovery-

对。

Right.

以及情感性能之间搭建桥梁。

And emotional performance.

最后一个问题。你们是否已经制造出正在销售的产品，或者距离拥有可销售产品还有多远？

And last question for you. Have manufactured a product yet that you're selling, or how far are we from you getting to having a product to sell?

是的，好问题。我们还没到那一步。我们是一家相当年轻的公司，大约成立18个月，所以尚未全面生产材料。目前我们是在公司实验室里制作材料。

Yeah. Good question. We're not there yet. We're a pretty young company, about 18 old, so we haven't manufactured full materials yet. We are making materials here at the company in our lab.

我们在几个不同领域已经制造出一系列不同的合金，但制造步骤是在我们称之为验证阶段之后进行的。我们制造的许多材料必须经过实际测试。如何

We have a bunch of different alloys that we've made in a couple of different areas, but that manufacturing step comes after what we call validation. A lot of these materials that we make have to be tested in real How

你认为要多久才能开始产生收入？

do you think it'll take to finally start making revenue?

我们有多种创收方式。大规模销售材料可能需要大约18个月到两年左右的时间。我们必须经历验证和制造流程。我们认为这大约是一个24个月的过程。

We have different forms the way we make revenue. Eighteen months to probably two years, give or take, to sell materials at scale. We have to go through that validation and that manufacture process. We think that's about a twenty four month process.

太好了。这样吧，约瑟夫，下次请你回来上节目时，不如在实验室里做采访？你可以带我们参观实验室，给我们展示讲解一下。也许我们可以用iPhone摄像头，通过FaceTime连线，让我们看看各种不同的设备。

Great. Well, I'll tell you what, Joseph. Next time we have you back on the show, why don't you do the interview from inside the lab? You can show us around the lab that you can give us a bit of a show and tell. Maybe we'll get, you know, an iPhone camera, you can sort of FaceTime us in and show us all the different devices.

我相信这听起来非常酷，而且内部肯定更酷。非常感谢你来参加节目。这位是Radical AI的首席执行官约瑟夫·克里斯特。今天的节目就到这里，各位。

It's I'm sure it sounds very cool. And I'm sure it's even cooler on the inside. Thank you much for coming on the show. That is Joseph Christ, the CEO of Radical AI. With that, that does it for today's show, folks.

提醒一下，我们的直播时间是周一至周五太平洋时间上午10点，东部时间下午1点。我要感谢本次节目的首席赞助商亚马逊网络服务。感谢大家的收看，我们非常重视每一位观众。我已经开始期待明天的节目了。

A reminder, we are 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. I want to thank you for tuning in. We really do appreciate your viewership. I am already excited for our next show tomorrow.

祝大家周二愉快。暂时先到这里，再见。

Have a great rest of your Tuesday. Bye bye for now.