Robinhood CEO 谈让 AI 替你交易

Vlad，我刚看了 Robinhood Agents，问题很多。对我来说最重要的一点是，这大概是很多人第一次接触到 AI 替自己下单这个概念。这件事对你们、对整个行业、对 AI 本身、对人们怎么用 AI，都是大事。我想听你讲这个想法从哪来，你们为什么决定做。你们不久前刚做了 MCP，人们可以把 Robinhood 的数据接到自己的 AI 工具里，那些都有了，但这个感觉更大。

Vlad, I just got to see Robinhood Agents and have a lot of questions. I think the big thing for me is it's the first time that a lot of people will probably be introduced to the concept of AI making trades on their behalf, and it seems like a big deal for not only you guys, but the industry, AI in general, how people use AI. I'd love to hear from you the genesis of this idea and why you guys decided to do it, 'cause you did the MCP thing recently where, you know, people can use Robinhood data in their own AI harnesses and all of that, but this is. This feels, like, bigger.

是。它的起源和动机是，我们想让交易者拥有最好的工具，想把一家成熟对冲基金或算法交易公司的全部能力放进你的口袋。我职业生涯的起点是算法交易，所以我多少知道那是什么样。我觉得这会是一项很长的使命，某种程度上永远不会结束，因为富人总会找到办法拿到更好的东西，所以我们的工作永远做不完，这也是它有趣的地方。我们开始让 Robinhood 变成智能体式的，把这个智能体金融品类做成真的。

Yeah. I guess the origins of it, the motivation is that we want to empower traders with all the best tools, and we want to put all of the power of a sophisticated hedge fund or algorithmic trading firm in your pocket. the beginning of my business career, I was in algorithmic trading, so I have a little bit of experience of what that looks like and what it is. And I think this is,. It's gonna be a long mission that, to some degree, never ends because the wealthy always find a way to get better stuff which means our job is never done and that's part of what makes it very fun. So the. When we started our path, to making Robinhood agentic and making this agentic finance category real.

而且几乎没有别人在做这件事，所以我们像是在雾里走，一边搭基础设施，一边改产品。今年早些时候我们推出的第一版，叫 Robinhood MCP，也就是 Robinhood for Agents，有两个部分。一个是交易，另一个是智能体信用卡。交易从一开始就放在单独账户里，隔得很干净。你必须主动把钱从银行，或从你的另一个账户，转到智能体账户。它和退休账户分开，也和你的主 Robinhood 账户分开。起初它只能做没有保证金的股票，也就是不加杠杆、现金交割的股票。

And by the way, nobody else is really doing this, so it's sort of like we're wandering through the fog and iterating on the infrastructure and the product at the same time. So the first version of it, which we launched with our Robinhood MCP, Robinhood for Agents, earlier this year, had two components. One was for trading, the other was an agentic credit card. And for trading, it started in a separate account, so it's very cabined off. You have to affirmatively move money from either your bank or from another one of your accounts to your agentic account, so it's separate from your retirement, separate from your main Robinhood account. At first, it could only trade equities without margin, so unlevered, cash-settled equities.

然后随着时间，因为我们想从轻的地方开始，先看清客户怎么用、他们想要什么，资产范围扩大了。现在可以做期权，可以做加密货币，我们也加了有限的保证金。我们发现，首先客户很喜欢。已经有超过 10 万客户真正开通了智能体交易。指标在持续、稳定地上升，尤其是我们加上新能力、打开有限保证金之后。轨迹和留存都非常好。接下来有几件明显要改进的事，它们一直在添摩擦。其一是，没有多少人愿意用 Claude Code 或 Codex。

And then over time, because we wanted to start light and kind of figure out how customers would use it and what they wanted, and so over time, we've expanded assets. Now you can do options. Now you can do crypto. We've added limited margin capabilities as well. and what we found was first off, customers love it. We have over 100,000 customers that, have actually onboarded onto our agentic trading offering. and as we look at the metrics, it's, like, continually and steadily increasing, especially when we add new capabilities and we unlock things like limited margin. And so the trajectory and the retention has been really good. What people. there's a couple of things that are clear next steps that we wanna improve that have been just adding a little bit of friction. One is not a lot of people wanna use Claude Code or Codex.

在我们这个圈子里，大家技术上更熟一点。我们大概认识很多用 Claude Code 和 Codex 的人，但对普通个人来说，它还是太复杂，仍然像给程序员用的开发工具。你从 Claude 出发，接上 MCP，连到 Robinhood，还得在另一个地方盯着自己的策略，这个摩擦不理想。这是我们用 Robinhood Agents 解决的一个问题。第二个问题是，通用模型接上 MCP 之后，很多时候不愿意交易。它们不自在，训练数据里看不到交易轨迹，本来就不是为这个造的。

And, you know, in our circles, we're a little bit more tech-savvy. We probably know a lot of people using Claude Code and Codex, but for your average individual, it's a little bit too complex. It's still kind of a developer tool for computer programmers. And the friction of going from your Claude, connecting it to an MCP, linking it to Robinhood, and having to, like, monitor in a different place where you're managing your strategy is not ideal. So that's one problem that we fixed with Robinhood Agents. The second problem is that a lot of times, the general models, when you plug in an MCP, they don't wanna trade. They're not comfortable with it. They don't see the, you know, trading traces in their training data. They're not built for it.

所以我们花了大量功夫，而且我认为这部分会继续做下去，让模型非常擅长用我们的工具，知道 Robinhood 能做什么，知道怎么利用这个巨大的生态，并且不断迭代，让它不只是愿意交易，而是把交易周围的每件事都做得很好。这就是 Robinhood Agents 的一大步。我们把它放进你的应用，让你在一个地方管理，给你一套完整的智能体交易体验。这只是开始。你可以想象更多资产类别、更多能力。你作为个人在 Robinhood 上能做的每一件事，都应该能被你的智能体用到。

And so we've spent a lot of work, and this is the part that I think will continue, to make the models very good at using our tools, being aware of what Robinhood can do, and how best to take advantage of the huge ecosystem and actually iterate to make it, make it not just wanna trade, but be really good at everything around it. And so, that's the big step in Robinhood for, Robinhood Agents. We're bringing it into your app, making it so you can manage it in one place, and really kind of giving you an integrated experience for agentic trading. And it's just the beginning. You imagine more asset classes, more capabilities. Everything that you should be able. Everything that you do as an individual on Robinhood should be accessible to your agent.

随着时间，我们想系统地过一遍，一家极其成熟的算法交易公司或对冲基金能用到的一切，再通过 Robinhood 交给我们的客户。这意味着在性能和延迟上继续改进。

And over time, you know, we wanna systematically go through everything that an extremely sophisticated algorithmic trading firm or hedge fund would have access to and give that to our customers through Robinhood, and that means improvements on performance and latency.

很多高频公司在争，谁能成为数据中心里、交易所或做市商那台交换机上的第一个端口，对吧？这件事可以走得很深。所以性能和延迟这边还有更多工作，我们已经取得很大进展。研究和新工具也还有更多工作。Agent apps 是个新东西，我们可以接到第三方开发者，他们给智能体提供工具，我们帮他们变现，再把这个能力交给用户。还有更多数据，以及和更广的 Robinhood 生态的集成。我们看到的世界是，北极星应该是客户相对对冲基金和算法交易公司拥有优势。要走到那一步，工作是做不完的。

you know, a lot of these high-frequency firms, they compete over who gets to be the first port in the switch in the data center where the exchange or market maker is, right? so the. Yeah, it goes quite deep. So more work on the performance and latency side, which we've been making incredible progress. More work on research and adding tools. - Agent apps is a new thing, where we can actually link to third-party developers who provide tools for our agents, help them monetize, and give that capability to users. And, more data and more integration with the broader Robinhood ecosystem. So yeah, we see a world where actually the north star is our customers should have an advantage over the hedge fund and the algorithmic trading firm, and there's just a endless amount of work, to get to that point.

对还没用过、没见过的人来说，设置大概只要 60 秒。你给智能体起名，选模型，决定要不要打开交易审批。我觉得你们把这件事做成默认需要选择，很重要，这样人们可以先习惯它，因为对很多人这会是一个新概念。尤其是普通的 Robinhood 用户。正如你说的，他们并不知道专业交易者能用的那些复杂工具和数据源。

And for people who haven't used it yet or seen it, I mean, it basically takes 60 seconds to set up. You name your agent, pick your model, decide if you wanna have trade approvals on for the agent, which I think is important you guys are making it opt-in for the experience so that people, I think, can just get used to it, 'cause it's a. It's gonna be a new concept for a lot of people. Especially the average Robinhood user. Like you said, they don't know about all these sophisticated tools and, data sources that professional traders have access to.

我们一开始放了很多护栏。有些客户会说，我想要全部权限。但在这个阶段，我们仍希望它是一个主动做出的决定。你要它自主，还是每笔都要你批准？默认情况下，我们会让你批准每一笔，但如果你想试自主，会有一条路。再加上它在单独账户里，你必须主动决定让它动你的钱。它不会自己跑偏去做。我们花了很多时间想这些取舍。我觉得我们有一个很好的底座，可以继续迭代，并领先这个新市场。我认为这会是一个非常大的市场。

And we put a lot of guardrails in at the beginning. And some customers actually say, "Hey, I wanted to have access to everything." But at this point, we still wanna make it to be a affirmative decision. Do you want it to be autonomous, or do you want to approve everything? Well, by default, we'll make you approve everything, but we'll have a path if you wanna try autonomous. And then the fact that it's in a separate account as well makes it so that you have to make an affirmative decision for it to trade your money. It won't just you know, go rogue and do that. So we spend a lot of time thinking through the trade-offs, and I think we have a good base from which we can continue to iterate and to lead this new market, which I think will be a really big market.

本期节目由 Mercury 赞助。这是包括我在内、超过 30 万名创业者喜爱的 AI 原生银行服务。访问 mercury.com 了解更多。Mercury 是金融科技公司，不是银行。详情见节目说明。节目也由 Atlassian 的 Jira 赞助。团队和智能体在那里获得上下文、协同和控制，把工作往前推。在 jira.com 免费试用，网址是 J-I-R-A.com。还要感谢 Granola，给一场接一场开会的人用的 AI 记事本。它跟着你的工作走，让你把注意力放在真正重要的事上。到 granola.ai/sources 试用，用优惠码 sources 可免三个月费用。

This episode is brought to you by Mercury, AI native banking that's loved by more than 300,000 entrepreneurs, including me. Visit mercury.com to learn more. Mercury is a fintech, not a bank. Check the show notes for details. This episode is also brought to you by Jira by Atlassian, where teams and agents get the context, coordination, and control to move work forward. Try it free at jira.com. That's J-I-R-A.com. Thanks also to Granola, the AI notepad for people in back-to-back meetings. It works everywhere you do, and lets you focus on what matters. Try it at granola.ai/sources, and use the code sources for three months off.

关于单独账户：你本来想做成一个账户吗？分开是因为监管，还是因为隐私？你觉得有没有一天，AI 和你会在同一个账户里？

The separate account piece: did you want it to be one account? Did you have to do separate account for regulatory reasons or privacy reasons? Do you see it ever being one account, the AI and you in your, in your account?

我觉得，把它连到你的主账户，对我们来说很容易。我们想这么做，一部分原因是先从小处开始，万一客户有反馈，我们可以小步试进这个体验。很多人一直在要求我们放进主账户。你也可以想象一个开关，打开之后就能把智能体接到你已有的账户上。这是我们在想的事，但我不觉得眼下这是产品的限制。我觉得很多客户其实喜欢把钱转过去、单独放着，所以至少眼下，默认仍会是单独账户。

I think that, I mean, it would be very easy for us to link it to your main account. I think part of the reason we wanted to do it is just to start small and to make sure that, you know, if we do get feedback from customers, we sort of tiptoe into the experience. A lot of people have been asking us for putting it in the main account. Or you could imagine, like, a flag so that you could flag and turn on an agent, plug it into an existing account that you have. So yeah, that's something we're thinking about, but I don't think at this point it's, a limitation of the product. I think a lot of, a lot of customers really like being able to move things over and keep it separate, so probably the default being separate account will continue, at least for the time being.

我们现在就在硅谷，在你门洛帕克的办公室。这里整家整家的创业公司都在用 AI 做自主交易，对吧？这是一个很有意思的品类，我也想了很多。AI 替人管钱，而且不只是替老手，是替每天的散户。你们现在就在做。我很好奇你觉得这会走到哪。我们会不会进入这样一个世界：大多数人让 AI 不只做决策，还做交易、执行，全部交给 AI？你觉得这是不可避免的吗？

Out here in Silicon Valley where we are, we're here at your office in Menlo Park, I mean, there's whole startups being created to use AI to do autonomous trading, right? And it's a really interesting category and something I've been thinking a lot about, and the idea of AI managing money for people, and not just, you know, sophisticated people, but everyday retail investors, and you're doing that now with this. I'm curious, like, where you think this is going. Are we going to a world where most people have AI make not only their decisions, but their trades, execution, all of that, have AI do that for them? Do you think that, this is inevitable?

我认为很大一部分成交量会由 AI 智能体产生，因为它们没有人类交易者的很多限制。它们可以 24 小时在线，对吧？可以同时跨多个资产类别。人类直接用 Robinhood 应用时，有一件事很难：不同资产类别要去不同的页面。我们往应用里加的东西越多，某种程度上就越得费劲让体验保持连贯。比如我们从股票起步，股票体验很简单，然后加了期权、加密货币。现在有期货、预测市场。在海外的一些产品里，我们还有 Robinhood Chain 上的 DeFi，去中心化金融的产品和资产。

I think a very large portion of the volume will be generated by AI agents, and I think that's because they don't have a lot of the limitations that we have as human traders. They can be up 24/7, right? They can be multi-asset class simultaneously. So you know, one of the things that are challenging for humans using our Robinhood app directly is you have to, like, go to different screens to get different asset classes. You know, the more things we add to the app, to some degree, the more work we have to do to make the experience coherent. So for example, we've got. We started with stocks, and it was a very simple stock experience, but then we added options, crypto. Now we have futures, prediction markets. In overseas in some of our other products, we've got DeFi with Robinhood Chain, decentralized finance products and assets.

我们还有永续期货、银行、退休、投顾。东西很多，这种复杂度对人类是挑战。对智能体来说，那只是更多上下文、更多机会、更多交易想法。我们已经在 Robinhood 智能体最早的版本里看到，人们在构建期权策略。就是说，我有一件非常具体的事想做，想下一笔交易，也许是铁鹰式。我只要告诉智能体，它会拿数据，把它拼起来。这件事变得很容易，这是一种自动化用法，像是去掉一个小痛点。但你可以想象，我们把它接到所有资产、所有数据源上，用例会更复杂，策略会丰富得多。

We've got perpetual futures, banking, you know, retirement, advisory. So it's a lot of stuff, and that complexity is a challenge for humans. But for agents, it's just more context, more opportunity, more trade ideas. And so, you know, we've started seeing the earliest iterations of Robinhood agents, we see people constructing option strategies, right? And it's like, I have a very specific thing that I wanna do a trade that I wanna put on. Maybe it's an iron condor or something and, you know, it's just I can tell my agent, and it'll it'll take the data. It'll put it together. It makes that really easy, and that's kind of a automation use case. It's like removing a paper cut. But you imagine we plug it into all assets, we plug it into all data sources, and you'll see the use cases getting more complex, the strategies getting a lot richer.

直接让它做。

Just do it

是的，我看到一个世界：对个人投资者、对我们整个客户群来说，很大一部分交易量由 AI 智能体产生。但这不一定意味着绝大部分资产都会被智能体拿去交易。我觉得绝大部分资产大概仍会长期持有，但交易活动会越来越多地走向智能体。

Yeah, I see a world where a large portion of the trading volume for an individual investor, for our customer base writ large, is generated by AI agents, but that doesn't necessarily The vast majority of the assets, are gonna be agentically trading. I think, I think the vast majority of the assets will probably be, you know, held long term, but the trading activity, will probably go more and more towards agentic.

时机也很有意思，因为人们开始习惯 Muse、Instinct 这类产品替他们做购买，用 AI 省钱、付钱。所以我觉得你们的时机其实很好。普通的日常 AI 用户开始意识到，我可以用 AI 和我的钱打交道，而六个月前几乎没人这么做。

It's an interesting time too, because people are getting used to products like Muse, Instinct, et cetera, making purchases for them agentically, and using AI to save money, pay for things. So I feel like your timing's actually pretty good, because the average everyday AI user is starting to realize, like, "Oh, I can use AI to interact with my money" in a way, in a way that six months ago no one was really doing.

完全同意。我觉得当时仍有一道缝：我不舒服把邮箱交给 AI，也不舒服把券商账户的钥匙交出去。我们正在经历这个转变，其他新技术也发生过。我记得互联网刚成为大事的时候，人们非常紧张，不敢在网站上输入信用卡号。

Yeah, absolutely. I think there was still that gap of, "I don't really feel comfortable," you know, giving my AI my email or you know, the keys to my brokerage account. I think we're going through the shift, and it happened with other new technologies too. I mean, I remember when the internet first became a big thing, and people were very nervous punching their credit card number into a website.

我记得。是的。

Yeah, I remember. Yeah.

感觉不安全，像是一个安全问题。大概过了 10 年，人们才看清，这其实比在实体收银台把信用卡递出去更安全。手机也一样。我们开始做手机的时候，奥弗顿窗口还没移到“手机是管理财务的默认方式”。人们仍觉得手机是打游戏、做不严肃事情的地方。所以“我的财务在手机上”几乎像是亵渎。我觉得这里会看到同样的事。我们已经开始看到了，所以我们可能偏早，但我觉得那个时刻很快会到。

Didn't feel secure. It was like a safety concern, It took probably 10 years before it became clear that it's actually more secure than handing your credit card over at a brick-and-mortar register. Same with mobile, you know? When we got started with mobile, it was still the Overton window hadn't shifted to mobile being your default way of managing your finances. Still, people thought of it as, "It's where I play games or do not serious stuff." So the idea of my finances being on my phone seemed sacrilegious almost. Yeah, I think, I think we'll see the same thing here. We're starting to see it, and so we're probably early, but I think, I think the time is soon.

你们团队给我看演示的时候，我也在想，这正好碰到你们公司最初的使命：把金融民主化。因为 AI，你说的那些小痛点，下复杂的单，这些事，可能让更多人用自己的钱做更复杂的事，甚至不必完全理解所有复杂之处。

When your team was showing me the demo, I was thinking too, this really gets at the original mission you had for the company, democratizing finance, because AI, those paper cuts you described, putting on complex trades, all these things, potentially a lot more people could be doing more complex things with their money and not even needing to fully understand all the complexities.

是的，而且 AI 可以向你解释，比读一本书好得多。期权交易大概是个好例子。期权交易非常难。

Yeah, and the AI could explain it to you know, much better than reading a book. Yeah, I mean, options trading is a, is probably a great example. Options trading's very difficult.

界面变简单了，但它本身仍然很复杂。你们曾经、我想现在仍然有学习期权方面最好的界面，但我得老实说，它还是复杂。

The interface becomes simple, but it's so complex. And you all had, I think still do have the best interface in terms of learning options but I gotta be honest, it still is complex.

是的，尤其是多腿策略。我们可以尽量画出来，说这笔交易上行有限、下行有顶，并用图形展示。我们也可以生成这些可视化，交给智能体。智能体能做的，是接住你的意图。比如，我想对这只股票下一笔交易，我想封住下行，我想要无限上行。它们在为你搭出期权交易的正确几条腿这件事上，做得相当好。

Yeah, particularly if you go into these multi-leg strategies. I mean, we can do what we can to illustrate and say, "Okay, well, this trade, you know, you've got limited upside-" capped downside, you know, and we'll show that graphically. And, you know, we can generate all of those, all of those, visualizations and make them available to agents as well. What the agents can do is take your intent. It's like, "Hey, I wanna put on a trade for this stock. I wanna cap my downside. I wanna have, you know, unlimited upside," those sorts of things, and it'll actually. They do a pretty good job of constructing the right legs of the options trade for you.

我很好奇，AI 扩散之后，你觉得投资和交易里的优势会在哪。你这场谈话一开始就说，我们一直在追赶那些大机构，想让普通人占一点上风。大机构也一直在找自己的优势。这场猫鼠游戏感觉不会很快停。但如果把图画大，一个所有人都用 AI 管钱的世界。

I'm curious where you think the edge will be in investing and trading as AI proliferates. I mean, you started this conversation by saying, "We're always trying to play catch up with the big guys, trying to get people a leg up." The big guys are always gonna be trying to get their edge. so this cat and mouse game feels like it's not stopping anytime soon. But if we think big picture, a world where everyone's using AI to manage their money.

我觉得优势会一直在移动。不会有“一切都静止了，优势就在这里，alpha 就在那里”这种事。人们一直在做新产品。就说我们，Robinhood Chain 是我们新推出的、为现实世界资产优化的区块链。它一上线，开发者活跃度和 DEX 成交量就大得让我们吃惊。其他链上的很多交易者立刻桥接过来，因为有这些新资产可以交易，股票代币，还有一些 meme 和其他资产。那整个就是一个新生态。所以你会看到新生态继续形成，新的资产类别继续出现。

I think, I think the edge will be always moving. There's not gonna be like, "Oh, everything's still," and, you know, the edge is right here, the alpha's right there. I mean, people are always building new products. You know, even us, you look at, Robinhood Chain, which is our new blockchain that we rolled out optimized for real-world assets. I mean, as soon as that went live, it surprised us with just a huge amount of developer activity, DEX volume. A lot of the traders, from other chains immediately started bridging over, because there were all of these new assets they could trade, the stock tokens, but also, you know some of the memes and other assets. All of that was just a new ecosystem. So I think you'll see new ecosystems continue to form new asset classes.

智能体交易的兴趣里，有很大一块在预测市场。你不只拥有典型的金融资产，还可以交易现实世界的事件。你可以交易美联储的决定、体育。加密货币已经非常流行。所以你会看到新的资产类别进来。我认为 Robinhood 的优势是，我们是少数、甚至可能是唯一能在一个地方做所有这些事的：一个容易共享的余额，统一的 KYC，简单的资金移动。你的智能体可以同时用到这些，然后你可以做跨多种资产的策略。我觉得有意思的地方从这里开始，因为这种复杂度和这种策略，一个没有辅助的人很难自己拼出来。

You know, a lot of the agentic trading interest has been in prediction markets too where you not only have your typical financial assets, but you can trade on real-world events. You could trade on Fed decisions, sports. crypto has become very popular. So you'll see new asset classes coming in, and I think the advantage that Robinhood has is we're one of the few, if not the only, where you could do all of these things in one place with one easily shared balance and, you know, uniform KYC, simple money movement. So for your agent, they can just have access to all these things, and then you can have strategies that incorporate multiple assets. And I think that's where it's gonna start to get really interesting, 'cause that type of complexity, that type of strategy's gonna be very difficult for a human, unassisted, to be able to put together.

所以你不担心这样一个世界：每个人都有同样强大的 AI 替自己交易，然后实际上就没有优势了？

So you don't worry about a world where everybody has the same kind of super powerful AI that can trade for them, and then there's effectively no edge?

任何事情里的优势都来自差异化。如果你和所有人做一样的事，按定义你就没有优势。所以你得想办法变得不一样。我觉得交易一直都是这样。

The edge in anything is built by differentiation. If you're doing the same thing as everyone else, kind of definitionally, you have no edge. So you'll have to figure out how to be different, and I think that's been always true for trading.

听起来并不只是“我的 AI 比你的好”。周围还有很多产品上的东西。

It's not as simple as like, "My AI is better than yours." There's a lot of, like, product stuff around it sounds like.

如果你的 AI 比别人好，我觉得那会是一种优势。但我不认为会有很多人去训练自己的 AI。把智能体接到独特的数据上，智能体应用商店里的应用会大量出现。现在数量还比较少，但它会开放，我们会加进更多。事情在某种程度上会集中，因为这些模型现在什么都在看；但我们能给它们的东西、上下文和工具，也在增加。

I mean, if you have a better AI than other people, I think that would be one form of edge. But yeah, I don't think very many people will be training their own their own AIs. But I mean, connecting it to unique pieces of data, we'll have a proliferation of apps in the agent app store. which now we have a relatively small number, but it's gonna be open, and we're gonna be, we're gonna be adding a lot more. Things in some degree, to some degree do get centralized because we have these models that are now looking at everything, but also the amount of things we can give them, and the amount, the context, and the tools are increasing

我想问智能体市场，但先快速问一下模型。你们在对 OpenAI、Anthropic 那些大模型做很多后训练、微调吗？Robinhood 不会为这件事自己做一个模型，对吧？

And I wanna ask about the agent marketplace, but on the model side quickly, are you guys doing a lot of post-training, fine-tuning on the big, you know, OpenAI anthropic models? Are you. Robinhood's not gonna make its own model for this, right?

对。目前我们没有在做微调，但我们在内部对某些模型做过。我们也为自己预训练过一些专门模型。这一版，我们想从客户已经认识的模型起步，让他们可以选择部署哪一个。但这不意味着产品不会变。你已经能看出来，这个产品在演进，六到十二个月之后，它看起来大概会完全不同。

Yeah. Currently we're not doing any fine-tuning, but we have done that internally, for certain models. We've also pre-trained some specialized models for ourselves. Yeah. with this one, we wanted to start with, you know, the models that our customers would already know and give them some choice for what to deploy. But yeah, that doesn't mean that. I mean, you can tell already this product has evolved, so it's gonna look completely different probably six to 12 months from now.

这个市场，是不是和你们谈其他业务线的方式一样？它是下一条业务线吗？你觉得它会不只是让 Robinhood 智能体变得更好，而是公司真正的一门生意？

The marketplace, is this in the same way you talk about all the other business lines you guys have? Is this the next one? Do you see this being more than just a way to make Robinhood agents better, but an actual business for the company?

我确实认为，Robinhood 绝大部分生意会来自交易和资产，也就是客户在平台上做事、把钱带进来，以及由此产生的二阶效应。订阅对我们越来越重要。你看主应用里的 Robinhood Gold，那是一年数亿美元收入的生意，从一个很小的基数长起来的。我们想把它做得尽可能好，并做出这个市场，让开发者和服务提供商有动力把数据接进来。它会不会和苹果 App Store 一样大？我不确定。我从不预设、也不预先规定我的每门生意相对另一门该有多大。不是那种事先定好的事。

I do think the vast majority of Robinhood's business is gonna be from transactions and assets, so just customers performing activities on the platform and bringing their money in and kind of the second-order effects there. subscription is, a bigger and bigger thing for us. I mean, you look at Robinhood Gold for the main app, that's a multi-hundred million dollar a year revenue business started from a small base. We wanna make it as good as possible and create this marketplace where developers and service providers can come in and have an incentive to plug in their data. Is it gonna be as big as the Apple App Store? I'm not sure. but yeah, I guess I never try to presuppose and preordain how big each of my businesses will be relative to the other ones. It's not like, you know.

人们总问我，你希望加密货币占你生意的百分之几？我希望我的加密生意是最大的加密生意。我希望股票生意是最大的股票生意。我希望它们互相争，看谁更大。我为什么要说，加密就保持在 7% 吧。

People always ask me, "Well, what percentage of your business do you want crypto to be?" I want my crypto business to be the biggest crypto business. I want the equities business to be the biggest equities business. I want them to compete with each other over which one's bigger. Yeah, like why would I be like, "Oh, no. Yeah, crypto I wanna keep to, like, 7%."

对。对。[笑]

Right. Right. [laughs]

那真不错。7%，咱们就别停在那儿。

That's really good. 7%, let's just not keep it there.

这是一个组合，Vlad。不过，所以这个市场——

It's a portfolio, Vlad. Well, no, but so the marketplace

这是一个组合，里面每一项都必须尽可能大。

It's a portfolio where everything just has to be as big as possible.

但这个市场，商业模式是某种抽成，还是订阅分成？你们一开始放了 Unusual Whales，还有一些很好的数据源。除了接触到 Robinhood 用户，他们得到什么？你希望它有多复杂？你觉得它会一直这么简单吗？

The marketplace though, I mean, is the business model some kind of take rate or subscription sharing component? How do you. 'Cause you've got, like, initially you've got Unusual Whales in there. You've got some great data sources. What is in it for them except exposure to the Robinhood user? How complex do you want that to be? Like, do you think it stays that simple?

我觉得会，因为一旦你做基于交易或基于资产的分成，会有几个原因让它很难。其一，很多客户不想把那些信息分享给应用。你不想让整个市场看见你账户里有多少钱。人们对自己的交易策略非常私密。你不会想让别人检查你的每一笔交易。所以我们提供的一个好处是一层隐私。人们已经放心在 Robinhood 里交易，因为他们知道我们不会把他们的交易数据随便分享给所有人。

I think so because whenever you get into, like for example, it's hard to do a transaction-based or an asset-based revenue share for a couple of reasons. One is a lot of customers don't want that information shared with the apps. You know, you don't want, you know, the entire marketplace to see how much money you've got in your account. You also don't want them. People are very private about their trading strategies. You're not gonna want people, like, inspecting all of your trades. Right? So yeah, one of the benefits we provide is we provide a layer of privacy. People know that. people are already comfortable with their trading in Robinhood because they know we're not just sharing their trading data with everyone.

所以我觉得订阅是干净的。它不会把我们带进那些麻烦事：要么让合作方必须拿到牌照，要么让客户觉得自己的数据流到了他们不想去的地方。

So I think subscription is clean, and it, yeah, it doesn't get us into, like, these messy things of either making sure that the partners have to get regulated or that we don't want customers to feel like their data is going anywhere they don't want it to go.

当然。那些大机构、对冲基金、量化，他们在做的事情里，有没有让你觉得“我现在就想把它放进这个市场”的？有没有你可以透露的、人们还不知道的、这些大机构的优势，是你打算引进来的？

Of course. Yeah. Is there anything that the big guys are doing, the hedge funds, et cetera, where you're like, the quants, all that you're like, "Man, I'm trying to get that in right now to the, to the marketplace"? Are you, are there things that you can tease that people may not know about the edge that these big guys have that you're like, "I'm, I'm gonna bring this in"?

一个很好的北极星是这样。我再次说明，我不会给你任何时间表，说我们什么时候会有类似的东西。但大机构很多策略是被低延迟打开的。你必须在物理上靠近交易所和撮合引擎。Robinhood 现在不提供这个。我自己是从高频交易起步的，我觉得这是一个值得问自己的问题：那会是什么样？我们能把这件事民主化吗？我们能不能做到？我们现在先从整体上把性能和延迟做得更好开始。但有一天，你能不能把一个交易策略部署到和交易所共置的地方，而且方式非常简单，你不必是数据中心或网络方面的专家。

I mean, a great North Star would be this. And again, I'm not giving you any timeline for when we would have anything like this, but a lot of the strategies that the big guys have are unlocked by low latency. So you have to be physically proximate to the exchanges and the matching engines. And right now, Robinhood doesn't offer that, but, you know, of course, my beginnings were as a high-frequency trader, and, you know, I think it's an interesting question to ask ourselves, what would it look like? Can we democratize that? Can we actually. And we're starting with just making the performance and latency generally better. but, you know, one day can you deploy a trading strategy co-located with the exchange, and do that in a very simple way where, you know, you don't have to be an expert in data centers or networking.

你不必理解那种基础设施工程的细节。但我们能不能真的让它对你变得容易？

You don't have to understand, you know, the details of that type of infrastructure engineering. But can we actually just make that easy for you?

我很好奇你自己怎么用智能体，管公司，也管你的钱。我听你的团队说你非常喜欢 Loops。你在做很多 Loops。这个我们可以谈，但我好奇这几个月你怎么用它、它怎么演变的，因为现在事情走得太快了。

I'm curious how you're using agents in your personal life for running the company, also managing your money, and I heard from your team you really love Loops. You're doing a lot of Loops. We could talk about that, but yeah, curious how you're using it and how it's evolved for you these last few months, 'cause things are moving so fast right now.

我最开始做的就是写代码。我又变回工程师。从那里，我的北极星是，每个人都应该有一支在他们睡觉时仍然有产出的智能体团队。所以我们应该是一家 24 小时运转的公司，而且不只是加总起来，是每个人。

The first thing I started out with was just writing code. I became an engineer again, and, you know, through that, I guess my North Star is that everyone should have a team of agents that's While they sleep. So we should be like a 24/7 company, and not just a, an aggregate, but each person

这有点像你们在交易时间上推的 24 小时。是的。

It kind of mirrors the 24/7 push you're doing on like trading windows and all that. Yeah.

每个人都应该是一个 24 小时的智能体经理。你的工作变成管理你的智能体，确保它们吃饱了、不饿，并且有上下文。你要问，智能体要完全自主地跑，需要什么？它们必须有上下文。它们必须能测试自己的工作。因为如果它们不能测试自己的工作，最终会停下来，说，好了，我做完了，测试交给你，人类。所以这真的要求我们重新想整个基础设施，确保从持续集成、持续部署到测试，这些都非常好，而且能扩展，这样我们才能把几乎所有事自动化。

Each person should be a 24/7 agent manager. Your job becomes managing your agents, make sure they're, you know, well fed and they're not hungry, and they have the context. And you know, you ask, all right, well, what does it take for your agent to be running fully autonomously? They have to have context. They have to be able to test their work. Because if they don't have the ability to test their work, eventually they stop, and they're like, "All right, I'm done. You know, you do the testing for me, human." So it really just requires rethinking our entire infrastructure and making sure everything from, the continuous integration, continuous deployment, the testing, like that's all really good, and it can scale, and that, you know, we can actually automate pretty much everything.

我觉得未来所有产品开发都会以这种方式进循环。你的瓶颈会是，改动能多快拿到真实客户身上测试，以及你多快能判断，这是一个好改动，还是一个该扔掉的坏改动。全行业的软件工程正在彻底改变。你会看到产品开发和基础设施开发的端到端自动化。最早出现的地方之一，其实是 AI 研究。

And I think that the future is all product development will be looped in this way, right? And your bottleneck will be how quickly can your changes be tested against live customers, and how quickly are you gonna be able to tell, is this a good change or a bad change that we should throw away? Industry-wide, software engineering's completely transforming. You're gonna see an end-to-end automation of product development and infrastructure development. One of the first places you're seeing it is actually AI research.

人们以为那会是最难、最后被替代的，因为它是复杂工作。但你仔细想，它的沙箱其实相当干净。AI 模型已经有办法在不上线的情况下测试，我们有评测。所以你有自动化评测。实验在某些方面相当窄。你可以跑实验，改模型，上一个新功能，有时重新训练。所以最早被自动化的工作里，有一部分是 AI 研究。我觉得最终产品也会这样。但产品工作更难一点，因为你必须从真实用户那里获得采用。

People thought that would be the hardest and the last to go because it's just complex work. But really if you think about it, it's pretty neatly sandboxed. I mean, AI models, there's already a way to test them without having them live, but we have evals. So you have automated evals. The experiments are pretty sort of like narrow in some ways. Like you can run experiments, change the AI model, deploy a new feature, retrain it in some cases. So it turns out some of the first work that's being automated is the AI research work, and I think eventually it'll happen with products too. But product work is a little bit harder because you have to go and get adoption from real users.

他们必须在真实环境里测试。人类的采用和行为改变，比纯智能体世界里的事情慢一点。

They have to test it in the wild, and human adoption and behavior change changes a little bit more slowly than, you know, things in the, in the pure agent world.

产品里有一种直觉，真的有一种技艺，我觉得 AI 也很难复制。

And there's an intuition to product and an art really to it that I think also is gonna be hard to replicate by AI.

也许吧。我不知道。人们说过，艺术会是最后被复制的东西。但 ChatGPT 一出来，你最早看到的东西里就有 Stable Diffusion。你可以说，做成梵高的风格。它已经相当好。这当然不意味着它有品味，人们也会习惯，并认出这是 AI 生成的图像。但我觉得很多设计工作也会受益于 AI。

Maybe so. I don't know. I mean, they said, AI would be the last to, art would be the last thing to be, to be replicated, but you know, some of the first things you saw right when ChatGPT launched was stable diffusion. You could say, make this, but in the style of Vincent van Gogh, right? And it was, it was, it's pretty good. now again, that doesn't mean that it has the taste and people kind of get used to things and realize that this is an AI-generated image. But, yeah, I think a lot of the design work benefits from AI as well.

我很想听你对更大范围的 AI 讨论、以及人们感受的看法。人们对它的感觉非常强烈。他们要么爱它，在用它，它在改善生意和生活；要么恨它，厌恶它，拒绝用它。非常两极。数据中心是个问题。问题很多。我在想不平等这个概念，它怎么连到 Robinhood 的使命，把平等带进金融、把金融民主化。我开始觉得，围绕 AI 的很多负面情绪，也许其实和不平等有关。不知道你同不同意。

Yeah. I'm really interested to hear your view on the conversation around AI, bigger picture, and how people feel about it because I was thinking about it's so visceral how people feel about it. They either love it, and they're using it, and it's improving their businesses and their lives, or they hate it. They detest it. They refuse to use it. Very polarizing. And the data centers are a problem. There's a lot of problems. And I was thinking about, you know, this concept of inequality and how it connects to like the Robinhood mission and, you know, bringing equality to finance and democratizing finance, and I wonder if you agree or not that I'm starting to think a lot of the negativity around AI may actually have to do with inequality.

完全同意。这也是我们如此用力推动个人投资者能够投资私营公司的原因之一。我们有 Robinhood Ventures，可以把它想成在美国、在纽交所上市的风险投资基金。我们已经有两只基金完成了 IPO。后面还有。

Completely agree. Yeah. And this is one of the reasons why we push so hard for individual investors being able to invest in private companies. And we have Robinhood Ventures, which is our US-based, think of it as, publicly traded venture funds listed on the, on the NYSE. Yeah, we have two funds that have IPO'd. More are coming as well.

你们现在在第二只了，对吧？我看到你们刚放进了 Crusoe。

You're on the second one now, right? You just had Crusoe on there I saw, and yeah.

对。Crusoe 是 Robinhood Ventures 第一只基金的投资。Robinhood Ventures 第二只基金是种子轮和 A 轮。所以第一只是各自行业里领先的前沿公司，通常是后期。OpenAI 在里面。

Yeah. Crusoe was Robinhood Ventures One investment. Robinhood Ventures Fund Two is seed and Series A stage. So fund one is frontier companies that are leading their industries, typically late stage. so OpenAI's in there.

Databricks。

Databricks.

Databricks，没错。还有 Revolut。都是很难进的顶级后期名字，通过一个像任何股票一样在纽交所交易的产品，提供给大众市场。好的地方在于，因为我们有这么多零售客户，我们直接提供了零售分销。我们上市，募集资本，大部分来自 Robinhood 的个人投资者，然后这笔资本投进这些公司。结构很高效，对吧？中间层非常少，而且 Robinhood Ventures 第一只基金没有 carry。第二只基金，我们想再往前走，做一件以前没人做过的事，就是种子轮和 A 轮的公司。那些在成为家喻户晓的名字之前，没人听说过的公司。

Databricks, exactly. Revolut. So really just top quality late stage names that are hard to get into, available to mass market audience through a product that, is traded on the NYSE like any stock. And the great thing about it is because we have so many retail customers, we provided retail distribution directly. So we went public, we raised capital, the bulk of which came from Robinhood individual investors, and then that capital was invested into these companies. It's an efficient structure, right? There's very little middle layers and no carry for Robinhood Ventures Fund One. Robinhood Ventures Fund Two, we wanted to take it further and do something that nobody's done before, which is seed and Series A stage companies. Companies nobody's heard of before they become household names.

这很让人兴奋，但我现在作为一个新投资者可以说，风险也很大。

Which is exciting but I can say now as a new investor, also very risky.

是的。那些阶段风险最高，但潜在上行和机会也最大。而且有研究。PitchBook 有一项研究，你是投资者，大概看过，种子投资作为一个资产类别，合计回报最高，即使把更高的公司失败率算进去，也高于其他类别。

Yeah. I mean, the risk is highest in those but also the potential upside and the opportunity is greatest. And actually there have been studies. There's a PitchBook study that looked at, you're an investor, so you've probably seen this one, seed investing as an asset class had the highest aggregate returns, even accounting for, a higher rate of company failures than any other Any other class

幂律是件漂亮的事。

The power law's a beautiful thing

是的，这很有意思。我们用一个广泛的一篮子分散，解决了单家公司失败的问题。Robinhood Ventures 第二只基金上市时，我们给组合放进了 80 家公司。投资数量会很大。我们正在造做这件事的机器。当然，海外我们有代币化。我们已经向欧洲客户演示过，把 OpenAI 和 SpaceX 代币化，当作一份礼物。现在链已经出来了，我们显然非常想把代币化股票、股票代币扩展到其他资产。所以你看 AI，一个大问题是，你把它和加密货币对照：加密从一开始就向个人投资者开放。

Yeah, so it's interesting. I mean, we solved the problem of individual company failures by diversifying in a, in a broad-based basket. And when Robinhood Ventures Fund two went public, we seeded the portfolio with 80 companies. So it's gonna be a large, a large, a large number of investments. And you know, we're building the machinery to do this. And of course, overseas we have tokenization, which we've, demonstrated with our, tokenizing OpenAI and SpaceX for our European customers as a gift. And then, you know, now that the chain is out, we're obviously very keen on taking our tokenized stock products, stock tokens, and, expanding that to other assets. So yeah, when you look at AI, a big problem, and you juxtapose it with something like crypto, which was available to individual investors from the beginning.

从比特币区块链的第一个区块起，它就是一个个人的、去中心化的现象。普通人比机构更早采用它。

From the first block of the Bitcoin blockchain, it was like an individual decentralized phenomenon. Normal people adopted it before institutions.

并从中变得非常富有。

Became incredibly wealthy from it

是的。现在你有了加密货币的拥护者群体。只要有一点政府越界的迹象，这个群体就能左右选举，对吧？2024 年的选举，很大程度悬在个人对加密货币的巨大支持上。对照 AI，没有人有敞口。

Yeah, and now, you know, you've got crypto army. if there's a hint of government overreach, the crypto army is, like, swinging elections, right? The 2024 election hung on the balance of huge individual support for crypto. Juxtapose that with AI, you know nobody has exposure.

没人对这些有敞口。

No one has exposure to these

如果后院的数据中心只是让这些风险投资人赚钱，我为什么要它？

Why do I want the data center in my backyard if it's just gonna make you know, these venture capitalists

它只是昂贵、消耗资源、稍微好一点的 Google。你要是这么想，当然会恨它。

It's just expensive, resource-consuming, slightly better Google. And if you think of it that way, of course you're gonna hate it.

是的。我觉得这是一个大问题：人们没有切身利益，也不会替它站台。如果在最早的阶段，这些 AI 公司有更多个人股东，他们会替它辩护，因为那是我的 AI 公司。我想让它更好，想让它扩张、成长，因为我受益，我有切身利益。所以我们这么在意 Robinhood Ventures，以及让 AI 公司在上市之前就把一部分切身利益给个人投资者。私人市场现在资金充裕，这件事被加剧了。公司觉得没那么需要上市，直到估值已经到了万亿美元以上。[提示音]

Yeah. I think, and yeah, I think that it's a big problem with, people not having skin in the game, not rooting. If AI had, if there were more individual owners of these AI companies at the earliest stages, they would be defending it, 'cause it's my AI company. You know, I want it to do better. I want it to, I want it to expand and to grow because I benefit and I have skin in the game. So yeah, that's why we care so much about Robinhood Ventures and our efforts to get AI companies to, even before they go public, give individual investors some skin in the game. And you know, it's exacerbated by the fact that private markets are getting flush with capital. And companies feel less of a need to go public until they're now in trillion dollar plus valuations. [chime]

AI 的用处取决于它拿到的上下文。上下文散落在工具、讨论串和私信里的时候，你的团队和 AI 智能体就是在盲飞。Atlassian 的 Jira 解决的就是这个问题。你的项目绑着什么目标？上周在 Slack 私信里决定了什么？Atlassian 的协作图谱把 Jira、Confluence、GitHub、Slack 等地方真正有价值的碎片拉到一起，不让事情从缝里漏掉。结果准确率提高 44%，token 用量减少 48%。有了 Jira，你可以轻松把工作上下文分享给已经在用的 AI 智能体，比如 Claude、Cursor 和 GitHub Copilot。直接给他们派活，或者通过 MCP 把工具接上。这样你就少花时间在没完没了的链接和消息里挖，追谁做了什么决定，而把更多时间花在真正把东西做出来。

AI is only as useful as the context it has, but when that context is scattered across tools, threads, and DMs, your team and your AI agents are flying blind. That's the problem Jira by Atlassian solves. What's the goal tied to your project? What got decided last week in Slack DMs? Atlassian's teamwork graph pulls all of the valuable pieces together, from Jira, Confluence, GitHub, Slack, and more, so nothing falls through the cracks. You get 44% more accurate results with 48% less token usage. With Jira, you can easily share your work context with the AI agents you already love, like Claude, Cursor, and GitHub Copilot. Assign them work directly or connect your tools through MCP. All of this lets you spend less time digging through endless links and messages, chasing down what got decided and by who, and spend more time actually shipping.

详见 jira.com，网址是 J-I-R-A.com。[提示音] 我花很多时间在会议之间切换，常常上一场还没消化，下一场就开始了。好在 Granola 一直在后台跑。它是好用的会议 AI 记事本，到处都能用，连电话也行。我用 Granola 回想会议上说了什么，并做出有用的摘要。我每天用它掌握自己和团队要做的事。它连着我的邮箱，建议我快速看过再发出的跟进，省下宝贵时间。Granola 不只是我工作流的核心，它基本上是我的第二大脑。到 granola.ai/sources 试用 Granola，用优惠码 SOURCES 可免三个月。[提示音] 运营 Sources 最大的痛之一，是管账。

Learn more at jira.com. That's J-I-R-A dot com. [chime] I spend a lot of time context switching between meetings, often with no time to process one before the next starts. Thankfully, Granola runs in the background the whole time. It's an easy-to-use AI notepad for meetings that works everywhere, even on phone calls. I use Granola to recall what was said in meetings and create helpful summaries. I use it every day to stay on top of what I need to get done with my team. It connects to my email and suggests follow-ups for me to quickly review and send, saving me valuable time. Granola isn't just a core part of my workflow, it's basically my second brain. Try Granola at granola.ai/sources, and use the promo code SOURCES for three months off. [chime] One of the biggest pains I've had running Sources has been managing my books.

我不是会计，市面上的软件都不直观、很笨，而且和交易真正发生的地方脱节，直到现在。Mercury 最近推出了 Mercury Books。这是 AI 驱动的记账，能接上你的 Mercury 银行和信用卡交易，也接外部卡和薪酬系统。你可以选收付实现制或权责发生制，并快速生成现金流、损益表和资产负债表。和 Mercury 做的每件事一样，Mercury Books 的设计非常好上手、很干净。我终于不再被生意的全貌压倒。我喜欢 AI 扛起大部分重活，包括自动给交易分类，也喜欢 Mercury 的指挥 AI 智能体能把任务从头做到尾。

I'm not an accountant, and all the software out there is unintuitive, clunky, and detached from where my transactions actually live, until now. Mercury recently launched Mercury Books. It's AI-powered accounting that works with your Mercury banking and credit card transactions, plus external cards and payroll systems as well. You can choose between cash and accrual accounting, and quickly generate reports for your cash flow, P&L, and balance sheet. Like everything Mercury does, the design of Mercury Books is super approachable and clean. I finally don't feel overwhelmed when I'm trying to understand the big picture of my business. I love that AI does most of the heavy lifting, including for things like auto-categorizing transactions, and that Mercury's command AI agent can handle tasks end to end.

你不必再为单独的会计软件付钱，甚至可以请会计师或 CPA 和你一起用 Mercury Books。这不额外收费，需要的话 Mercury 还会帮你找一位人类会计师。访问 mercury.com/books 了解更多，几分钟内在线申请。Mercury 是金融科技公司，不是 FDIC 承保的银行。银行服务由 Choice Financial Group 和 Column NA 提供，二者均为 FDIC 成员。[提示音] Framer 是驱动 Sources 播客网站 podcast.sources.news 的 AI 建站工具。Framer 把 AI 智能体放进设计和发布网站的同一块画布，让你走得更快，同时不交出品味或控制权。

You don't need to be paying for separate accounting software anymore, and you can even invite your accountant or CPA to work alongside your Mercury Books with you. This doesn't cost extra, and Mercury will even help you find a human accountant if you need one. Visit mercury.com/books to learn more and apply online in minutes. Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column NA, members FDIC. [chime] Framer is the AI website builder that powers the Sources podcast website at podcast.sources.news. Framer brings AI agents into the same canvas where your website is designed, managed, and published, so you can move faster without giving up your taste or control.

我用 Framer 把 Sources 的播客网站做成你能找到这档节目的所有地方的目的地，你也可以在那里看到我通讯的最近几期。今天就到 framer.com/sources 免费开始用智能体建站，Framer Pro 年付计划可享 30% 折扣。framer.com/sources。可能适用规则与限制。

I use Framer to make the Sources podcast website be the destination for everywhere you can find the show, plus you can also see recent issues of my newsletter. Start building with agents for free today at framer.com/sources for 30% off a Framer Pro annual plan. framer.com/sources. Rules and restrictions may apply.

我很好奇你怎么想。我的意思是，我们刚有了 SpaceX，对吧？但我好奇你怎么看 Anthropic 和 OpenAI，看它们接下来要上市。

I'm curious how you're thinking. I mean, we just had SpaceX, right? But I'm curious how you're thinking about Anthropic and OpenAI and looking at those coming down the pike.

当然，我们看到机会，因为我们有这么大的零售客户群，很可以成为散户的声音，并在这些公司生命周期的每个阶段帮助它们。上市前是 Robinhood Ventures。现在我们是 IPO 的承销商。我们宣布了 Oura，这是我们承销的第一笔 IPO。然后——

Well, of course, we see an opportunity because we have, such a large retail customer base that we can very much be the voice of retail, and help these companies at each stage in their life cycle. You know, pre-IPO with Robinhood Ventures. Now we're an underwriter in IPOs. We announced Oura which is the first IPO that we're underwriting. And then

我猜这只是开始。你还想做更多承销？

I assume that's just the beginning. You wanna do a lot more underwriting?

是的。我们打算在这个领域也造成颠覆。问题还在于，很多承销商收费很高。所以我觉得我们可以成为零售分销方面的专家。

We intend to be disruptive in the space, too. The problem is, too, like, a lot of these underwriters charge high fees. So I think we can be kind of like the retail distribution experts.

你怎么打破那个承销小圈子？说到底就是这个。[笑]

How do you break that underwriting cabal, really, is what it is? [laughs]

嗯。不过说句公道话，有些事情——

Yeah. Well, I mean, to be fair, there's some,

它很重要。这是一份重要的工作。我不是想贬低它的重要性。

It's important. It's an important job. I mean, I'm not trying to downplay the importance

是的。有些事我们其实没有很好的位置去做，至少在可预见的未来大概也不会聚焦。我不认为我们有什么特别的竞争优势，去把一大批大型买方机构或主权财富基金拉来当公司的投资者。高盛、摩根大通，这些机构很擅长这个。我们没有同等程度的私人财富客户。但我们有的是大量热爱投资这些公司的普通人，所以我觉得我们可以成为默认的零售选项，并在股东基础里代表散户。我们还有很好的数据。IPO 过程一直非常手工。

Yeah. There's actually some things that we're not really well positioned to do, and we probably won't focus on, at least for the foreseeable future. So I don't think we have any particular competitive advantage going and getting, a bunch of large buy-side institutions or sovereign wealth funds to be connected to the company as investors. So, you know, like, Goldman, JP Morgan, all these other guys, they have. They're great at that. we don't have, like, a private wealth to nearly the same degree. But what we do have is everyday people in large numbers that love investing in these companies, and so I think we can be the default retail option. and we can kind of, represent retail, as a part of the shareholder base. We also have really good data. You know, the IPO process has been very manual.

你作为 CEO 坐下来讨论定价，这东西该定在什么价，很多时候就是凭感觉，手指往上一指。我不知道。Cathie Wood 也许不愿意付那个价。而你想要 Cathie Wood，对吧？这对人们很重要。但我们给发行人的，是每一个价格上精确的需求曲线。这个数据来自整笔发行端到端的自动化。所以通过 IPO 申购和承销，我们可以在那个阶段帮助公司。即便上市之后，通过财报和股东类产品，我们也可以帮他们以这种方式联系自己的股东。

When you sit down as a CEO and, you're having the pricing discussion, where should we price this thing at, a lot of it is just, you know finger in the air. I don't know. You know, Cathie Wood might not be willing to pay that price. And you want Cathie Wood, right? which, you know, is important to people. But what we give issuers is just a precise demand curve at every price, which is data we get because everything is you know, automated end-to-end for that offering. So we, through IPO access and underwriting, we can help companies at that stage. And even post-IPO, through our earnings and, say, shareholder products, we can, we can kind of help them connect with their shareholders that way.

眼下市场里有没有宏观上让你担心的事？美联储三年来第一次加息。10 年期国债接近 5%。通胀的担忧我觉得还在，而且仍然偏高。又有这些巨大的 IPO 要来，一切都感觉脆弱。当然还有中期选举。我很好奇你现在怎么看这些大因素。

Is there anything macro in the market that concerns you right now? I mean, the Fed just raised rates for the first time in three years. The 10-year Treasury is near 5%. Inflation concerns, you know, I think are still there. I think it's still elevated. and we've got these huge IPOs coming, and everything feels fragile. And midterms, obviously. I mean, I'm curious how you're thinking about all these big factors right now.

是的。很多人说，通胀在 3% 以上，或者是 2%？我在共产主义的保加利亚长大，出生在 80 年代末共产主义的最后一段。1991 年是我在保加利亚的最后一年，通胀率是 180%。

Yeah, I mean, a lot of people say, "Well, you've got the inflation rate at above," what is it, 3%? 2%. I grew up in, communist Bulgaria, so I was born, I was born during the final swan song of communism in the, in the late '80s. And, 1991, which was my last year in Bulgaria, the inflation rate was 180%, right? So,

还没到那个地步。

Not quite there yet.

是的，那很艰难。生活还是继续。1997 年，我搬到美国几年之后，保加利亚不幸成为全世界通胀最高的地方。是几千个百分点，他们 真的在往钞票上加零。票印得不够快，票面上需要更多的零。你会非常真切地感到整个系统的脆弱。眼下这个局面，离那种情况差得远。再说宏观，利率上升，利率下降。这个国家有很多工具，有很多出色的创新。美国经济的基本面相当强。原料都在。我有理由更乐观，而不是更悲观。

Yeah. So that was rough. And, you know, life went on. In 1997, which was a couple years after I moved to the US, Bulgaria had the unfortunate distinction of the highest inflation rate in the world. It was in the thousands of percent, so it was literally they were adding zeros. They couldn't print the bills fast enough. They needed more zeros, in the bills. And I think, I think it was very, like you viscerally felt the fragility of the entire system. This is, like, nowhere remotely close to that. so yeah. And again, like the market, the macro, rates go up, rates go down. I think we have a lot of tools in this country, a lot of great innovation. the fundamentals of the economy in America are quite strong. The ingredients are there. I have reasons to be more optimistic than pessimistic. So yeah.

听到这个很好。

That's good to hear.

我现在感觉相当好。你看 AI 产业的中心，以及推动这么多创新的东西，它仍在这里。不在欧洲。还没有搬到中国。

I feel pretty good now. I mean, you look at the center of the AI industry and what's driving so much of this innovation, it's still here. It's not in Europe. Hasn't moved to China.

我也在想，我们是不是低估了 AI 的通缩一面，以及它能打开什么。尤其是当你开始进入机器人，以及 AI 在物理世界里做更多事。劳动力成本，这些东西。我们甚至还没有一个感觉。

And I do wonder how much we're maybe underwriting the deflationary aspect of AI and what that could unlock, right? Especially when you start getting into robotics and all these things, and AI doing more things in the physical world. Cost of labor, all these things. Like, we don't even really have a sense of that yet.

还有，你得把另一件事算进去：很长一段时间，我们只是在把产业出口出去。但现在有了 AI，我们在这里建数据中心。作为一个国家，我们在认真想怎么在这里获得更多能源。我们该不该建更多核电，真正利用那种创新？我觉得这些其实是长期的正面因素，因为我们必须想办法打破那种让我们什么都建不了的泥潭，而现在势头很大。政府这边出来的很多事情也非常正面。我们在做的另一件事是特朗普账户。

Well, and also you gotta, you gotta factor in that, you know, for a long time we were just exporting our industry. But now with AI, we're building data centers here. we're trying to really figure out as a country how do we get more energy here. Should we be building more nuclear and really taking advantage of that, of that innovation? And I think those are actually long-term positive things, because we have to figure out how to, you know, break the morass that got us stuck to a place where we couldn't build anything, and now there's just, like, huge momentum. yeah, a lot, a lot of the things coming out of the administration have been very positive, too. One of the other things we've been working on is the Trump accounts.

对，是的。

That's right, yeah.

Robinhood 做很多事。我们在做智能体交易。我们有 Robinhood Ventures。我们在做代币化。这一切都围绕一个我认为非常重要的想法：所有权。对高质量金融资产的所有权极其重要。不只是对个人重要，意思是我能赚更多钱、我有切身利益、我能保障财务未来。这些都很好，但它还有社会层面的好处：一个有更多个人成为所有者的社会，是一个更稳定的社会。人们对身边产业的成功有一份利害，对国家的成功有一份利害。所以我们所有产品都通向这个想法。

Robinhood does a lot of things. We're doing agentic trading. We have Robinhood Ventures. We're doing the tokenization. all of it goes around this one idea, which I think is a very important one, it's ownership. Right? And I think ownership of high quality financial assets is incredibly important, right? And, not just important for the individual in the sense of I can make more money, I have more skin in the game, secure my financial future. That's all great, but there's a societal benefit to it as well, which is that a society where more individuals are owners, is a more stable society, right? I mean, people have a stake in the success of the industries around them. They have, they have a stake in the success of the country. And so all of our products really ladder to this idea.

私人市场的所有权，那是 Robinhood Ventures。全世界的所有权，那是代币化。特朗普账户是另一件和政府有关的事：从出生就开始的所有权。你怎么让一个人从零岁就立刻接进我们的经济，从而获得最大的好处？他们能比正常情况下多享受几十年的复利。所以这大概会是这届政府最重要的遗产之一。这个计划有希望永久延续下去，并带来更稳定的投资者基础，以及一个更稳定的国家。

Ownership of private markets, that's Robinhood ventures. Ownership all over the world, that's tokenization. Trump accounts, one of the other things, talking about the administration. Ownership from birth. How do you get someone to be immediately plugged into our economy from age zero so that they have the largest benefit? They get to benefit from compounding for decades more than they would have under normal circumstances. So, that's probably gonna be one of the most consequential things that, most consequential legacies of this administration. Like, this program will survive, hopefully in perpetuity, and lead to a more stable investor base and a more stable country.

我觉得你、Brad Garlinghouse，以及所有参与的人，都值得很多肯定。我同意你，显然也同意政府这边。但我很高兴硅谷的公司真的站出来，帮了这件事。而且这个应用很好，对吧？体验不错。

I think you and Brad Garlinghouse and everyone who's involved in that deserve a lot of credit for that. Yeah, I think it's. I do agree with you, and the administration, obviously. But, like, I'm glad that, the companies in Silicon Valley and really stepped up with this and helped. And the app's great, right? It's a good experience.

你喜欢就好。我们为此非常自豪，也有一点不服气，因为在某种意义上，这是我们为国家服务的方式。显然，这不是拿起武器去打仗，我不想做成那种比较。但我们在这里做的事，影响可能是一代人的。另外，政府并不以做出好产品著称。事实上，我们多少都被 healthcare.gov 这类东西伤过。

I'm glad you like it. Yeah, I mean, we, I think we take a large amount of pride, and we have a little bit of a chip in our shoulder, 'cause this is our way of, like, serving our country in a sense, right? And it's, obviously, it's not like we're, picking up, you know, arms and going to battle, right? So I don't wanna compare it to that, but, you know, the impact of what we do here could be generational. And also, the government is not known for building great products. In fact, we're all sort of like, scarred to some degree by things like healthcare.gov and

还有车管所，这些都是。

The DMV, I mean, all, yeah.

是的。不过我最近去了红木城的车管所，流程顺得让我印象很深。

Yeah. Although I went to the DMV recently in Redwood City, and I was, like, very impressed at how smooth it was, so.

你去了车管所？

You went to the DMV?

我去了车管所，然后意识到，我得把自己的标准提高。我以前觉得政府还在石器时代，但事情已经变得非常顺。

I went to the DMV, and then I realized, holy shit, I gotta, I gotta up my game. I had this view that government was kind of in the Stone Age, but things have gotten very smooth.

红木城的车管所挺像样。

The Redwood DMV's got it together.

我得办 Real ID。我去排队，很快就办好了。

I had to get my real ID and, you know. I went and stood in line and got my real ID really quickly.

向车管所致敬。我没想到。

Shout out DMV. I did not expect that.

所以如果你最近没去过，我强烈推荐。去看看你们当地的车管所，他们也在努力变好。

So if you haven't been recently, I highly recommend it. Just try to, try to pay your local DMV a visit and, they're trying to get better too.

好。[笑] 我们一直在绕，也许可以在这里落地。我不敢相信我们现在才谈到这个。你整场都在说代币化。我得问，你在搅这锅水，你知道吧？AMC 那场争执，公司们的反应非常强烈。我觉得你们刚刚——

Okay. [laughs] Well, we've been dancing around this, and maybe we can land it here, and I can't believe we're actually just now getting to this. You've been saying tokenization throughout this conversation. I gotta ask, man, like, what. You're stirring the pot with this, you know? Like, the AMC feud, companies have been reacting very viscerally to this. I think you all just

其实只有一家公司。

Well, there's one company.

一家，但确实。你们在私募那边也有过一点，对吧？OpenAI 和其他公司有过反应。而且就在我们录这期的几天前，你们在美国拿下了一点进展，创新豁免，对吧？它开始把美国这边打开一点。但我还是想明白，你为什么推这个？我们为什么需要把股票代币化？

One, but yeah. You had it on the private side a little bit, right? You had some reactions to that, OpenAI and others. and I think you all just notched a bit of a win, as we're recording this a few days ago here in the US with the innovation exemption, right? Which starts to unlock the US a little bit. But I'm still trying to wrap my head around, like, why are you pushing on this? Like, why do we need to tokenize stocks?

主要原因是，如果你在海外，拿到美国股票其实非常难。有些市场相对容易，它们有健全的本地资本市场和交易所。如果你在南美的很多地方，就是难。拿到美元也曾很难。有一段时间，比如在一个本币恶性通胀的南美国家，那些人在买美国运通的旅行支票。这就是稳定币变得这么流行的原因。稳定币的大量使用来自海外，来自那些只是想要一条更好、更便宜的路去拿到美元的人，因为他们要把钱放在美元里。同样，很多人想拿到美国股票。那是他们听说过的公司。

Yeah. So the main reason is that it's actually really hard to get access to US stocks if you're overseas. And, you know, in some, in some markets it's relatively easy, and they have robust local capital markets and exchanges. If you're in South America, in a lot of places, it's just tough, right? And it was also tough to get dollars. For a while, you know, if you're in a South American country, for example, with a hyperinflating local currency, those people were buying American Express traveler's checks. Right? And that's why stablecoins became so popular. A lot of the usage of stablecoins was overseas from people that just wanted a better, cheaper way to get access to US dollars, because they would keep their money in it. So in the same way, a lot of people want access to US stocks. It's companies that they've heard of.

有 SpaceX，有英伟达，还有很多。代币化让他们能拿到，而且有一些更好的属性：全天候，可以拆碎，你可以把代币拆成很小的块，不必买一整股。它可以在你的钱包里，特别是，没有人能把它从你这里没收走。它不在某个本地账户、某个也许没有我们美国这里这么强保护的金融体系里。所以某种意义上，我觉得公司会非常高兴，因为它从全球市场带来对它们股票的净新增买入。Robinhood Chain 和我们的股票代币已经在 120 多个国家上线。那些地方的人本来大概不会买这些美国股票。他们没有一条容易的路径。

Got SpaceX, you've got Nvidia, so many others. And tokenization, allows them to have access with actually some better properties, 24/7, fractionalizable, so you can break up the token into little itty-bitty chunks. You don't have to buy a full share. it can be in your wallet, so in particular, nobody can confiscate it from you. It's not in some, you know, local account at some financial system that maybe isn't as developed with as strong protections as we have here in the US. So in a sense, I think the companies will be very happy about this, 'cause it creates net new buying of their stock from, a global market. Robinhood Chain, with our stock tokens, are live in over 120 countries, where, you know, people probably wouldn't have been buying these US stocks. They don't really have an easy mechanism.

所以它把这件事彻底打开，打开了一个以前不存在的、全新的买家资金池。

So it just cracks that wide open and opens up an entirely new capital pool of buyers that didn't exist before.

但然后你有 AMC 的 CEO 和其他人在反击。我觉得有一种反驳是公平的。我不知道，你被问过这个问题——

But then you've got the AMC's CEO and others pushing back, and I do think there's a fair pushback that. I don't know, you've gotten asked this question

但你看到他昨天改口了，对吧？

Well, but you kind of saw he reversed yesterday, right?

是的。他……好吧，确实。但是——

Yeah. He. Well, yeah, sure. But

我觉得 AMC 这件事，你得考虑到里面有 meme 式的热闹。而且我们和 AMC 有一点可以追溯到 2021 年的历史，有点特殊。所以我其实不确定，其中有多少是对代币化到底是什么的困惑。因为这是一种复杂的新技术。你可以想象最好的情况是，我在经营我的生意，我为什么要去对付资本市场周围正在变化的技术？这不在我的优先列表上。我在努力卖电影票和爆米花，对吧？或者，其中有多少只是在迎合一群散户，他们一看到有人吵架就会兴奋。

I think a lot of the AMC thing, you have to take into account that, - There's like meme activity. And there's, we have a little bit of a history with AMC dating back to 2021 that's a little bit idiosyncratic. So I'm actually not sure how much of it is confusion about what tokenization exactly is. Because this is like a complicated new technology, and, you know, you can imagine best case scenario, I'm running my business, why do I have to deal with like changing technologies around capital markets? It's like not high on my list. I'm trying to sell movie tickets and popcorn, right? Or also, how much of it is just, you know, appealing to a retail base that, you know, gets excited when there's just people fighting.

但我觉得我听到的公平反驳是投票权。你并不真正拥有那只股票，治理这一面怎么解决？

But the, I think the fair pushback I've heard is the voting. How do you solve the governance aspect of not actually owning the stock?

我们有解决方案。我们宣布打算加上投票。现实是，也没有多少人会给自己的股票投票。

Oh, yeah. I mean, we have a solution for that. I mean, we announced that we intend to add voting. The reality too is not a lot of people vote their shares.

他们不会。但我觉得这像是一个要勾上的格子：我确实拥有它，我确实以这种方式有敞口。

No, they don't. But I think it's like, it's a nice box check to go like I actually have it, I actually have exposure that way.

如果你在委内瑞拉，想获得美国股票的敞口，投票权是好的，但大概不是更高阶的那一件。所以你可以找出各种小挑剔。人们对股票的电子化交易也有过同样的话。他们说，我拿不到真正的纸质凭证。有一张真正的凭证很好。那才是真正的股份。Cede and Company 是什么？它甚至不在我的名下，是街名持有。股票开始以电子方式交易和经纪、而不再是纸质凭证的时候，这些同样的批评都出现过。所以我觉得木已成舟，这不可逆。现在的问题只是，我们怎么让这些发行人、利益相关方，以及美国的交易所感到舒服？

If you're in like Venezuela and you're looking to, you know, get exposure to US stocks, voting's nice, but it's probably not the higher order bit, you know? So I think, yeah, you can find all sorts of little nitpicks. People had this about electronic trading of stocks as well. They're like, "Well, I don't get the real certificate. It's really nice to have the real certificate. That's a real share. What is this Cede and Company thing? It's not even in my name, it's in street name." So all of these same criticisms actually landed when stocks started being traded and brokered electronically rather than physical certificates. So I think the cat's out of the bag, like the, this is not reversible. Now it's just a question of how do we make all of these issuers, and stakeholders and, you know, the exchanges in the US comfortable?

我们怎么让他们觉得这对他们是好事？并且理解，因为你看，如果你是发行人，比如 AMC 或其他公司，这基本上全是上行，几乎没有下行。所以很清楚，这会成为一件事。创新和变化从来不会被一致接受，但它也是不可避免的。我们在和所有利益相关方一起做，确保每个人都被带着走。但它正在发生。现在 SEC 站在代币化这边。我们从欧洲开始，一部分原因是，我们在法国南部的活动上说，我们要为国际客户交付好产品。我们立刻在增加价值，我们在交付产品。

How do we make them feel that this is like, this is good for them? And understand, because actually if you look at this, if you're an issuer, like a AMC or others, it's basically all upside, next to no downside. so it's just very clear that it's gonna be a thing, and it's innovation and change is never just uniformly accepted, but it's also inevitable. So, yeah, and we're working with all the stakeholders to make sure everyone's kind of brought along. But, yeah, it's, it's happening. So, yeah. And now the SEC is behind tokenization. So, yeah. And it, part of the reason we actually started with Europe, and when we were in our event in south of France, we said, "We wanna deliver great products for our international customers." Like that's, we're immediately adding value, we're shipping product.

我们不想只是谈论代币化。有些人喜欢谈，它有点变成一个时髦词，他们接受采访之类。但我们交付。我们想交付真正的产品，交到人们手里，那才是最有说服力的。

We don't wanna just like talk about tokenization. You know, some people like to talk about it's kind of become a buzzword and, you know, they have interviews and stuff like that. But we ship. We wanna ship real product, put it in people's hands, and that's really the most persuasive thing.

因为我可以告诉你代币化是未来，但你也许不信。那我把它做出来呢？人们用它。我们看到它在证明价值。如果它给美国以外的这些客户增加了价值，也许也能向这里的人证明，我们应该采用这项技术，并找出一条让它在这些市场里行得通的路。我觉得基本上就是这么发生的。很多年前我们就把话说在前面了，我们在尽所能把行业往前推，不只是海外，这里也是。我告诉你未来。它会发生在每一种资产上。私人市场会来。我们会有私募的代币化。这是必然。任何一个富人的资产组合里有的，我们都会有。葡萄酒的代币化，房地产的代币化，对吧？

'Cause, yeah, I could tell you tokenization's the future, but, maybe you won't believe me. How about I ship it? People use it. We see it demonstrating value and if it adds value to, these customers outside the US, maybe that also demonstrates to folks here that, hey, we should, we should adopt this technology and figure out a path for it to work in these markets. And I think that's basically what's happened. I mean, we kind of called our shot many years ago and we're doing what we can to kind of push the industry forward, not just overseas, but here as well. And I'll tell you the future. It's gonna happen to every asset. Private markets coming. You'll be able to. we'll have tokenization of privates. It's an inevitability. We'll have any wealthy person's asset base. We'll have tokenization of wine, of real estate, right?

它们都会像股票一样，在无需许可的网络上 全天候交易。从技术上看，这会很酷。对一个在海外建组合的人来说，能以低成本拿到一个美国高净值个人能拿到的任何资产。

It'll all trade 24/7, on permissionless networks, like a stock. Yeah, and I think it'll be really cool. I mean, from a technical perspective, it's interesting. But, yeah, for someone building a portfolio overseas to have low cost access to any asset that a high net worth individual here in the US would be able to have.

我想跟你确认预测市场的现状，尤其是走向中期选举。我觉得 Kalshi 和 Polymarket 在观感上打得很厉害，Robinhood 是一个很大的玩家。你对预测市场一直很乐观，我们以前谈过，你说它是新闻的未来。你是我听到的第一个说“我把它看成新闻的未来”的人。你预期会怎样，尤其是中期选举会是一个大时刻。每次大选举都是。但未来几个月，你怎么准备，怎么看这个品类？

I wanted to check in with you on the state of prediction markets, especially going into the midterms. I think you've got Kalshi and Polymarket kind of battling out in the court of perception, and Robinhood is a huge player. And you've been very bullish on prediction markets as we've talked about this in the past, the future of news. You were the first person I actually ever heard that said that, like, "I think of this as the future of news." What are you expecting kind of, especially, I think the midterms will be a big moment. Every big election is. But how are you gearing up and how are you thinking about that category over the next several months?

我们在做一个很好的产品。预测市场在 Robinhood 应用里，基础设施这边我们也做了很多。但作为客户、作为 Robinhood 应用里的用户，它已经好了很多。现在我们为所有主要类别做了专门的体验和中心。世界杯，我们有一个很好的世界杯体验，所有信息在一个地方。金靴奖、最佳射手，各个小组都有。中期选举有一个很好的原生地图体验。如果你在交易，它显然有用、也好用。但如果你只是想看中期选举进行得怎么样，印第安纳州谁会赢，也可以。我们那里的体验不错。

Well, we're building a great product. Prediction markets are in the Robinhood app, and we've done a lot of stuff on the infrastructure side too. But as a customer, as a user in the Robinhood app, it's gotten way better. And, you know, now we have bespoke experiences and hubs for all the major categories. World Cup, we had a really nice World Cup experience where we had all the information in one place. You know, Golden Boot winner, top goal scorer, we had the different groups. Midterms, there's like a really nice native map experience where if you're trading, it's obviously useful and easy. But also, if you just wanted to see, like, how are the midterms going? Who's gonna win the state of Indiana? You know? And, yeah, we have a nice experience there.

所以，要改进产品和基础设施，工作量永远做不完。我们看到新的类别在出现。加密市场非常受欢迎。所以我们开始看到预测市场和其他资产之间的连接和协同，比如加密货币，股票这边后面也会更多。你看到这个行业在成熟，变成一个真正可交易的资产类别，而且有很大的零售成分。

So, it's constantly, the amount of work that we have to do to improve the product and improve the infrastructure is never-ending, right? - And we're seeing new types of categories come up. Crypto markets have been very popular. And so increasingly we're starting to see connections and synergies between our prediction markets offerings and, different assets like crypto and we'll start to see more with stocks as well. You're seeing the maturation of the industry into a true tradable asset class, and with like a huge retail component.

这次中期选举，你甚至不会开着电视跟进展吗？你就打算——

Are you not even gonna have the TV on this midterm in terms of trying to follow things? Are you just gonna be

我觉得我不会开很久的电视。我觉得那已经过去了。

I don't think I'm gonna have the TV on very much. Yeah. I think, I think that's gone away.

我们已经在离开那种方式，这不是很野吗？

Isn't it a wild concept that we're already moving away from that?

是的。即便先不谈预测市场和跟选举，有一个让我觉得“事情在变”的时刻，是几个月前我想看 NBA 季后赛。我习惯换台。好，它会在 ABC 或 ESPN。然后我说，这场比赛在哪？我找不到。然后我问了我的 AI 智能体，我个人的 AI 智能体。它说，你得去 Amazon Prime。我就想，当 NBA 季后赛出现在 Amazon Prime 上，我觉得线性有线电视到头了。我自己还有有线。但我觉得它很快会变得像固定电话。

Yeah, I mean, I just realized even aside from prediction markets and following the elections, I think a big, moment to me where I was like, "Wow, things are changing," is when, I was trying to watch the NBA playoffs a couple months ago. And I'm used to just switching the channels. All right, it'll be on ABC or ESPN, and I was like, "Where are this where's this game? I can't find it." And then, I went and I asked, my AI agent, my personal AI agent. It's like, "Oh, you gotta go to Amazon Prime." So I'm like, man, okay. When the NBA playoffs are hitting Amazon Prime, I think, I think it's the end of the line for linear cable. I still have mine. but yeah, I feel like it's gonna be like a landline phone before too long.

我同意。最后一个问题。我碰巧有机会见到 Tudor，你在 AI 实验室 Harmonic 的联合创始人。他们在做很多很有意思的事，而且是从数学的角度看。在建基础模型。我不剧透即将到来的东西，但他给了我一点预告，非常有意思。他给了我一个问题来问你。他说，你怎么看 AI 公司用自己的平台去解数学，而不是把工具交给其他所有人去做？

I agree. Last question. I fortuitously had the chance to catch up with Tudor, your co-founder at your AI lab Harmonic, which is doing a lot of really interesting stuff, and approaching it from the view of math. Building foundational models and, I won't spoil what's coming, but he gave me a tease and it's very interesting. He gave me a question to ask you. Yeah. He said, "What do you think of AI companies using their platform to solve math versus giving the tools to everyone else to do it?"

这对我简直是一个轻松球。我喜欢这个问题。我本来就打算主动谈。我们谈了很多，问题之一是普通人没法投资这些公司。这带来了所有这些负面效果，AI 的领导者会说，为什么所有人都在生我的气？我只是在经营我的生意。你的股东基础，不是你的错，但他们中的一些人——OpenAI 是个好例子。他们和我们合作，让个人可以投资。但大多数人并不真的想这件事。大多数人想的是，我怎么把生意尽可能快、尽可能大地做起来，并打败竞争对手。

Oh gosh, what a, what a softball to me. I love that question. I was gonna bring this up proactively, right? so we talked a lot about one of the problems being that normal people just don't have access to invest in these companies. And that's led to all these negative effects where the AI leaders are like, "Why is everyone mad at me?" You know, "I'm just running my business." Well, I mean, your shareholder base, through no fault of your own, but, you know, some of them, actually OpenAI great example. They partnered with us and, made investing available for individuals. but yeah, like most people don't really think about this and most people are very much thinking about, "All right, how do I grow my business as fast and big as possible and beat my competitors," right?

但对 AI 这样的东西，你必须明白，到头来你需要一张经营许可。如果人们反对你的技术和你的公司，会非常非常难。我觉得有一件事是缺的，坦白说。我也能看出他们是怎么走到那里的：AI 公司把这些先进模型留在内部，自己做发现。他们的员工跑这些模型来做发现。

But with something like AI, you have to understand that you have to have, a license to operate at the end of the day. Like, if the people are against your technology and your company, it's gonna be very difficult. And I think one thing that is just, amiss, frankly, is. And I can kind of see how they got there, is AI companies keeping these advanced models in-house and just making the discoveries. Their employees running them to make discoveries. And

OpenAI 刚刚解出了千禧年难题，诸如此类。

OpenAI just solved the millennium problem and all of that.

我觉得有一种未来是，人们社区里立着巨大的数据中心，只是在跑 AI 公司员工的工作负载，发现全由 AI 公司做，发明全由他们创造，也许药也由他们研制和生产。那是一个非常集中、几乎反乌托邦的未来。出于很多原因，我觉得这是可能性更低的未来。更鼓舞人、也更可能的一种，是把工具造出来，让普通人能用到。所以如果是一位数学家在用公开的 OpenAI 模型做出这些发现，而不是只由员工跑着、花掉几千万美元的算力，结果会好得多。

Like, I think the future where there's just these giant data centers in, in people's neighborhoods that are just running workloads from the employees of the AI companies, and the AI companies are making all the discoveries and creating all the inventions and, oh, maybe they're creating the drugs and producing them, right? That's like a very centralized, almost dystopian future. And I think for a multitude of reasons, it's the less likely future. I think the more, the more inspiring one is building the tools so that normal people get access to them. So I think it would've been a much better outcome if it was, you know, a mathematician that was using the OpenAI model that was, you know, that was coming up, that was coming up with these discoveries and not just employees running it and spending, what was it? Tens of millions in compute.

所以这是我相信的未来，也是我认为更可能的那个。

So yeah, I think, I think that's the future I believe in, and that's I think the more likely one.

公开可用的。好，谢谢你的时间。

Publicly available. Well, thank you for your time. Appreciate it.

永远是件愉快的事。

Always a pleasure.

银行服务应该有现代软件的感觉。你需要的一切放在一个地方。访问 mercury.com 了解更多，几分钟内在线申请。Mercury 是金融科技公司，不是银行。详情见节目说明。Granola 是我用过最好的 AI 记事本。它到处都能用，视频、电话、当面，或者 Apple Watch。现在到 granola.ai/sources 试用，结账时用优惠码 sources 可免三个月。Atlassian 的 Jira 是你的团队和智能体共享同一份上下文的地方。在 jira.com 免费试用，网址是 J-I-R-A.com。Framer 是 AI 原生的网站构建器，让你建得更快，同时不交出控制权。访问 framer.com/sources 可享 30% 折扣。可能适用规则与限制。

Banking should feel like modern software. Get everything you need in one place. Visit mercury.com to learn more and apply online in minutes. Mercury is a fintech, not a bank. Check the show notes for details. Granola is the best AI notepad I've tried. It works everywhere, on a video or phone call, in person, or an Apple Watch. Try it now at granola.ai/sources and use the promo code sources at checkout for three months off. Jira by Atlassian is where your team and your agents work from the same context. Try it free at jira.com. That's J-I-R-A.com. Framer is the AI native website builder that lets you build faster without giving up control. Visit framer.com/sources for 30% off. Rules and restrictions may apply.