重塑交易世界的 Citadel 校友
- 核心产品主张是：Lighter 专门为金融构建了定制 ZK 电路，而不是改造通用电路；证明这些电路所需的算力“实际上只有通用计算机所需算力的1%”。 Vlad 对这一机会的判断是：金融占经济的30-40%，所以“想象一下，你找到一种方法，只用1%的成本运行40%的经济”。他称，与“几乎所有竞争对手”相比，Lighter 在低成本、低延迟、可验证性和安全性这几个维度上“都更强”，同时也认可早期参与者相较此前带来了“阶跃式变化”。
- Vlad 的判断是，排名前5至前10的去中心化交易场所对大多数客户而言已经强于中心化交易所。 他认为大部分迁移将发生在这里——从 CeFi 转向 DeFi。
- Lighter 的价值归属从设计上就没有歧义。 Lighter 从第一天起就是美国 C corporation，没有双基金会架构；代币发行前融资轮已承诺是“我们永远进行的最后一轮股权融资”，尽管该轮超额认购5倍，最终也只有股权表中1%的份额选择退出。所谓“所有价值都归属于代币”，主持人总结为股权端不获得收入或利润，Vlad 回答“Okay”；在 Clarity Act 通过且 SEC 继续推进创新的前提下，理想终局是代币也同时代表股权。
- 2004年，Vlad 加入 Citadel 的第一周就指出 Madoff 的可疑高比率（很可能是 Sharpe 比率），Ken Griffin 回应：“哦，不，那是个庞氏骗局，别担心”（"Oh, no, that's a Ponzi. Don't worry about that one"）。 这比 Madoff 案件被揭穿早了5年。Madoff 通过交出一个从未有人拨打核实的虚假 DTCC 编号躲过 SEC 核查；而在链上，“所有这些交易都会记录在链上，任何人都可以验证”。
- 路线图是在26年Q3推出真正的链上期权。 期权将在 Lighter 内部与永续合约和现货共用一套资产负债表及风险抵押品模型；公司希望用户可以通过 Robinhood 交易这些期权，Vlad 将这段合作归因于技术尽调、Ethereum 互操作性以及前置部署工程师。
- 值得注意的意外是，ZK 本可以实现暗池和交易隐私，但 Vlad 说：“我其实很意外，市场需求没有我原本以为的那么大”。 Lighter 使用 ZK 主要是为了扩展性；隐私“还没有成为客户经常提出的需求”。
- 创始团队层面的机制是，Lunch Club 转向 Lighter 后通过“一次内部 YC”保住了80%的工程团队。 做法是让3个项目并行推进，再举办内部 demo day 和路演比赛；叠加充足的资金储备，以及团队真正偏好的问题领域——密码学、扩展性系统和量化金融——最终实现了产品彻底转型下的工程团队留存。
1. 16岁进哈佛的奥赛少年——班级平均分37，他考了94分
- Vlad 8岁就开始编程，用 QBasic 和 Pascal 编写机器人，练习 Durak 等纸牌游戏，好在家人面前赢牌；但他说，12岁发现的竞赛才真正让他找到突破口：“直到接触这些竞赛……我才真正知道自己有竞争性”，你会得到“非常清晰的反馈闭环”，可以练习并不断进步。竞赛也是他最早加入的社群，虽然竞争激烈，但协作性强得多。
- 按他的讲述，哈佛专业选择本质上是一次值得复盘的组合决策：编程能力“已经开始出现边际收益递减”，所以大学阶段要学的是“该做什么”；物理学获得了大量资金支持，但未必处于快速增长期；生物学意味着他不喜欢的实验室工作；经济学则“有很多令人兴奋的事情正在发生”，包括期权定价和投资组合理论，而他的数学与编程能力都能派上用场。
- 他的第一份工作源于一场微观经济学期中考试：他考了94分，原以为只是差一点拿到 A——“高中只看绝对分数”——结果发现班级平均分只有37分，在这门课历史上属于标准差最高的成绩之一。教授当场就把他招进了研究项目。他在2年半内完成了哈佛学业。
- 他把自己意外进入的那张人脉网络形容为：“如果阿甘是个数学家，大概就是这种感觉。”其中包括信息学奥赛伙伴 Adam D'Angelo——Zuckerberg 的高中室友、Facebook 首任 CTO，如今可能在 OpenAI 董事会任职；2位物理夏令营同学，他们后来共同创办了 Anthropic；以及一位高中校友，他认为很可能就是 Robinhood 的 Vlad Tenev。
2. Ken Griffin 在他18岁时说服他加入——并于2004年称 Madoff 是庞氏骗局
- 关于招聘传闻，Vlad 说：“我不会说是他亲自招募了我，但确实是他亲自把我说服了。”Griffin 的说服方式非常直接：Vlad 在大学期间自行交易的期权策略“确实有效了很多年，但我已经把它们套利掉了，所以那套东西现在不行了。你应该来和我们一起工作。”当时正值市场刚转向电子化，Citadel 正在搭建最初的交易台。
- 这期节目的最佳故事发生在他入职的第一周：Vlad 在审阅竞争对手的业绩记录时，问既然 Madoff 的比率异常突出，为什么不去反向拆解他的策略。Ken 回答：“哦，不，那是个庞氏骗局，别担心。”（"Oh, no, that's a Ponzi. Don't worry about that one"）这比 Madoff 案件被揭穿早了5年。SEC 确实核查过 Madoff，但他交出了一个虚假的 DTCC 编号，“他们从来没有打那个电话。如果他们拿起电话，事情当时就会被揭穿。”
- 这构成了 Vlad 为链上金融提出的完整论据：Madoff 根本没有进行交易，而案件能否被发现“完全取决于有没有人拿起电话”。在链上，“如果你在30美元买入 Microsoft、40美元卖出并因此赚了10%，所有这些交易都会记录在链上，任何人都可以验证”。
- 主持人追问：人们并不希望自己的交易公开。Vlad 坦承，ZK 证明可以在不暴露交易内容的情况下完成验证，暗池也完全可以构建出来，“但我确实没想到，市场对这类需求没有我原本预期的那么大”。Lighter 使用 ZK 主要是为了扩展性；隐私“还没有成为客户经常提出的需求”，而如今在链上交易的机构“在那个环境中仍然拥有大量 alpha”。
3. 硅谷绕行，以及那些曾经逆势、后来成了常识的经验
- 他离开华尔街，是因为量化交易虽然有趣，却意味着“我们并没有真正打造新产品和新市场”，而 D'Angelo 等朋友正在这么做。之后他先在 Quora 担任机器学习负责人，再在 Joe Lonsdale 旗下的 Addepar 负责工程，随后与另一位奥赛校友 Scott Wu 共同创办 Lunch Club。
- 他见证了2个从荒谬走向显而易见的想法：Lonsdale 提出的前置部署工程师，把工程师作为客户成功团队的尖兵；以及 Robinhood 的零佣金零售产品——“当时这被认为是疯狂的”。他的更高层判断是：好想法在于“逆势而行，而且最终证明是正确的——之后它就不再逆势了”；但“每天可能有10,000个人都在学这套框架，因为他们可能才18岁”。
- 面对精英圈层的问题，他说：“这是一个大俱乐部，但它是一个开放的俱乐部……也更看重能力。如果你真的擅长客户成功、运营，欢迎加入这个俱乐部。”
4. 转型打法：一次内部 YC 保住了80%的工程团队
- 董事会批准转型后，Lunch Club 没有直接拍板定下方向，而是运行“一次内部 YC”：让3个项目并行推进，举办内部 demo day 和路演比赛，确保“团队里经验最少的人也有机会决定公司的走向”。再加上充足的资金储备——“不是只剩最后3个月的资金”——最终在产品彻底改变的情况下，保住了80%的工程团队。
- 第三个留人因素更能说明问题：团队加入 Lunch Club 是因为看重那里的人，但 Lighter 所处的问题领域——密码学、扩展性系统、量化金融——才是他们真正想做的事情。
- 对于“没人关心技术，一切都是分发”的说法，他的反驳是：分发和用户体验都依赖技术。Telegram Wallet 和 Robinhood “绝对对技术做过尽调”；即便用户永远不会知道执行延迟被压低了100毫秒，也“绝对在意这件事确实发生”。Bitcoin 是同一个逻辑：很少用户真正理解它，“但他们知道它确实有效，也知道如果它出了问题，总会有人找出错误”。
5. Lighter 的优势：用1%的算力运行40%的经济
- 从工程角度看，交易所是一个3维或4维优化问题：低成本、低延迟、可验证和安全。不同指标之间通常需要取舍，但“有时你会找到一些非常酷的想法，让所有指标都得到改善……代码实际上更便宜、更快、更安全。这才是真正让工程师兴奋的东西”。
- 突破口在于，他们意识到可以为金融构建定制 ZK 电路，而不是改造通用电路，将证明所需算力压缩到通用计算机所需算力的1%。既然金融占经济的30-40%，那就意味着：“想象一下，你找到一种方法，用1%的成本运行40%的经济。”
- 面对 Hyperliquid 的直接比较，他称 Lighter 在所有这些维度上“都比几乎所有竞争对手更强”，同时承认早期参与者相较此前带来了“阶跃式变化”。更大的判断是，他认为排名前5至前10的去中心化交易场所对大多数客户而言已经强于中心化交易所，因此大部分迁移将从 CeFi 转向 DeFi。
- Ethereum 是整个体系的承重梁：它是“DeFi 最安全的账本”，相当于“链上的清算所”，10年来“完全没有出过岔子”；迁出 Ethereum 不是路线图上前10件、甚至前100件要做的事。下一步是26年Q3推出链上期权，与永续合约和现货共用同一套资产负债表及风险抵押品模型，最好能让用户通过 Robinhood 交易；Vlad 称这同时是技术、营销和合作伙伴关系问题。
6. 一个代币，不留歧义：所有价值都归属于代币
- Lighter 从第一天起就是美国 C corporation，没有双基金会架构，其基础判断是：“监管来得慢，但最终会到来——而且这件事正在发生。”代币发行前融资轮已承诺是“我们永远进行的最后一轮股权融资”；尽管超额认购5倍，最终只有股权表中1%的份额选择退出。主持人问股权是否不享有任何收入或利润，Vlad 回答：“Okay。”所有价值都归属于代币，团队完全押注于此。
- 他对终局的描述带有明确前提：Clarity Act 获得通过，且 SEC “继续推进创新”。届时股票本身将被代币化，Lighter 的这枚代币“也会是股权……这样这些事情就不会再有任何歧义。从我们的角度看，我们已经处于这一状态”。
- 谈到自己在 Twitter 上的活跃表现，他把信息流形容为会议间隙进行的“问答”，主要是在“指出这个行业的荒谬之处”。但评论区的回复甚至比他原本想嘲讽的对象还要荒谬。
I read that you were personally recruited by Ken Griffin to work at Citadel. What was that experience like?
Guest
Like, my first week, Ken brought me these track records of other funds, and Ken said, “Oh, today we’re interviewing boy genius Vlad.”
1. Harvard At 16 & The Olympiad Network Behind Anthropic
Vlad, thank you so much for joining us today. It’s a pleasure to have you here at the 1000x studio at the Seleni conference. This is going to be a fun one.
Guest
Thanks for having me.
We were talking a little bit before the pod, and one thing that struck me is that you’ve had a very interesting life, man. You’ve crossed paths with some very interesting people, and I think you’ve lived a life that a lot of other people have no conception of.
It’s kind of interesting because it starts in Northern Virginia. You went to high school at a magnet school called Thomas Jefferson High School for Science and Technology, and I heard that you were actually there with the other Vlad, Robinhood’s Vlad Tenev.
Guest
That’s right.
That’s right. It’s like I’m Forest Gump if he had been a math person, in terms of encountering different people who ended up shaping the world.
And while you were there, you were very active in the Physics Olympiad and the Informatics Olympiad as well.
Guest
Yes.
What drove you to do this? Did you always have an innate interest in math and science?
Guest
Yeah, for sure. I think I always liked numbers, structures, and things like that, and I got into programming when I was 8. But, as it turned out, I’m a very competitive person. I didn’t really know that I was until I discovered these competitions when I was around 12.
Before that, I thought, “Okay, this math and science stuff is kind of interesting. I’m kind of good at it, but I don’t know exactly what I’ll do. Maybe I’ll become a scientist or something, but I don’t know exactly how that works.” With competitions, you have a very clear feedback loop, and you can practice and get better. It all clicked for me and accelerated the progress.
You said you were programming. What could you possibly have been programming at 8?
Guest
Well, back then I played card games with my family, and I wrote some programs to basically program the computer to play the card games, so I could practice in my spare time, get better, and beat my family at the card games.
What was this Durak that you were programming?
Guest
Initially, QBasic and then Pascal.
Oh, I was talking about the card game.
Guest
I don’t know if you’ve ever played Durak.
I haven’t.
Guest
Okay, so it’s a very Russian game, but—
Oh, actually, that was one of the games. You’re right.
Guest
Yeah. There were 3 different card games.
That’s cool. So, you end up going to the Math Olympiads. You said you started competing at 12. What was that experience like, and what were the other people like as well? Was there a community?
Guest
There absolutely was. I don’t know how it was for the other kids—probably similar—but for me, it was the first time I felt like I was part of something. Before that, I was mostly interacting with adults. It was hard to find common ground with the other kids.
When I started doing these competitions, I was part of the community. Maybe someone who joins a sports team feels the same way, but it was competitive and much more collaborative, with everyone learning together. Especially for the team competitions, where we represented our school, our state, or our country, that brought us together closely as well.
Then you graduated from high school at 16 and ended up at Harvard, where you chose to study economics, which is kind of orthogonal to everything that you were doing in high school. Thomas Jefferson is a science school, and you were doing physics and informatics. Why economics?
Guest
When I got to Harvard, I was thinking about what to concentrate in. Actually, most people don’t know this about Harvard, but a lot of the curriculum is more general. If you pick a concentration, that doesn’t mean 95% of your classes are in that concentration. You’re still learning a lot of other areas, but you do need to pick a concentration.
For me, the way I looked at it was, “I’m pretty good at building stuff in terms of programming. That’s probably not an area—I probably already hit diminishing returns when it comes to that.” What I really wanted to take out of the college experience was what to build and learning more about the world as far as what to build.
The 3 areas I considered were physics, biology, and economics. Physics at that time, in the early 2000s, was kind of after a lot of funding had gone into physics, and it wasn’t necessarily growing very fast.
Economics, on the other hand, had a lot of exciting stuff going on with things like option pricing and portfolio theory. A lot of my math and programming background could actually be applied there to work on some interesting stuff. Biology wasn’t for me simply because I wasn’t somebody who loved being in a physical lab environment. So, I picked economics based on that frame of thinking.
While you were at Harvard, what were you doing? Just the normal economics course load? Were you doing specific types of research while you were there?
Guest
Yeah. Funny story: I took this microeconomics course in the first semester. We had this midterm, and I got the grade back. It was a 94. At Thomas Jefferson, and probably at most high schools, 94 is the minimum for an A. So, 90 to 93 is a B+, and so on. I thought, “Okay, this is kind of close. I almost didn’t get an A on this.”
Then I got an email from the professor asking me to come see him. I went to see him, and he said, “I was looking at your scores.” I thought, “Maybe I didn’t know how well I did.” He said, “Actually, did you know how well you did?” I said, “No. It was a 94.” He said, “The class average was 37.”
He said that, historically, since they had run the class, it was one of the highest standard deviations they had seen. So, he asked me to come do research with him. That was actually my first job.
Oh, that’s cool. And what kind of research was that?
Guest
It was microeconomics research.
And then you started working with other faculty the following year on more financial economics. But that’s how you got into research—just from that one exam, right? And that was the first exam?
Guest
That was the first.
What was it like being a 16-year-old boy genius in college?
Guest
I thought I had just barely gotten an A, because I didn’t know that they graded on a curve. In high school, it was just based on the absolute value.
No, for sure. I’m sure the professor was blown away by that, which is kind of fun. Did you do research with him all 3 years?
Guest
No. I did research with that professor the first year. I finished in 2.5 years.
It was 2.5 years. Did you find that you got the classic college experience in any way? You were quite young at the time. What was being at Harvard as a 16-year-old like? Were you the youngest on campus?
Guest
No. There was one girl who was 15, and there were 2 other 16-year-olds. So, there were 3 of us in our age range.
Was the experience—
Guest
I think people say that skipping grades means you didn’t get a proper college experience. The reality is that if you’re somebody who’s really into math and all these things, and you’re doing competitions, then if you go to college at 18, you’re hanging out with the same kinds of people anyway.
It’s not like if you’re 16, you’re hanging out with the math nerds, and if you’re 18, you’re hanging out with football players. It’s not that different. You still hang out with the same kinds of people.
Right. Was there anyone, basically up until this point, that you ran into who remains relevant in your life today? I think you mentioned a few.
Guest
Yeah. So, like I was saying, it's kind of like a Forest Gump experience looking back, right? In those early 2000s, one of my good friends from the computer Olympiad, aka the Informatics Olympiad, was Adam D'Angelo, who was actually Mark Zuckerberg's high school roommate, helped him build Facebook, was the first CTO there, and is now on the board of what was likely OpenAI. Two of the people from the physics camp are now co-founders of Anthropic. Somebody from my high school is Vlad Tenev from Robinhood, and that's just talking about founders, right?
I think at Harvard in the early 2000s, 2 of the people there came pretty close already to becoming president on either party. Another one served in the White House at a very senior level, and so on.
That's very cool. And so when you graduated at 19, I read that you were personally recruited by Ken Griffin to work at Citadel. What was that experience like?
2. Ken Griffin Closed Me At 18
Guest
Yeah, so it was when I was 18, and I think Ken—I wouldn't say he personally recruited me, but he personally closed me, right? The firm was already talking to me at the levels of the quantitative research team and portfolio management, and I was deciding between them and a few other firms. That's when Ken really got involved and spent a lot of time with me, walking me through how he thinks about markets.
That was, again, a very interesting time because markets had just become electronic a couple years prior, and he was building out some of these trading desks for the first time. One funny thing was that, based on some of the research I had done, I was trying to trade options on my own in college. Ken was like, “Yes, actually, some of those strategies did work for many years, but I just arbitraged them away, so that stuff doesn't work anymore. You should instead come work with us.”
He was very good at convincing me.
What were some of the—do you remember any takeaways or specific frameworks that he mentioned that have stuck with you? What can you tell me a little bit about?
Guest
Well, I probably shouldn't talk about any specific trading strategies, even now, 20 years later.
They'd come after you.
Guest
Well, I don't want to imply that, but I just don't want to give other people listening the idea that it's appropriate to talk about that kind of stuff in general, right?
But I will say one funny thing. My first week, Ken brought me these track records of other funds so I could check them out and see if there was interesting stuff in what these other guys—our competitors—were doing, right? I said, “Okay, I'll look at that.”
3. "That's A Ponzi": Ken Called Madoff In 2004
This was 2004, right? I had done some of my own research on other funds and track records, and I said, “Well, shouldn't we be looking at Madoff? His Sharpe ratio is very high, and he's trading options. Shouldn't we be figuring out what that strategy is all about?” And Ken said, “Oh, no, that's a Ponzi. Don't worry about that one.”
So he knew somehow? When was this?
Guest
2004. So this was a full 5 years before, right?
Right.
Guest
And then later on, when it was uncovered, the SEC published the full report and said that 3 anonymous hedge fund managers reported Madoff to the SEC. Maybe it was—I don't know exactly how that went down—but the SEC did look at it and thought they had checked it out. They thought it was okay, right?
Because he gave them a DTCC number. Madoff gave them a DTCC number to call, or to look at, and when they called DTCC, that was a fake number, and they never made the call. So had they just picked up the phone, the call would have uncovered it based on those suspicions.
I mean, stuff like that, I guess, turning it back to what I do now, is what got me really excited about using blockchain for finance, right? Something like Madoff would have never happened if the stuff was on-chain, where him getting called literally depended on somebody picking up the phone and asking DTCC a question.
How do you see that? How would blockchain have stopped Madoff from happening specifically? What about it would have stopped him?
Guest
Because the thing with Madoff, right, is that he actually never did any trading. He just said he did. He raised whatever it was, X number of dollars, and just said, “Okay, we made 10% last year, made 12% this year,” and just kept going and raising more money.
Like classic funds, there were no actual trades happening. He sent reports of trades to his LPs, but it was all fake. When the SEC tried to check it, they said, “Okay, what's the DTCC number for this account?” He gave them a fake number.
On-chain, all the stuff could just be verified. If you're up 10%, and the reason is because you bought Microsoft at 30 and sold at 40 or whatever, all those trades would be on-chain. Anybody could verify them.
Yeah, that makes sense. But one pushback that is often made is that people don't actually want all of their trades to be public in any meaningful way, and so you have to figure out a way to obfuscate that. How do you approach that problem?
Guest
Well, in terms of that, there are ways to do it with zero-knowledge proofs, actually, right? You could prove that certain trades were done without the exact trades being made public. There are different ways of doing that, right?
For one thing, it's not real-time, right? It's not like you're publishing the trade at a time when somebody else could front-run it, and there are issues like that. I mean, we're talking about longer-term investments here, but there are ways to implement dark pools or add additional levels of privacy.
I've actually been surprised that there's not as much market demand for that as I would have thought, because the way Lighter works, we use ZK more for scaling. We could use it for privacy and add these kinds of features, but it hasn't been much of a customer request. Once it becomes a priority, we can build more along those lines.
At this point, a lot of firms, including the one hosting the summit and others, trade on-chain, right? They're perfectly fine. They still have a lot of alpha in that environment, and they're perfectly fine executing.
I want to get into Lighter, but before we pivot to that, you spent a decent amount of time in finance. You were working at Citadel and then another quant shop, right? But after that, you actually took a break from finance and pivoted out. You worked at Quora as the head of ML, and then you also started an AI company. Then you came back to finance, so I want to hear: why did you leave finance, and then why did you come back to finance?
4. Why He Left Wall Street For Silicon Valley
Guest
Yeah. I certainly thought that going from Wall Street to Silicon Valley was going to be permanent. I certainly didn't come back to Wall Street in its traditional form, but at the time I thought, “Okay, this is really interesting. Trading, building out quant strategies, building out early forms of electronic trading—very interesting work. But at the end of the day, we're not really building new products and new markets.”
I saw that happening in Silicon Valley, by people like my friend who was CTO of Facebook and then started Quora, and others as well—early people at Google and early people—
So that's how you ended up at Quora, through your friend?
Guest
Well, yeah. I knew the company through him and others. I knew a lot of the early team there, but I considered a few different places. After Quora, I joined Joe Lonsdale to head up engineering at Addepar. There we recruited Scott Wu, also from the Olympiad world, and then Scott and I started a company a couple years after that.
Right. It's like one big club. What did you—are there people—
Guest
It's one big club, but it's an open club, right? It's not like an old boys' network. It's more merit-based. If you're really great at what you know—and it's not just math—if you're really good at customer success, if you're really good at operations, you're welcome to the club.
Maybe because up until you started Lighter, you've met a wide cast of characters and a lot of really interesting people. What are some of the biggest lessons that you've taken away from specific people that you think are valuable to you today?
Guest
Yeah. I wrote about this recently, and one of the big lessons from Joe Lonsdale, who also famously co-founded Palantir, was really the idea of having engineers be the tip of the spear, part of the customer success team—forward-deployed engineers. That idea was new.
It's funny because a lot of these learnings, if I say them now, some of you listening might think, “That's obvious. Of course it works that way.” But it wasn't at the time, right? That idea was an important learning.
Something I learned from the Tenev and Robinhood team was the idea that you can have a product with zero fees. At the time, that was considered crazy. Now it's like, “Yeah, of course you want to have zero fees for retail,” right?
I don't know. I'm trying to think: is there something that I learned from one of the industry greats that's still considered controversial?
That's a good one.
If you think about it at some point.
I feel like a lot of these ideas are about being contrarian and right—
Guest
And then it's no longer—
Right, it's no longer contrarian.
Guest
No, it's a problem with old advice, but I think it's also something that sometimes people just don't necessarily think about.
So, it might be commonplace knowledge among successful people, but if you're just starting out, you have to learn that framework at some point, right?
Guest
Yes.
It's like every day, there are probably 10,000 people learning the framework of a forward-deployed engineer because they've never come across it before, because they're maybe 18 years old.
Guest
That's right.
So, I think that's important. And I want to get to—actually, you mentioned Scott Wu. So, you've co-founded Lunchclub with him.
Guest
Yes.
That's right. And you worked with him for a few years before you ended up pivoting Lunchclub to Lighter.
Guest
Yes.
One thing that I noticed is that when you pivoted, even though it's a completely different concept, you managed to retain 80% of the workforce. Is that right?
Guest
That's 80% of engineering.
80% of engineering. How did you convince them to stay through it? I think this is very important for founders. How do you convince 80% of your engineering workforce to make a complete pivot into a different product line?
5. How To Pivot And Keep 80% Of Your Engineers
Guest
Well, hopefully because we've earned some trust from them. That's a starting point, but I think more specific things we did, right? One, they were part of the process. So, once we got approval from our board to do a pivot, we didn't just say, “Okay, we're going to pivot to this.”
We actually ran an internal YC within the company, right? So, we had 3 projects going at the same time, and then we did an internal demo day and pitch competitions and all of that, right? So, everyone felt like if someone—even the least experienced person on the team—came forward with an idea, “Hey, the whole company should do this,” that would have been considered too, right? And so they felt like they were part of the process, for one thing.
Two, they still knew that we had healthy runway, strong investors, and it's not like we were down to the last 3 months of funding or something like that. I think a third thing was, for a lot of our team, the technical problems they would have wanted to work on were actually closer to what we're building at Lighter. They joined the original team more because they wanted to work with other smart people. The problem space itself, they were somewhat interested in, but they were actually more interested in the problem space of Lighter, which is cryptography, scaling systems, quant finance, that kind of stuff.
That makes a lot of sense, and I guess that also is part of why that idea won the internal pitch competition, right? But, yeah, can you maybe talk about some of those technical challenges that you were solving? What were they, and what made them interesting? Because I think a lot of people hear the technical challenge and they go, “Well, I don't understand anything. I don't understand why that would be interesting,” right? But maybe you can translate that. What was interesting about these problems?
Guest
Right. Before going into specifically what was interesting there, I want to make a general point, which is, I think oftentimes there's this narrative: “No one cares about tech. It's all about distribution, it's all about user experience,” and this and that. I'll push back on that.
First of all, things like distribution and user experience can and do depend on the tech, right? When we've done partnerships with Telegram Wallet or Robinhood, they absolutely did diligence on the tech and cared a lot about that, right? Or user experience—shaving 100 milliseconds off trade execution, even if the user doesn't know how we're doing that, they absolutely care about that happening.
So, again, if you look at something like Bitcoin, how many people who use Bitcoin actually fully understand how it works? Not many, but they know that it does work, and they know that if it broke, somebody would find an error. So, I would make that meta-level point.
Why is the problem space of Lighter interesting? Because we're solving a problem—we're optimizing across a few different dimensions, right? We want to build an exchange that's low-cost, low-latency, verifiable, and secure. That kind of 3- or 4-dimensional optimization problem is very interesting intellectually, right? Because a lot of times when you make decisions, you trade off being strong in one of those areas and doing less well in another.
But sometimes—and these are the moments that really get engineers excited—you find novel ideas for a system that actually improve all of them, right? You write a piece of code where, after you deploy it, it's actually cheaper, faster, and more secure. That's the stuff I think that really gets engineers going.
So, you found that mostly these things are at odds with each other, and you have to make decisions, right? I mean, that's true of any—I mean, not specifically to us, but in general, when you're building systems, optimizing one goal usually means you have to compromise another goal, right? But sometimes you find these really cool ideas where all of the metrics improve.
So, can you talk me through some of these breakthrough moments in the process of building Lighter? When did they happen, and what product do you think you have today that really differentiates itself from the market?
Guest
Well, I guess one breakthrough moment was just realizing that we could actually build custom ZK circuits. We don't have to adapt general-purpose circuits to our use case; we can actually build custom circuits specifically for finance, right?
And we realized that the amount of compute you need to prove those circuits is actually only 1% of what you need for a general-purpose computer. And that's to run finance, whereas finance is, what, 30–40% of the economy. So, in a world where everything is on-chain, imagine you find a way to run 40% of the economy at 1% of the cost. That's pretty cool, right?
6. Running 40% Of The Economy At 1% Of The Cost
And is Hyperliquid doing these things as well, or do you feel like you've really created a new space in finance?
Guest
So, I think, again, going back to the point that there are a few axes here: low cost, low latency, verifiable, and secure. I would argue that we're stronger on all of those dimensions than pretty much any of our competitors.
Some of them were there earlier and built a strong track record. I think some of these guys had a step-function change relative to what came before them, right? So, I think we have to give them credit, too, and they had good products at the time when there were very few other options.
Any of these products in the decentralized space right now, at least in the top 5 or top 10, I think, are stronger than centralized exchanges for most customers, and that's where I think most of the shift will happen, also, from CeFi to DeFi. But we certainly believe that our technical choices have led us to being stronger across several different dimensions.
And we're seeing that being adopted either by direct users, but also through integrations, institutional use cases, and so on and so forth.
Do you think your partnership with Robinhood was directly a result of the technical choices, or was there something else? What were the conversations like there?
Guest
Well, the technical choices were a big part of it, for sure. Being on top of Ethereum—and, again, implementing what we've been able to implement while retaining Ethereum's security is a hard problem, right? Robinhood Chain is also on top of Ethereum, so that interoperability is really important.
Just low latency was important, too. I think in the demo we did last week, everybody there was pretty impressed with how fast it was, right? “Good speed on that” was kind of the quote. But it's not just that, right?
We forward-deployed engineers, right? Actually having people on the team who are there, working with teams at Robinhood, Telegram Wallet, and other ongoing partner conversations, that's important too. Or, like, being a good dance partner, as somebody put it.
It's not like—I think the idea of, “Okay, here's our tech; just throw it over the wall and use it as is,” can work in some contexts, but it doesn't if you're working with a large financial institution, whether they're on-chain or not, right? There are people there, and there are fine details to get right.
They have a lot of their own customers whose priorities need to be considered. So, I think it's much better to do that in a collaborative way.
Makes sense. Is there a world in which you ever move off Ethereum, or how important is Ethereum to your model?
7. Robinhood, Ethereum & Options Onchain
Guest
Right. Well, Ethereum is the most secure ledger for DeFi, right? If you think of that ledger as the on-chain equivalent of a clearinghouse in TradFi, it works, right? I mean, there's no reason why Ethereum wouldn't serve that function very, very well.
It's been secure and stable since the beginning. Over the last 10 years, there have been no hiccups at all. I mean, in a very hypothetical sense, if that stopped being the case, would we move?
If there was a new Ethereum that was even better in every possible way, would we move to that?
Guest 2
Maybe we could, right? But that's not, I guess, in our technical roadmap. That's not in the top 10 or the top 100 things that we're looking at, right?
What is the next problem for you to solve? Is it technical, or are you now focused on marketing and liquidity? Where is your focus right now?
Guest 2
Well, we have the capability of doing multiple things at once, but—
If you had to pick one, really, one—
Guest 2
Well, right now, one thing we're really excited about in Q3 of 26 is options: building real options on-chain and having that be on the same balance sheet and the same risk-collateral model as perps and spot, and doing that within Lighter.
Vlad Tenev
I mean, that's a really exciting problem, and it's technical, but we also need to get the right market makers involved. We need to tell customers about it. So it's a technical problem. It's also a marketing problem. It's also a partnerships problem.
Are people going to be able to trade options on Lighter through Robinhood?
Vlad Tenev
That's the goal.
That's—yeah. But do you consider yourself competitive with them, or is Robinhood to you just a layer for users?
Vlad Tenev
They've been very good partners. I think for a lot of these things, our approach is to work with a partner rather than go it alone. Our strength is in the technology part, right? So if there's a partner that already has very good distribution, relationships with institutions, and maybe some regulatory aspects—which we're also working on, too—we'd rather work with a partner in that case than go it alone.
Yeah, and the partnership, by all accounts, has—and reasonably, it looks like it's going to go very well so far. So congratulations on that. I think, generally, as somebody that's been in crypto for a long time, the financialization—the actual implementation of crypto in the world of finance—is very welcome. We're finally getting things that are useful and used, and people seem to enjoy them.
One question I think that a lot of people have, though, is the perennial issue with crypto: where does the value accrue, right? And I know you've talked about this before, but I think it's important for anyone new listening to get that from you. What is the value of the equity versus the token, and why do you have the token, I guess?
8. All Value Accrues To The Token
Vlad Tenev
Sure. So Lighter has a pretty unique structure in that we're a U.S. C-corp. We've always been a U.S. C-corp. There is no dual foundation or anything else like that. We built in the U.S. from day 1.
I guess our thesis at the time was that regulations are slow to come around, but they will come around, and that's been happening, right? To answer your question, we raised funds through venture funding. I think it's healthy to do that when you're building something as massive an ambition as Lighter. But we committed that the round we raised before token launch is the last equity round we'll ever raise, and everyone knows that. All the stakeholders at the time—we were 5x oversubscribed—so anyone who didn't like that and didn't think it was fair to them had an opportunity to get out at a very healthy return for them.
Only 1% of the cap table took that opportunity. So everyone else is all in on the token model, and the team is all in on the token model, with all value accruing to the token.
All value accrues to the token. Okay.
Vlad Tenev
Yes.
So the equity doesn't get any revenue or profit at all?
Vlad Tenev
Okay. Now, I will say that, in a longer-term time frame, let's say that we are in a healthy place with the CLARITY Act passing and the SEC continuing to innovate how they think about tokens, right? I think where we want to get to is where equities are tokenized.
So then let's say there was a project that wanted to have a native utility token and equity, for whatever reason that makes sense for them. Well, then that equity would also be tokenized. It would be like they have 2 different tokens.
Right?
Vlad Tenev
So I think now, for us, it would be 1 token, which we have now, and that token would also be equity.
That makes sense.
Vlad Tenev
So I think that's the world we'd like to get to, and from our perspective, we're already there. But from a regulatory perspective, once CLARITY passes and the SEC continues to innovate, to me, that's the exciting end state here. Then there will be no ambiguity about these things.
I think that's a good place. We have room for 1 last question. I've seen you reply on Twitter to a lot of people, and I think you have a very unique communication style.
My question for you is: is this just you shooting from the hip, or is this a strategy? Is this a marketing strategy, or are you just having fun out there?
Vlad Tenev
Yeah. Sometimes, if I'm taking a break in between meetings, I just like to—when I was at Quora, I did a lot of writing there. So to me, the feed is just Q&A. If I can answer a question, I'll answer it.
I think some of the stuff is pointing out the absurdity of this industry. It's funny to see some of the responses, right? Because I think the responses are even more absurd than the thing I was trying to make fun of in the first place.
I mean, that's welcome to crypto. Vlad, thank you so much for coming out today. This was awesome. Really enjoyed this.
Vlad Tenev
Thank you so much. Yeah.