63亿美元劫案后，赵长鹏牛津大学闭门演讲：我没让一个用户亏钱！（完整视频+全文）

编辑：米奇 来源：财经会议圈

2026年11月3日，牛津大学辩论社（Oxford Union），这座成立于1823年、曾迎接过无数国家元首与诺贝尔奖得主的辩论殿堂，迎来了一位特殊的客人：币安创始人赵长鹏（CZ）。

时机微妙。

就在五天前，10月29日，币安刚刚经历了人类金融史上按金额计最大规模的黑客事件之一，约63.4亿美元数字资产失窃，全平台暂停提现整整两天。

演讲前一天，提币通道才刚刚重新打开。台下坐着数百名牛津师生，主持人开场便单刀直入：一次如此重大的失误之后，用户为什么还要相信这个平台？

赵长鹏没有回避。

他承认漏洞真实存在，承认某些基础安全程序没有被严格执行，也拒绝给出“绝不再犯”的承诺。但他同时亮出了底牌：平台全额兜底，没有一个用户损失一分钱。这份坦诚，恰恰是当下AI攻击泛滥、黑客事件几乎每天发生的市场环境里最稀缺的资产。

而这场对话的价值远不止于危机回应。

从AI智能体之间即将爆发的海量微支付，到美元稳定币如何暗中扩大美国国债需求；从香港作为数字资产“试验田”的精准定位，到大国面对资本管制与货币国际化的两难，赵长鹏给出了一幅少见的全局图景。

他甚至罕见地谈及了自己的婚姻与离婚，坦言“找商业合伙人已经很难，找人生伴侣还要难得多”。

几个值得先读的观点：

其一，关于安全。“攻击者只需找到一个入口，防御者却必须排查所有入口。”他认为AI让攻击能力出现阶跃式提升，而防御方还停在被动模式，整个行业必须联合起来补课。

其二，关于AI与加密的交汇。未来AI智能体将代替人类完成海量小额交易，“我保证那会是区块链，至于是不是比特币，我不能保证。”这句话出自全球最大交易所的创办人之口，分量自明。

其三，关于稳定币的地缘逻辑。99%的稳定币由美元支撑，等于全球用户在替美国国债接盘，这解释了美国为何转向拥抱加密。

其四，关于采用周期。跨过20%采用率后加速，超过80%后放缓，初步普及还需五到十年，最后20%可能要一百年。

以下为这场近一小时访谈的完整实录，中文原文与官方英译逐段对照，推荐给所有关心加密行业、AI浪潮与全球货币格局的读者。

牛津赵长鹏（CZ）活动通知，2026年11月3日（周二）晚6点，牛津辩论社，赵长鹏（人称CZ）是币安（Binance）创始人，2017年创立币安并将其打造为全球交易量最大的加密货币交易所。2023年，他与美国当局达成和解，卸任首席执行官并服刑四个月，2025年获总统特赦。他目前仍是币安大股东。

地点：牛津辩论社，St Michael’s Street, Oxford OX1 2DU注意：注册不保证入场，先到先得，注册仅可用于向个人日历添加提醒。请提前到场，携带会员卡并遵守入场指引。

牛津辩论社访谈实录：赵长鹏对话全文（中英对照）

开场 / Opening

EN:“Without further ado, please welcome the President… and his guest, Mr. Changpeng Zhao.”

中：“废话不多说，有请辩论社主席……以及他的嘉宾——赵长鹏先生。”

EN:“Wow, okay, that’s a lot of you. So thank you so much everyone for coming. I’m sure we’re all very excited. What I’m going to do is a quick rundown of today’s events: there will be about thirty to thirty-five minutes of interview by myself, and afterwards we’ll go to audience questions for about half an hour. Very quickly as well—I always say this before every event—CZ has very kindly said that no particular questions are off the table. However, it would be best if you phrase these questions in a constructive and productive manner. No questions are off the table, but they should be evidenced and actual questions, rather than simply statements. So let’s get to the interview.”

中：“哇，人真多。非常感谢大家到场，相信各位都很期待。我先简单介绍一下今天的流程：先由我进行约30至35分钟的访谈，之后是约半小时的观众提问环节。另外——我每次活动前都会说——CZ非常友善地表示没有任何问题是禁区。但请大家以建设性、有成效的方式提问。没有问题是禁区，但提问必须有依据，是真正的问题，而不是单纯的声明。我们开始访谈吧。”

EN:CZ: “First of all, thanks for having me. I’m really honored to be here. I think the Union is actually kind of a true source of freedom of speech. We work in a slightly adjacent industry that deals with freedom of money—which is actually the name of the book. Freedom of speech is important and is the necessary foundation for freedom of almost anything else. Over the last few hundred years, civilization has evolved from physical freedom, like freedom from slavery, to freedom of speech, freedom of press, freedom of information. And now we’re entering the freedom of intelligence—well, not potentially, I think most likely it’s going to happen. So it’s an honor for me to be here to discuss it. This is the intersection of freedoms.”

中：CZ：“首先感谢邀请我，我真的很荣幸来到这里。我认为牛津辩论社实际上是言论自由真正的发源地之一。而我们从事的是一个略微相邻的行业——与金钱自由相关，这也正是我这本书的书名。言论自由非常重要，它几乎是实现其他一切自由的必要基础。过去几百年，人类文明从人身自由（比如摆脱奴隶制）演进到言论自由、出版自由、信息自由。现在，我们正在迈入‘智能自由’的时代——不是可能，我认为大概率会发生。所以能在这里讨论这些问题是我的荣幸。这里是不同自由理念的交汇之处。”

EN:Host: “Let’s get right to the elephant in the room. Over the last couple of weeks, Binance has had a security breach. What would you like to say to the users of Binance and the members of the public here, to either reassure them or inform them of the strength of Binance now and in the future?”

中：主持人：“我们直接谈那件绕不开的大事。过去几周，币安遭遇了安全入侵。您想对币安的用户和现场公众说些什么，来安抚他们，或说明币安现在和未来的实力？”

EN:CZ: “Obviously this is one of the largest hacking instances in human history by dollar amount. First of all, I want to say there are a number of deficiencies on Binance’s part—it’s very clear. And this is preventable. Every hack, no matter how sophisticated, no matter how long it took, is preventable—there are multiple ways to prevent it. So there are significant shortcomings on the Binance platform.”

中：CZ：“显然，按金额计算，这是人类历史上规模最大的黑客攻击事件之一。首先我要说，币安方面存在诸多缺陷——这非常明确。而且这是可以预防的。每一次黑客攻击，无论多么复杂、耗时多久，都是可以预防的——有多种方式可以阻止它。所以币安平台存在明显的重大短板。”

EN:“What they did do well: the instant handling and the communication was very clear, very transparent, real-time. Everything is on social media and broadcast very transparently. People see the transactions; third parties can actually analyze transactions sometimes even faster than I could. And the second, the most important part: Binance is responsible—it has enough to cover all the losses. That’s a principle I set when I was the CEO and founder, and I’m really proud that they followed that to this day. From that perspective, no users got hurt, no user funds got impacted—Binance absorbed the impact. So it is safe, it is secure, but there’s a lot of work to do, especially now with AI coming up.”

中：“但币安做得好的地方是：即时处理和沟通非常清晰、非常透明、实时进行。一切都通过社交媒体实时公开。人们能看到交易，第三方分析交易有时甚至比我还快。第二点，也是最重要的：币安承担责任——有足够的资金覆盖所有损失。这是我担任CEO和创始人时定下的原则，我非常自豪他们坚持到了今天。从这个角度看，没有用户受损，没有用户资金受影响——币安吸收了冲击。所以平台是安全的，但还有很多工作要做，尤其是在AI崛起的当下。”

EN:“It used to be we saw one or two hacks a year. Then it became one or two hacks every week. In the past week, we saw three or four hacks on different crypto platforms every day—and not just crypto platforms, banks as well. And actually, I can probably tell you that many banks are getting hacked, but they’re not public about it—you only hear about them six months later. With blockchain, at least you see what’s happening—you see the $40 million leaving a Brazilian exchange today or yesterday. It’s actually more transparent. In the short term it’s going to hurt, but the truth is AI is actually making our system safer, like it or not—by covering all these security issues, we can now fix them. So short term it’s a double-edged sword, but long term I strongly believe AI is going to make our systems much safer. The transition will just be very painful.”

中：“过去一年只见一两次黑客攻击，后来变成每周一两次。而上周，我们每天都能看到三四个不同加密平台被攻击——不只是加密平台，银行也一样。实际上我可以告诉你，许多银行也被黑了，但它们不公开——你六个月后才会听说。而区块链至少让你看到发生了什么——比如今天或昨天，4000万美元从一家巴西交易所流出。这其实更透明。短期内会有伤害，但真相是：不管你喜不喜欢，AI实际上正在让我们的系统更安全——正因为暴露了所有这些安全问题，我们才能修复它们。所以短期是双刃剑，但长期我坚信AI会让系统安全得多，只是过渡会非常痛苦。”

EN:Host: “You said this hack is preventable—and in some way this is true, all hacks are in some way preventable. What can you say to give assurance to both the people trading on it as well as the general public that this will not happen again? I’m approaching this question in good faith—there’s no way you can prevent everything, but I’m sure you can say something to ensure a smaller probability.”

中：主持人：“您说这次攻击本可预防——某种意义上这没错，所有攻击在某种意义上都可预防。您能对交易用户和公众说什么来保证这不会再次发生？我善意地提出这个问题——我知道没人能预防一切，但相信您能给出降低再次发生概率的保证。”

EN:CZ: “I think there are multiple levels of things we can do. First, something similar should be completely prevented. To be very frank, there were some basic opsec security procedures that were not followed by some internal employees—which is really bad. It should not happen for a platform that handles this amount of money. Then, the hackers’ sophistication is increasing very much. They used both traditional methods like fake interviews and social engineering, and very advanced methods using AI—they were able to use AI to analyze our systems in a lot of detail very quickly. Based on the information I have and what’s been made public, the entire hack took them about five to six months to piece together. They tried many, many times to social-engineer their way into the company. Much of that could have been quickly prevented with much stronger operational security procedures—and this part doesn’t have much to do with AI.”

中：CZ：“我们可以做很多层面的事。首先，类似事件应该被完全杜绝。坦率地说，有些内部员工没有遵守一些基本的安全操作流程——这非常糟糕，处理这么多资金的平台不应发生这种事。其次，黑客的复杂程度大大提升。他们既用了假面试、社会工程学等传统手段，也用了非常先进的AI手段——能够用AI非常快速地详细分析我们的系统。根据我掌握的信息和已公开的信息，整个攻击他们花了大约五六个月才拼凑完成，期间尝试了无数次社会工程渗透。其中大部分本可以通过更严格的操作安全流程迅速阻止——这部分其实和AI关系不大。”

EN:“But once they found one weakness, their ability to understand our system is very fast. That part we actually need a lot of help from AI players to make the system safer. We’ve had quite a few meaningful discussions already, and as an industry we need to come together to make all our systems more secure. As you said, I cannot promise a hundred percent that no other hack will happen—anybody telling you that is just false. This is a very intellectual group: the technology for attacking will continue to advance, and it’s harder to defend than to penetrate. To penetrate, you only need to find one crack; to defend, you need to find all cracks and fix yourself. Now we’re in a step function of AI on attack capabilities, and defense hasn’t really woken up—we’re in reactive mode. We need the whole industry, including the financial sector, to move this up: work closely with frontier AI labs, with law enforcement, regulators, and security experts. There’s a whole mountain of work ahead, but I’m very confident it’s possible to make an extremely secure system.”

中：“但一旦他们找到一个弱点，他们理解我们系统的速度就非常快。这部分我们确实需要AI公司的诸多帮助来让系统更安全。我们已经进行了不少有意义的讨论，作为行业，我们需要团结起来，让所有系统更安全。正如你所说，我无法百分之百保证不再发生攻击——任何这么说的人都是在撒谎。在座各位都很懂行：攻击技术会持续进步，而防守比渗透更难——渗透只需找到一个裂缝，防守却要找到所有裂缝并修复。现在攻击能力因AI出现阶跃式增长，而防守方还没真正醒过来，处于被动应对模式。我们需要整个行业——包括金融部门——一起提升：与前沿AI实验室紧密合作，与执法部门、监管机构和安全专家合作。前方有大量工作，但我非常有信心能打造出极其安全的系统。”

EN:“There’s a silver lining: with this amount—like six billion dollars—being hacked, not only is the company mobilizing, we’re seeing national security apparatuses mobilizing to help us. Because the incident is so big, the drive to make things happen, the ease of actually making things happen, is much easier now—everyone’s on board. I’m always very confident that the positive forces in our society always win. There are some bad players, but the positive forces will win. Lastly, on the recovery piece: the company, plus industry players, plus many law enforcement agencies did a fantastic and vital job to save a significant chunk of the damage, and that effort is still ongoing. I’m actually very impressed by the amount of help we got. Overall, I do think we will always be able to make the systems secure enough for people to use, so that people’s funds are always protected.”

中：“这事也有好的一面：正因为失窃金额高达约60亿美元，不只是公司在动员，国家安全机构也在动员帮助我们。因为事件太大，推动事情落地的动力和便利程度大大提高——所有人都在参与。我一直坚信社会的正向力量总会获胜。有坏玩家，但正向力量会赢。最后，在追回方面：公司、行业伙伴以及许多执法机构做了出色而关键的工作，挽救了相当大一部分损失，这项工作仍在进行中。我们获得的支持令我印象深刻。总体而言，我相信我们始终能把系统做得足够安全让人们使用，让用户资金始终得到保护。”

EN:Host: “Zooming out—you founded Binance relatively late in your life, in your late thirties or early forties. What is special or unique about cryptocurrency and blockchain, and why did you find this to be—perhaps not your life mission—but a very large part of what you wanted to do?”

中：主持人：“我们把视角拉远一些。您创立币安时相对较晚，约在三十七八岁或四十岁出头。加密货币和区块链有什么特殊之处，让您认定这虽未必是人生使命，却是您想做的很重要的事情？”

EN:CZ: “For most founders, you need to have a mission. If you’re just trying to build some products and make money, that usually doesn’t last long. Most founders have a very passionate need for a mission. For me, as you said, it took a long time to find that. When you first enter the working world, you’re just trying to make money, support yourself, then take care of your family. It took a while for me to get past that stage.”

中：CZ：“对大多数创始人来说，你需要一个使命。如果你只是想做点产品赚钱，通常难以长久。大多数创始人都怀有对使命的强烈热情需求。对我而言，正如你所说，找到它花了很长时间。刚进入职场时，你只想赚钱养活自己，然后照顾家庭。我花了相当一段时间才跨过这个阶段。”

EN:“I’ve always grown up with freedom in my head—I just didn’t realize it when I was fresh out of college. I was born in China, moved to Canada when I was twelve. Canada is a very freedom-oriented country, with a lot of freedom of speech. I was the beneficiary of that. Then I worked in different parts of the world, so I had a worldview—I didn’t lock myself into any specific country. When the Internet happened, there was freedom of information, but I was too young to capture that—I was just about to leave college. Then, while I was on a job, I came across Bitcoin, and I thought: this is the technology of the future. Once I really understood that, I knew it would have huge positive impact. I knew the road would be rough—if you look at all the important technologies, they all go through rough periods at the beginning. Printing was banned in most places when it first came out; freedom of press was a big debated topic. All highly impactful technologies or advancements in society are always fought—they’re not easy. So it was very clear to me then, and even now, that blockchain does bring freedom of money, and I think this will have huge impacts on society—we’re just beginning to see some of that.”

中：“我脑子里一直装着自由——只是刚毕业时没有意识到。我出生于中国，12岁移民加拿大。加拿大是一个非常崇尚自由的国家，言论自由很多，我是受益者。之后我在世界不同地方工作，形成了世界观——没有把自己锁定在某个国家。互联网出现时带来了信息自由，但我当时太年轻没能抓住——那时我快毕业了。后来在打工时，我接触到了比特币，我想：这是未来的技术。当我真正理解这一点，我就知道它会有巨大的积极影响。我知道这条路会很艰难——纵观所有重要技术，它们在初期都经历过艰难曲折。印刷术刚出现时在大多数地方被禁止；出版自由也曾是激烈辩论的话题。所有对社会有重大影响的技术进步都经历过斗争——都不容易。所以无论当时还是现在，我都很清楚：区块链确实带来金钱自由，它将对社会产生巨大影响——我们才刚刚看到其中一部分。”

EN:Host: “I’m a skeptical person. What’s this ‘freedom of money’ you’re speaking about? What’s wrong with the Federal Reserve, with using money issued by governments? Why do I need a separate blockchain, and why are people making so much money out of it?”

中：主持人：“我是个怀疑论者。你说的‘金钱自由’是什么？美联储有什么问题？用政府发行的货币不好吗？为什么需要单独的区块链？为什么有人从中赚了那么多钱？”

EN:CZ: “That’s quite a few things—I’ll break it down one by one. First is usability. In most developed countries like the UK—one of the strongest and oldest financial centers—you probably don’t feel the lack of financial access. But if you go to parts of Africa or Southeast Asia, you’ll feel it. In Africa, only eleven percent of people have bank accounts, and it takes them three days to pay a bill. In those situations, financial access is really important. Now with crypto, a guy with a smartphone in a village—the entire village can pay that bill within three seconds. There are real use cases here which are not obvious to established financial centers.”

中：CZ：“这包含好几个问题，我逐一说。第一是可用性。在英国这样的发达国家——世界最强、最古老的金融中心之一——你可能感受不到金融服务缺失。但如果你去非洲或东南亚的部分地区，就会感受到。非洲只有11%的人有银行账户，他们付一笔账单要花三天。在这种情况下，金融服务的可及性非常重要。如今有了加密货币，村里一个有智能手机的人，就能让全村在三秒内付清账单，不用花三天走到镇上另一头去办理。这里有真实的使用场景，只是成熟金融中心的人不易察觉。”

EN:“Second, nobody is free from inflation, and inflation is quite arbitrary. A little bit of inflation actually boosts the economy—that’s the argument—but too much inflation involuntarily means you’re getting poor very quickly. People should have tools to protect themselves from that, and blockchain and crypto is a tool. I’m not saying all crypto is good for that, but certain crypto assets are. Some early adopters of Bitcoin got rich because more people realized its value. But early adopters also took the risk—Bitcoin may not be the final one. Early investors in AI companies, in all the major companies, were rewarded disproportionately. Some people become rich because they took the risk and joined early—and this is just the beginning.”

中：“第二，没有人能免于通胀，而通胀相当随意。少量通胀确实能刺激经济——这是它的理由——但通胀过高就意味着你在不知不觉中快速变穷。人们应该有保护自己的工具，区块链和加密货币就是这样的工具。我不是说所有加密货币都适合，但某些加密资产是。一些比特币早期采用者致富，是因为更多人认识到了它的价值。但早期采用者也承担了风险——比特币未必是最终赢家。AI公司的早期投资者、所有大公司的早期投资者，都获得了不成比例的回报。有些人致富是因为承担风险、早早入场——而这才刚刚开始。”

EN:“Third, global payments. Paying somebody in the UK is probably not difficult, but paying across the country or across borders with larger amounts gets complicated very quickly. Payments haven’t really taken off in crypto, to be frank, but there are special use cases—cross-border remittances—growing very quickly. Then there’s tokenizing real-world assets: today there’s no reason for markets to only trade between nine and three. Events can happen on weekends—a war, strikes, other things—people need liquidity. The current technology could handle it, and blockchain does it much better and easier. I’m not political—I very much wish for a free economy globally, meaning global liquidity. That can be much more easily done on blockchain than through the banking system today. I’m not against banks—I actually think banks should adopt blockchain technologies to provide that too. I could go on, but I’ll be conscious of time.”

中：“第三，全球支付。在英国付款可能不难，但跨境或跨地区支付金额稍大就会迅速变得复杂。坦率说，加密支付尚未真正起飞，但有一些特殊场景——比如跨境汇款——正快速增长。此外还有现实世界资产的代币化：今天的市场没有理由只在9点到3点之间交易。周末也可能发生大事——战争、罢工或其他事件——人们需要流动性。现有技术本可以支持，区块链能做得更好、更简单。我不搞政治，但我非常希望全球经济自由化，这意味着全球流动性——这在区块链上比今天的银行体系容易实现得多。我不反对银行——我实际上认为银行也应该采用区块链技术来提供这些服务。这个话题我可以继续讲下去，但得注意时间。”

EN:Host: “Any new technology experiences hiccups. Based on what you see, how long do you think this process will take, if it will happen at all, and what are the primary obstacles?”

中：主持人：“任何新技术都会遇到波折。基于你的观察，这个过程需要多久？会实现吗？主要障碍是什么？”

EN:CZ: “Overall it’s usually a curve—initially slow, then once you cross the twenty percent adoption threshold, it picks up; once you reach eighty percent, the last twenty percent takes forever, maybe a hundred years. To reach twenty percent probably takes five to ten years in my opinion—I could be wrong—and the middle part, sixty to eighty percent adoption, probably one or two decades. Initially everyone is skeptical. I think we’ve just passed that stage—we’re seeing very good growth now. Namely, the US is now pro-crypto. When the US is pro-crypto, the rest of the world will follow—they will align with the US. That’s a very significant change in just the last two years.”

中：CZ：“总体上通常是一条曲线：初期缓慢，跨过20%的采用率门槛后加速；达到80%后，最后20%要花很久——也许一百年。我认为达到20%可能需要五到十年——我可能不准——中间60%到80%的部分，大概一到二十年。初期人人怀疑。我认为我们刚跨过那个阶段——现在增长很好。尤其是，美国现在支持加密货币。当美国支持时，其他国家会跟进——它们想与美国保持一致。这是仅仅过去两年里的重大变化。”

EN:“But upgrading the financial system takes a long time. Large organizations are resistant—not deliberately resistant to change, they just are. The infrastructure for banks to integrate directly with blockchain involves a lot of friction, and that takes time. We’re seeing new players become very large quickly—Binance is one example—but integration with traditional finance is still just beginning. Binance has difficulty getting banking access in many parts of the world, though it’s improving.”

中：“但升级金融系统需要很长时间。大型组织具有抗拒性——不是刻意抗拒变革，而是天然如此。银行与区块链直接集成的基础设施充满摩擦，需要时间。我们看到新玩家迅速做大——币安是一例——但与传统金融的融合才刚刚开始。币安在世界许多地方仍难以获得银行服务，虽然正在改善。”

EN:“Also, countries are used to operating by controlling the currency. Countries used to control religion and press—actually, the countries doing best right now are those that separated religion from state, and countries that didn’t control speech grew economically really well. But any entity, any government, has a self-protecting reaction: they want more and more power. They may say they don’t want it, but whenever something happens, they’ll say ‘we need this power’—over time it just grows. That’s a natural tendency. For them to let go—to say ‘we don’t need to control money, let’s have a real free economy’—there will be problems, ups and downs, and every time there’s a problem, the government wants to step in to fix it. But usually governments are not the most efficient at solving economic problems—not always, but in many cases. So we’ll see a lot of resistance. But crypto is not perfect either—we see a six-billion-dollar hack, we see many problems in this industry, so this industry needs to mature as well. There are considerations on both sides, but the trend is clear: the trend will continue, and technology improvement will continue.”

中：“此外，各国习惯于通过控制货币来治理国家。它们曾控制宗教和出版——实际上，现在治理最好的国家恰恰是政教分离的国家，而不控制言论的国家经济增长得很好。但任何实体、任何政府都有自我保护的反应：权力会越来越大。它们也许口头上不想要，但每当有事发生，就会说‘我们需要这项权力’——久而久之权力自然膨胀，这是天性。要让它们放手——说‘我们不需要控制货币，让经济真正自由’——过程中会有问题、有起伏，而每次出问题，政府就想介入解决。但政府解决经济问题通常不是最高效的——不是绝对如此，但很多情况下如此。所以我们还会看到很多阻力。不过加密行业也不完美——我们刚见证了60亿美元的失窃，这个行业有很多问题，它自己也需要成熟。两边都有考量的因素，但趋势是清晰的：趋势将继续，技术进步也将继续。”

EN:Host: “No interview is complete without an obligatory question about AI. How is AI and its rise impacting cryptocurrency, and do you have any other general thoughts about AI?”

中：主持人：“没有AI问题的访谈是不完整的。AI的崛起如何影响加密货币？你对AI还有什么总体看法？”

EN:CZ: “Let me share the general part first. I was just at DeepMind—one of my two bets here. AI is already making software development much easier, and it will continue at an exponential scale. I tried vibe coding before ChatGPT—it didn’t quite work, issues here and there. Now it can complete an entire app with barely any input from you. I haven’t tried it on large systems, but the trend is very clear. If you want to do a startup, the coding part becomes much easier. So everything around coding now comes into play: product design, market analysis, problem-solving. Twenty years ago, an internet startup had to set up its own servers; now you have AWS, Azure—so now coding becomes easy. The downside: competition will be much fiercer—everyone will be able to code, so you really have to differentiate your idea. But overall this is good: we’ll see a hundred, a thousand, maybe a million times more startups, entrepreneurs, new ideas, innovation. Very good in general.”

中：CZ：“先说总体看法。我刚去了DeepMind——我两大押注之一。AI已经让软件开发容易得多，而且会以指数级速度继续。ChatGPT出现之前我就试过vibe coding——不太灵，问题不断。现在它几乎不需要你输入什么就能完成一整个App。我还没在大型系统上试过，但趋势非常清晰。想创业的话，写代码的部分变得容易多了。于是代码之外的一切变得关键：产品设计、市场分析、解决问题。二十年前做互联网创业要自建服务器；现在有AWS、Azure——如今写代码也变简单了。坏处是：竞争会激烈得多——人人都能写代码，你必须让你的想法真正差异化。但总体是好事：我们会看到一百倍、一千倍、甚至一百万倍的创业公司、创业者、新想法和创新。总体非常好。”

EN:“Now bringing this into the crypto industry: beyond the security aspects, coding speed and technology improvements will become very fast—we can build much better apps, faster blockchains. AI is solving math problems so quickly that we may need to upgrade quite a lot of the math or encryption algorithms around crypto—but again, that will just make things safer; short term there may be migration impact.”

中：“再把AI带入加密行业：除了安全方面，编码速度和技术进步会非常快——我们能造出更好的应用、更快的区块链。AI解题速度极快，我们可能需要升级加密货币领域大量的数学和加密算法——但同样，这只会让事情更安全；短期可能有迁移影响。”

EN:“More importantly, AI is going to conduct a very large amount of transactions. People don’t understand how many transactions AI can conduct. Right now AI agents are still at the information/software-building stage—they’re not really transacting with each other yet. They will. Once AI agents transact on our behalf, there will be many times more transactions—what we call micro-transactions. You could hire a developer in Asia on a very short-term contract, and once they’re done, agent-to-agent transactions settle automatically. There will be a lot more of that, requiring a native digital, high-throughput currency—which I guarantee will be blockchain-based, whether it’s Bitcoin or not, I cannot guarantee. It could be a new coin released by an AI company, or a convergence on one public chain—who knows. But it will be a blockchain-based currency if it’s to be used by everyone in the world. I think we’re just at the beginning—AI will really increase the speed of growth.”

中：“更重要的是，AI将执行海量交易。人们不理解AI能执行多少交易。现在的AI智能体还处于信息/软件开发阶段，尚未真正互相交易。将来会的。一旦AI智能体代表我们交易，交易量将是现在的许多倍——即所谓的微交易。你可以在亚洲雇一个开发者做短期合同，完成后智能体之间自动结算。这类场景会大量涌现，需要一个原生数字、高吞吐量的货币——我保证它会基于区块链，但是不是比特币我不敢保证。可能是某家AI公司发行的新币，也可能是各方向某条公链收敛——谁知道呢。但如果要被全世界使用，它必然是区块链货币。我认为我们才刚站在起点——AI将真正提升增长速度。”

EN:Host: “When you have AI, you need servers, data centers—and mining Bitcoin also needs servers and lots of water. Do you have faith that hardware will catch up to the software requirements you predict?”

中：主持人：“搞AI需要服务器、数据中心——挖比特币也需要服务器和大量资源。你相信硬件能赶上你预测的软件需求吗？”

EN:CZ: “I think so. There’s a simple logic—the most valuable company right now is a chip company, Nvidia, and the top five or ten most valuable companies are AI companies. There’s so much money in there that they’re going to develop pretty quickly. I actually think large parts of Earth’s surface will be covered by solar farms soon. Right now solar covers a very small percentage of the Earth’s surface—I think double-digit percentages of the surface will be covered by solar, data centers—basically silicon. Demand for compute will continue to grow exponentially given the speed AI capabilities are growing. There will be limitations—we may run out of some raw materials—but with the speed of innovation and research, they’ll find substitute materials. There will be newer chips, different types of chips, different types of data centers. We shouldn’t be limited by our imagination that we only need to modify the current design. Things will change pretty dramatically—especially if AI is going to change the encryption algorithms for our industry, it’s also going to change material algorithms for hardware.”

中：CZ：“我认为能。逻辑很简单——现在市值最高的公司是一家芯片公司英伟达，市值前十大多是AI公司。这么多资金涌入，发展会很快。我甚至认为地球表面很大一部分很快会被太阳能设施覆盖。目前太阳能覆盖的地表比例还很小——我认为未来会有两位数百分比的地表被太阳能和数据中心——也就是硅——覆盖。以AI能力增长的速度，算力需求会持续指数级增长。当然会有限制——某些原材料可能耗尽——但以创新和研究的速度，会找到替代材料。会有更新的芯片、不同类型的芯片、不同类型的数据中心。我们不应该被想象力限制，以为只需修改现有设计。一切都会剧烈变化——尤其是如果AI要改写我们行业的加密算法，它同样会改写硬件的材料算法。”

EN:Host: “You visited an Oxford startup incubator. With vibe coding, young entrepreneurs miss out on the nuts and bolts that make a successful entrepreneur—the trial and error. I remember a startup craze with a company selling prepackaged juice that wasn’t very useful. How much faith do you have that this would be a useful set of startups?”

中：主持人：“你之前访问了牛津的创业孵化器。随着vibe coding的普及，年轻创业者会错过造就成功创业者的基本功——试错的过程。我记得当年的创业热潮里有公司卖预包装果汁，没什么用处。你对这一代创业公司的质量有多少信心？”

EN:CZ: “That’s a very good question, but I don’t think that’s going to be a problem. People need to learn the fundamentals of how things work—that’s why institutions like this are very important. Even in university I learned transistors, AND gates, NAND gates, but I never had to use them in my work. Very few people actually go down to the gate or transistor level today—we’re becoming more and more specialized; only a small number of people need to go there. When I started programming, C and C++ were the main languages—you dealt with memory allocation and much more complex stuff. Now you don’t worry about it.”

中：CZ：“这是个很好的问题，但我不认为这会成为问题。人们需要学习事物运行的基本原理——这就是像牛津辩论社这样的机构如此重要的原因。我在大学里学过晶体管、与门、与非门，但工作中从没用过。如今真正下沉到门电路或晶体管层面的人极少——我们的分工越来越专业，只需要一小部分人做这件事。我学编程时，C和C++是主流语言——要处理内存分配等复杂得多的东西。现在你根本不用操心这些。”

EN:“Today, when AI writes code, you should check it—especially for important applications, you shouldn’t blindly trust it. But AI is very good at analyzing code: you can have multiple agents, one writing, one checking—you can even use agents from different companies. People already do this—Anthropic writes code and OpenAI checks it—swarms of three hundred agents per developer, checking and checking. At some stage, you just need to understand the basic concepts of how AI writes code; at some point you won’t even need to look at the code. To be honest, at some point machines may just write code that humans cannot easily understand. Right now you write Python or JavaScript—those are made for humans and are not machine-efficient. Assembly is much more efficient. It’s the same for AI—AI has the compiler, and we don’t need the compiler or interpreter stage anymore. Once technology is mature enough, we don’t need to understand the nuts and bolts—very few people understand compilers today; ninety-nine percent of programmers don’t deal with compilers directly. I’m less worried about it. Today when we drive a car, we don’t really understand what the engine does—that’s enough.”

中：“今天AI写代码，你应该检查它——尤其对重要应用，不能盲目信任。但AI非常擅长分析代码：你可以用多个智能体，一个写一个查——甚至可以用不同公司的智能体。人们已经这么做了——Anthropic写代码，OpenAI检查——每个开发者配有三百个智能体的集群，反复检查。到了一定阶段，你只需理解AI写代码的基本概念；某个时刻你甚至不需要看代码。说实话，未来机器可能直接写出人类不易理解的代码。现在你写Python或JavaScript——它们是为人设计的，对机器效率不高。汇编语言高效得多。AI也一样——AI自带编译能力，我们不再需要编译器或解释器环节。等技术足够成熟，我们不需要理解底层细节——今天也极少有人懂编译器，99%的程序员不直接接触编译器。所以我没那么担心。今天我们开车，也不真正理解发动机的原理——够用就行。”

EN:Host: “Over the last couple of years, while the value of crypto stayed the same, the hype declined—I remember 2020 to 2022 when even Reuters and Bloomberg talked about crypto constantly, and that has disappeared. Now governments around the world are cracking down—for example China banning crypto. Two questions: first, do you think part of the reason is crypto’s association with black markets and shady dealings? Second, what do you think is necessary to persuade governments to turn to crypto—it takes a lot for a government to give up control over debt, interest rates, and the money supply.”

中：主持人：“过去几年，加密货币价值基本没变，但热度下降了——我记得2020到2022年连路透社和彭博都在不停谈论加密，如今这些报道消失了。现在各国政府都在打压——比如中国禁止加密货币。两个问题：第一，你认为部分原因是加密货币与黑市、非法交易的关联吗？第二，你认为说服政府转向加密货币需要什么？毕竟让政府放弃对债务、利率和货币供应的控制，是很艰难的。”

EN:CZ: “First of all, the less coverage of crypto may actually be a good thing. When technology becomes boring, we don’t talk about it—there’s no coverage on the Internet today; we don’t talk about TCP/IP. Google made a new protocol on UDP—actually I forgot the name—and nobody talks about it because it’s a fundamental layer now. We don’t talk about roads either. We take them for granted, which is great—that’s when something becomes fundamental infrastructure. Crypto isn’t quite there yet, but it’s getting there. There’s now a very high level of acceptance of stablecoins, and Bitcoin is kind of accepted as it is. So the less excitement is, in some way, crypto becoming more mainstream—a bit more boring. Boring means normal—that’s actually beautiful.”

中：CZ：“首先，加密货币报道减少其实是好事。技术变得‘无聊’时，人们就不再谈论——今天没人报道互联网，没人谈论TCP/IP。谷歌基于UDP开发了一个新协议——名字我都忘了——无人谈论，因为它已经是底层基础设施。道路也没人谈论。我们视之为理所当然，这很好——当一件事被视为理所当然，它就成了基础设施。加密货币还没完全到那一步，但正在接近。稳定币的接受度已经很高，比特币也基本被按其本来面目接受。所以热度下降，某种意义上是加密变得更主流、更‘无聊’了。无聊意味着正常——这其实是件美事。”

EN:“Also, AI is the new shiny object—everyone’s focus is on AI, and it’s getting a lot of coverage. That’s good too. So less coverage doesn’t mean people losing interest—it’s just becoming more mainstream.”

中：“另外，AI是新的耀眼焦点——所有人的目光都在AI上，它获得大量报道，这也是好事。所以报道减少不代表人们失去兴趣，只是它变得更主流了。”

EN:“For country adoption, there are many different considerations. The US has basically realized that stablecoins are one of the best ways for people to buy US Treasuries—you buy USDT or USDC, they take the money and buy US Treasuries. Why wouldn’t a country want that? And second, ninety-nine percent of stablecoins are backed by US dollars, which actually increases the dollar’s influence globally—further expanding what the dollar already had.”

中：“关于各国采纳，考量各不相同。美国基本已意识到，稳定币是让人们购买美国国债的最佳途径之一——你买USDT或USDC，发行方拿这些钱去买美债。哪个国家不想要呢？其次，99%的稳定币由美元支撑，这实际上扩大了美元在全球的影响力——把美元已有的优势进一步扩展了。”

EN:“On China—from my own understanding, let me generalize a bit. Many countries have capital controls—they want to prevent their own money from flowing out. That’s fine, it’s a decision every country can make. The US doesn’t have that; many large economies don’t have that. When you have controls on money going out, it’s also very hard to get money in—foreign direct investment will wonder, ‘if I make money, how do I get it out?’ The process is longer, so you actually limit the amount of investment coming in. China specifically wants the RMB to be the global dominant currency—that intention is pretty clear. But only four percent of RMB is held offshore. It’s very hard to become the globally dominant currency with only four percent of your supply offshore. The US dollar doesn’t have this limitation. And if you throw crypto into the mix, it’s hard to balance. I don’t know what the answer is—different countries have different problems to solve; there’s nothing right or wrong about it. Some countries are taking a much more cautious approach.”

中：“关于中国——按我自己的理解，稍微展开讲讲。许多国家实行资本管制——想防止本国资金外流。这没问题，是每个国家可以做的决定。美国没有这种管制；许多大型经济体也没有。但一旦管制资金外流，资金流入也会很难——外国直接投资会想：‘我赚了钱怎么汇出去？’流程更长，实际上限制了投资流入。中国明确希望人民币成为全球主导货币——这个意图相当清楚。但人民币只有4%在离岸持有。只有4%的供应在海外，很难成为全球主导货币。美元没有这个限制。如果把加密货币加进来，就更难平衡。我不知道答案是什么——不同国家有不同的问题要解决，没有对错之分。一些国家采取了更谨慎的态度。”

EN:“I think China views Hong Kong as a sandbox—Hong Kong now has two stablecoins, by Standard Chartered and HSBC, both old-school traditional banks. So they are experimenting. Also, China is a very large country and everyone is very active. Last year when they said Hong Kong would promote stablecoins, everyone got really excited, then they tried to temper the excitement. But it’s a big country—big countries have to make policies that are more black and white. A small country with three hundred thousand people can say ‘we’re open, do anything you want’ and check each project individually—maybe ten thousand projects, they can look at each one. A country like China, if they open up, there’ll be ten million projects—there will be scammers, bad players, people who cause problems, and it’s very hard to control. So they have to make more black-and-white policies. Big countries move slower and are more cautious—those are valid reasons. It doesn’t mean they’re worse; they just have a different problem to deal with. I do think every country is looking at crypto in detail, and every country will have to adopt crypto in one form or another, sooner or later.”

中：“我认为中国把香港当作沙盒——香港现在有两只稳定币，由渣打银行和汇丰银行发行，都是老牌传统银行。所以他们在试验。另外，中国是一个非常大的国家，大家都非常活跃。去年香港宣布推动稳定币时，大家都很兴奋，随后官方试图给热情降温。但大国必须制定更加黑白分明的政策。一个三十万人口的小国可以说‘我们全面开放，随便做’，然后逐个项目审查——也许一万个项目，可以一个一个看。而中国这样的国家一旦开放，会涌现千万个项目——会有骗子、坏玩家、制造麻烦的人，很难控制。所以必须制定更黑白分明的政策。大国行动更慢、更谨慎——这些理由是成立的。这不代表他们更差，只是要面对不同的问题。我确实认为，每个国家都在仔细研究加密货币，每个国家迟早都会以这样或那样的形式采纳加密货币。”

EN:Host: “So even though there are these quite large challenges, countries will eventually be persuaded to adopt crypto?” CZ: “I think so. A country cannot refuse to adopt the Internet—they cannot refuse to adopt AI. If you do not adopt those things, you become less efficient over time. In my life, I think there are three fundamental technologies: the Internet, blockchain, and AI. Countries can miss the middle one.”

中：主持人：“所以尽管挑战巨大，各国最终都会被说服采纳加密货币？” CZ：“我是这么认为的。一个国家不能拒绝采纳互联网——也不能拒绝采纳AI。不采纳这些，效率会随时间下降。在我的人生中，我认为有三项基础技术：互联网、区块链和AI。国家们可能错过中间那一个。”

*（访谈正文到此结束，此后进入观众提问环节。）*