他如何在加密史上最大崩盘期间募资6.5亿美元（ft. Haseeb）
Dragonfly第四支基金募资6.5亿美元，是押注金融加密领域的逆周期投资，而不是宣告市场阵痛已经结束。Haseeb表示，公司目前管理的资产规模约40亿美元，并在《Fortune》称为“加密VC的灭绝级事件”期间完成募资。估值下跌、资本稀缺之际，“钱变得很有价值”：基金的任务是在“所有东西都打折”时买入耐用的金融基础设施。
Haseeb的底层判断是，加密行业只有触及货币才会成功，而大多数非金融愿景都是行业自我制造的“集体幻梦”。Bitcoin、Ethereum、ICO、DeFi、NFT、stablecoins、RWAs、perpetuals和prediction markets都获得了自下而上的采用；去中心化社交媒体、DAO、游戏和元宇宙土地，则是投资者彼此说服出来的叙事。他对Gensler之后行业的质问很直接：“游戏都去哪儿了？……没人想要这些东西。”
主持人关于机构全面接管加密行业的“黑暗结论”被驳回，因为传统机构的采用是面向过去的，也不会复制加密行业最核心的属性。NYSE全天候交易或机构将债券代币化可能有用，但两者都未必需要区块链，也无法提供“全球化、无需许可、全天候、可编程的金融”。Haseeb把人们担心BlackRock最终拥有加密行业，比作早期预测《时代》杂志或Walmart会拥有互联网：传统机构可以把既有产品搬到线上，却未必能占据新一层网络中稀缺的价值。
Haseeb认为，耐心以及避免被迫清算，是加密行业最持久的优势。仅仅4个月前，Bitcoin还接近120,000美元，市场却突然变成“剧烈宿醉”状态；他的解释是，ICO、DeFi、FTX和Luna崩盘惩罚了那些被迫卖出的人，而底层系统仍在持续增长。“耐心几乎是加密行业所有问题的解药。”
加密骗局永远不会消失，因为让赚钱无需许可，既释放了人性中最好的一面，也释放了最坏的一面；但Haseeb认为，自2017年以来，骗局与实质项目的比例已经显著改善。下一轮投机复兴不会只是memecoins卷土重来：就像线上扑克一样，每种游戏最终都会职业化，直到业余玩家退出，只剩下“鲨鱼互搏”（shark on shark action）。他预计下一种游戏很可能原生于AI，而今天的深度交易者一开始会把它视为愚蠢的东西。
Agentic commerce是一个“兼容并行”的市场：获批的购买可以继续使用信用卡，而真正自主的agents需要类似加密货币的结算方式。封闭式AI实验室不太可能愿意承担agents未经人类批准就自主消费的责任，卡组织也会抵制prompt注入后发生购买、再通过拒付追回资金的交易。Stablecoins提供了相反的API——不可逆、全球化、买方自负风险的支付方式——适用于agents向其他agents、未知交易对手或金融市场付款，尤其是在美国以外的地区。
Agent支付的论点仍远远领先于现有证据。Haseeb形容x402的活跃度很小，在谈到 farms 后给出的粗略数字从“不到100万美元”到“每天约100万美元”不等，并警告说今天给一个agent加密货币，它可能几天内就会把钱弄丢。他押注的是模型能力快速提升，以及人类希望自动化采购和工作的需求，而不是当前已经有意义的交易量。
在AI领域，Haseeb预计美国前沿实验室会进一步扩大领先优势，最终成为准国家安全承包商，并在真正的行业崩盘中获得政府支持。他暂时把这一转折点放在2028—29年左右：届时军事应用和经济扰动的规模足以要求安全许可，并带来类似Los Alamos的保密程度。他为两项技术提出的最终行动原则相同：“留在牌桌上”，因为持续参与的人往往能熬过那些更聪明、早期获胜后却退出的对手。
1. Dragonfly坚持押注金融加密，募得6.5亿美元
Dragonfly宣布其第四支基金规模为6.5亿美元之际，许多同行正在缩减规模，甚至完全无法募资。Haseeb称，对所有基金管理人而言，背景环境都很艰难，但Dragonfly“幸运的是，在钱很有价值的时候手里还有钱”。
该公司管理的资产规模约40亿美元，投资组合包括Polymarket、Bybit、Monad、MegaETH、Kaito和Ethena。《Fortune》的配套标题称这是“加密VC的灭绝级事件”，Haseeb认为，这一标题准确描述了当时的募资市场。
他对Dragonfly能够拿到下一支基金的解释，不只是公司熬了过来，而是它回答对了加密行业最重要的问题之一：“这个领域究竟是干什么的？”答案是货币、金融和“价值互联网”，而不是去中心化社交媒体、游戏，或围绕DAO重组社会。
历史演进就是他的证据：Bitcoin引入点对点支付和非主权货币；Ethereum加入可编程金融合约；ICO提供资本形成；DeFi则把这一命题明确化。NFT、stablecoins、RWAs、perpetuals和prediction markets，都是同一条货币脉络的延伸。
2. 金融加密由用户需求自下而上建立，其余部分则是投资者制造出来的
Haseeb强调，金融应用是“自下而上”获得采用的。VC并没有先写出一份宣言，再把stablecoins或DeFi凭空召唤出来；用户先开始使用这些产品，投资者随后发现，创业者值得获得更多资本，以加速已经在运转的东西。
相比之下，Web3游戏、DAO、元宇宙世界和乌托邦式非金融用例，都是“一场席卷整个行业的集体幻梦”。Decentraland市值数十亿美元，却只有约300名日活用户，这成了典型案例：投资者为缺乏使用需求寻找解释，因为买家仍然愿意给虚拟土地定价。
监管后的反击有意保持严苛。Chris Dixon认为，监管和Gary Gensler让非金融加密应用来得太早；Haseeb的回应是：“Gensler已经走了……游戏都去哪儿了？”如今连游戏创始人也告诉他，用户主要想要的是更好的产品，而不是Web3元素。
3. 加密行业正在攻击互联网被禁止改动的金融底层
Haseeb用PayPal举例说明这一结构性论点：Peter Thiel最初想打造一种原生于互联网、与银行脱钩的货币，但监管机构迫使PayPal与Wells Fargo合作。最终留下的主要只是“覆盖在银行系统之上的一层薄薄接口”，并不是新的货币底层。
银行业仍遵循有限营业时间，因为国家金融体系最终由政府控制。私人银行可以在正常时期运营这些体系，但危机暴露出国家的权力；控制这条“窄腰”，就能监管货币如何流动。
因此，互联网时代的创新只改善了外围——移动账户访问和支票扫描——却没有取代底层架构。加密行业之所以“像野草一样生长”，是因为它终于直接与金融机构竞争，即使所有触碰金融的公司都面临传票、威胁、关停和被银行体系断供。
4. 传统机构的代币化，不等于无需许可的金融
Thread Guy的黑暗结论是，BlackRock、NYSE以及大型Web2机构可能会在封闭花园中采用代币化，却不会把用户或资本导向现有的DeFi协议、应用或基础设施。
Haseeb借早期互联网的类比，称这一判断“显然错误”：传统机构当时拥有分发能力和开发者，但Exxon、Walmart和《时代》杂志并没有因为上线互联网就拥有互联网。它们继承下来的产品形态，是为上一个世界优化的。
2017年的Enterprise Ethereum Alliance已经上演过一遍机构入场剧本：大型金融公司提出Ethereum分叉、代币化债券和区块链结算项目，但Haseeb找不到这场倡议留下的任何重要遗产。“他们当时做的，和现在做的完全是同一件事。”
他承认，NYSE全天候交易是好事，但货币本来就全天交易，交易所改变营业时间也从未需要加密行业。真正的产品是“全球化、无需许可、全天候、可编程的金融”；代币化债券是必要的库存，却不是他认为价值会集中的地方。
5. 下行周期奖励那些没有被迫卖出的人
Haseeb把加密周期描述为：从免费酒水和喧闹音乐，突然切换到灯光亮起后的尴尬时刻。大约4个月前，Bitcoin还在接近120,000美元交易；如今“派对结束了”，参与者正在离场，留下的人则感到“剧烈宿醉”。
VC最关键的优势，就是永远不会成为被迫卖方。无论是ICO崩盘、DeFi寒冬、FTX、特朗普之后的宿醉，还是Luna崩盘，Haseeb认为，反复永久摧毁回报的机制不是暂时回撤，而是被迫清算。
这让投资变得异常直接：“别人恐惧时要贪婪……那当然。你就买垃圾。”Dragonfly正在集中投资已经运转的金融产品，包括大笔投资Polymarket和Rain、持有Ethena和Mesh的大额仓位，以及其领投的7,500万美元Mesh融资。
6. 每一场投机游戏，最终都会在专业玩家解决它之后结束
Haseeb说：“加密行业永远不可能摆脱骗局”，因为完全自由的货币既可以让人向公益事业捐款，也可以让人瞬间发起memecoin rug pull。不过，他认为，2017年包括BitConnect在内的大型骗局泛滥之后，如今骗局与合法项目的比例已经明显改善。
两位嘉宾都认为，“加密行业的声誉从未如此糟糕”的说法不符合历史。Haseeb称，FTX之后的时期是“地狱的更底层”；Thread Guy则反对那种暗示崩盘并不严重、或市场信心始终完好的事后改写。
Haseeb用职业扑克解释了零售游戏为何会消亡。早期线上扑克中，中等水平的玩家仍能获利，业余玩家也玩得开心；随着职业选手不断进步，每张牌桌最终都变得机械而不受欢迎，只剩下“鲨鱼互搏”。
加密行业也以同样方式轮动：从ICO到DeFi farming，再到NFT和memecoins，每一代人往往都会把下一代视为垃圾。如今memecoins已经有太多专业玩家，供应链也过于完整；下一场游戏可能与AI有关，但Haseeb的“诚实回答”是，他还不知道它会是什么样。
7. 自主agents将支付分成需审批轨道与不可拒付轨道
Haseeb拒绝把cards和USDC看成二选一。通过Shopify或Etsy进行ChatGPT电商可以保留熟悉的流程：人类批准购买，再由信用卡完成结算。Amazon缺席反映的是平台控制，而不是信用卡无法服务agents。
完全自主的活动则不同，因为大型实验室面临极高的责任风险。一个agent可能遭遇prompt注入，在主人睡觉时买下不需要的商品，或在2分钟内执行50笔交易；如果反复发生拒付，卡组织最终会禁止没有明确审批记录的消费。
开源agents可能会推进得更快，因为它们的用户愿意承担更多风险——这相当于有人在监管机构宣布实验性肽类安全之前，就先自行注射。于是，自主支付API变成“不可拒付”和“买方自负风险”，正是Haseeb赋予stablecoins的属性。
这一支付轨道还会扩大可触达的交易范围。Agent可以向另一个agent、世界任何地方的未知交易对手付款，也可以买入金融资产——这些都是信用卡无法稳定覆盖的场景。Haseeb预计早期采用会发生在美国以外，因为跨境电商和B2B stablecoins在那里已经更重要。
8. 当前agent支付规模很小，但能力正在复利式提升
Haseeb反复提醒，整个论点仍处在早期。他称Coinbase采用的支付协议x402正变得越来越主流，但在经济规模上仍无足轻重：一方面说交易量“不到100万美元”，另一方面在剔除 farming 活动后又给出“每天约100万美元”的粗略数字。
更重要的问题是，agents究竟在买什么；他的答案是，当前用例仍然很小。“如果你给模型加密货币……就不要指望几天后还能看到那笔钱”，因为现有系统仍然过于脆弱且不可靠。
他的乐观来自变化速度：通用agents直到最近才变得可信，而模型质量正在快速提升。只要有可能，人们就会自动化采购和自身工作的一部分；经济活动会“无论我们愿不愿意”迁移到agents上，而且往往早于观察者意识到这一点。
9. 前沿AI可能成为国家基础设施，但不是马上
Thread Guy认为中国和美国在AI领域接近得令人不安；Haseeb称这是“非常中庸的看法”。他表示，中国开源模型可能只落后几个月，但据称它们是在蒸馏美国系统，而前沿实验室则在不断积累专有合成数据和内部数据。
他拿Alibaba的Qwen作为样本：Qwen在旧版AIME数学竞赛中表现出色，但在新题目上的表现明显变差；他将其解读为刷榜证据。根据他的实际生产编码经验，中国模型“就是差得多”，尽管过去6个月已经大幅进步。
中国仍是唯一有分量的挑战者，而且远远领先于欧洲或日本，但Haseeb预计，开源模型与闭源模型之间的差距会扩大。随着基准测试失去意义，衡量能力会变得更困难；美国领先实验室积累的私有数据，跟随者无法复制。
一旦军事用途“垂直增长”，并且AI每年将GDP推高约10%，他预计美国的控制措施会收紧——可能在2028或2029年左右，但“我不认为会是2027年”。实验室会变得像政府承包商：敏感岗位需要安全许可，保密程度接近Los Alamos，竞争也会变得资本密集且监管密集到难以承受。
10. 真正的AI崩盘会带来政府买单，而不是例行的VC支持
Thread Guy提到，据报道，政府在抓捕Maduro的行动中使用了Claude，以及随后与Anthropic发生的公开争议。Haseeb称这并不意外，并在没有十足把握的情况下猜测，Anthropic最终会让步，因为美国政府是“世界上最大的企业”，是一个不能失去的客户。
他反对含糊地说华盛顿会“兜底”OpenAI。这个词可能意味着股权投资、债券支持、购买算力、资不抵债时国有化，或者仅仅是口头支持；如果不说明机制，这个说法几乎没有解释力。
在他的叙述中，OpenAI距离财务困境还很远：公司仍在更高估值上持续融资，而Nvidia不投资500亿美元，本身不会造成崩盘。公司完全可以直接削减算力承诺。
但如果真正的崩盘到来，Haseeb认为政府干预“几乎毫无疑问”。华盛顿可以买入债券、投资算力，或提前购买需求，理由是前沿算力具有战略必要性，即使消费者或企业尚未愿意为此买单。
11. 长期坚持胜过天赋，因为早期赢家终会退出
在谈到Kyle Samani离开时，Haseeb回忆说，自己进入加密行业时，身边有一些他认为高不可攀的基金管理人。Kyle一直是最后一批“每天都在挨打”的同行之一，为组合公司持续奋战，而其他早期领袖逐渐沉默，或干脆退出牌桌。
Naval Ravikant的教训毫不煽情：“聪明人早期会赢，然后退出。”只要坚持得足够久，那些看似不可战胜的对手最终会退休、懈怠、消失或崩溃；有纪律的经营者往往不需要直接击败他们，因为“他们会自己打败自己”。
人才转向金融科技让Haseeb担忧，而人才加入银行会让他“泪流满面”。但转向AI不同：他认为AI是本世纪最重要的技术，其进步会叠加在加密、智能手机和能源之上，而不是与它们进行零和竞争。
他最后的号召很简单：“留在牌桌上。”加密并不是唯一值得投入的技术，而理解AI会让人更擅长加密；在苦干了大约10年后，Haseeb“哪儿也不会去”，因为他相信这个行业还有很长的路要走。
完整逐字稿
Mr. Haseeb, welcome back to the stream, my friend. How are you?
Thank you for having me. I’m good.
Of course, man. Look, at a grim time in the crypto markets, I just want to congratulate you and Dragonfly on the big announcement. Do you want to tell the stream really quickly what you guys just announced, and maybe give us a one-second introduction to what you do at Dragonfly?
Yeah. We just announced that we raised our fourth fund: $650 million, which I think, in this market, is not an easy feat.
Not a lot of people are announcing new funds these days. I know the fund market very well, and a lot of people are downsizing. A lot of people are not able to raise new funds. It’s a tough environment for everybody, including VCs and fund managers.
We’ve been very fortunate, and we’re lucky to have money at a time when money goes a long way. Just for a quick background, for those who don’t know me, I’m Haseeb. I run a fund called Dragonfly. We manage about $4 billion in AUM and are big investors in a ton of things in the space. We’re investors in Polymarket, Bybit, Monad, MegaETH, Kaito, Ethena, and a bunch of other things in the space.
I think it’s been an interesting era for VCs because Fortune wrote this article alongside our announcement, and I think the title of the article was something like, “Dragonfly raised $650 million amidst an extinction-level event for crypto VCs.” That was the exact title.
So, very clearly, it’s tough out there for everybody. I think we’ve proven something, and I recently got into this tiff with Chris Dixon from a16z about non-financial crypto. That was literally the week before we made this big announcement. A lot of people out there are feeling the pain right now, and that’s also true on the fund-management side.
Yeah, a lot of people invested in a lot of shit that went nowhere. I think one of the things we got right, which earned us the right to manage more money and continue doing what we do, is that we got a lot of the big questions in crypto right.
One of those really big questions was: What is this space about? Is this space about decentralized social media? Is it about gaming? Is it about the idea that all of society is going to be reinvented to run on DAOs? We never really believed that stuff. We believed that the core of crypto was about money, finance, and the idea of creating this “internet of value”—the internet capital markets and all that stuff.
That’s what we believe crypto is about. It seems kind of obvious now, but at the time it was not obvious at all, and a lot of capital was destroyed investing in things that fundamentally didn’t make sense.
It was never obvious. Honestly, it didn’t seem obvious, especially when I remember the 2021 and 2022 cycles and what was happening with gaming. I think Farcaster was the last decentralized-social protocol standing, or at least the last popular one. Farcaster shut down.
I can imagine getting cynical about the space when you’ve torched infinite money on going nowhere. Honestly, your back-and-forth with Chris Dixon was awesome. Can you give me the take on crypto being money and what this space is about where we currently stand in 2026? To a lot of the smaller, on-chain-focused players, it’s pretty grim, dude, and I would imagine it’s the same way on the institutional side. What’s left?
Look, it is and it isn’t. It’s grim relative to the party we were all having, where everyone was showing up, the doors were open, club music was playing, and you could just come in for free drinks. That’s what it felt like for a good while.
To be clear, this is crypto. This is how it’s always looked, as long as I can remember. I’ve been in the space for a while now, and we go through these boom-and-bust cycles. The party is bumping, everything is free, everything is amazing, and everyone gets to make money. Then, all of a sudden, it’s, “No, you’re all idiots. Everyone has to go home and get a real job.” Then we show up again a couple of years later, but the people who show up again are a different set of people.
I think this is crypto amplifying human nature, and this is part of human nature. But if you look at the things that have actually worked in crypto from the very beginning, where did we start? We started with Bitcoin, which was about peer-to-peer payments and non-sovereign money—digital gold.
The next thing that worked in crypto was Ethereum, which is programmable money: smart contracts and smart financial contracts. The next thing that worked in crypto was ICOs, which are fundraising and capital formation. This is also about money.
Then the next thing that worked in crypto was DeFi. DeFi has “finance” in the name. The next thing that worked was NFTs—non-fungible tokens representing financial assets that you can then move around and trade. Again, it’s also about finance.
Almost everything in this space that has worked at scale is now in RWAs, tokenization, perpetuals, and prediction markets. Prediction markets are event contracts, right? A very old idea in finance. All this stuff that’s really worked is about finance.
It was all adopted bottom-up. This was not VCs getting together and writing a manifesto that said we should build financial applications with crypto. People just did it. People started doing all this shit. It was the same thing with stablecoins. We didn’t manifest this into existence. We noticed that it was happening, and we thought, “We should invest in these things and give entrepreneurs more firepower so they can make them grow even bigger.”
All the other stuff—Web3 gaming, Web3 utopianism, DAOs—was stuff that we collectively came up with. This was a whole industry-wide fever dream that we convinced ourselves of through podcasts, blog posts, and conversations. We talked ourselves into it.
Yeah. Fair.
It was not users showing us, “Man, I really want to hang out in Decentraland all day.” In fact, nobody did. I remember we poured billions of dollars into it, and it had something like 300 daily actives. We just ignored it. We said, “Look, it’s early. You don’t need the users right now. It’s okay because they’re buying the land. What do you need to use it for? Obviously, people believe the land is valuable.”
I think all this reinforces the same lesson over and over again: crypto is about money. Crypto has always been about money. Why? Because money is the one thing that everything else the internet disrupted was never allowed to touch. You were never allowed to touch it.
There’s a famous story from the early days of PayPal. Peter Thiel wanted PayPal to basically do what Bitcoin did. He wanted PayPal to have a form of money that was completely untethered from the banks.
Originally, they were going to go live with this thing without a bank. They weren’t going to partner with anybody. They were just going to have an untethered currency that would be the native currency of the internet, and people would use it on eBay. That was their idea.
Basically, the regulators slapped them down and said, “No, of course not. Are you an idiot?” They then had to partner with Wells Fargo, and the rest is history. Now they’re basically just a thin overlay over the banking system. But this idea is super old.
I’d never heard that about the very beginning. Why have I never heard that before?
Go listen to some of Peter Thiel’s talks. He talks about it when he talks about Bitcoin. Everybody in the early days wanted money on the internet to be digital, because why would we use the old banking system that doesn’t operate on weekends and doesn’t operate after 4:00 p.m. Pacific time?
Especially if you live on the West Coast, you’re like, “Man, shit, the hours are horrible.” The whole world revolves around New York hours. Why, in a 24/7/365 world, are we still operating on this architecture?
The answer is, “Well, we are because we are.” In every country in the world, the banking system is basically nationalized. It’s basically run by the government. We have for-profit banks, but we all learned, especially when we went through the banking crisis a few years ago, that the government runs the banks too.
You can run the banks for a while as private citizens, but when everything really goes wrong, it turns out that the governments run the banks. You can pretend you run the banks, but we really run the banks.
This is the narrow waist through which all financial regulation has always persisted in every country around the world. Because you control the banking system, you control the way finance works and the way money works.
The internet and all the stuff we invest in in Silicon Valley could only really play around the edges. It was, “Now you can visit your bank account on a mobile app.”
Isn’t that nice? Now you can scan your checks instead of having to go into the bank, but if there’s ever an issue, you have to go into the bank branch anyway. You cannot totally innovate on this stuff. And that’s why the NPS scores for banks are among the lowest for any financial product or consumer product anywhere.
People fucking hate their banks. They hate their banks everywhere. They hate them in America, they hate them in emerging economies, and they hate them everywhere. The ability to compete with financial institutions and financial products is why crypto started working everywhere and why it grew like a weed. You could not kill crypto. People tried to kill crypto, but you couldn’t kill it.
One of the things that Chris Dixon said in his blog post is that he believed the nonfinancial use cases of crypto—DAOs, media, and gaming—didn’t work because it was too early. The regulators were getting in their way. Gary Gensler, right? I liked your rebuttal. It’s like, “Bro, how long are we going to do this Gensler shit for?”
Exactly. Exactly. Look, Gary’s gone. He’s gone over here now. Where are all the games?
Yeah, I agree.
Where are all the games? Launch them. Okay, great. Let it rip. But the reality is that nobody wants these things. They’re not good games.
I had some gaming entrepreneurs in my DMs, or even in my replies, saying, “Hey, it was so hard to build this stuff during that time.” It’s like, “Okay, well, dude, these guys are gone. Where’s your product?”
Where’s your product?
And they tell me, “Well, the thing is, people don’t really care about the Web3 elements. They just want a better product.” It’s like, yeah, I could have told you that.
I know. Build a fucking product.
One of my favorite takes, by the way, is that there was somebody else who had a similar article. Maybe it was just Chris Dixon and I’m getting confused, but everyone wants you to say, “Gary Gensler this, Gary Gensler that.” It’s like, okay, well, we’re done. We’re done with Gary Gensler. If people want the thing, build the thing. They didn’t want it then, and they don’t want it now.
This is a take that I actually really agree with, and I want to get your perspective on this. Where we currently stand in crypto, I think the ultimate black pill has always been—and is maybe starting to play out—that the crypto community, core devs, and crypto as an industry built a superior financial layer. I think that’s a fact. The data says we built a better financial layer than what already existed on the internet.
But the black pill is that these massive funds—the BlackRocks of the world—and these big Web2 institutions, like what’s happening with Tempo, are all going to come in and build their own walled gardens. The New York Stock Exchange wants to do 24/7 stocks. Everybody’s going to come in, build their own walled garden, and none of the money, capital, users, or consumers is going to flow back to what already exists here—the DeFi protocols, the apps, the infrastructure layer, whatever it is.
None of the capital is going to flow back into consumer crypto. It’s going to be just this walled-garden Web2 crypto. It’s kind of happening, right? You see, “Oh my God, the New York Stock Exchange is going to adopt 24/7 tokenization.” All Larry Fink and BlackRock want to talk about is tokenization. But how that actually crosses over into what we do on a day-in, day-out basis is relatively unclear. How do you see that playing out?
I think this is just obviously wrong. You could have said the same thing about the internet: The internet is going to allow the biggest companies in the world to move their inventory onto the internet, use their already superior distribution, and rule the internet. Why would the internet help you and your stupid little startup? The internet is going to help the giants because they already have access to web developers. Web development is so hard. How are you going to hire a web developer?
This was the exact same argument people made in the early days of the internet. The internet was weird. It was a bunch of hobbyists and nerds and people in universities making their little homepages to talk about whatever it was they were doing.
Eventually, it became a consumer internet. There were early adopters and later adopters. But the reality is that the form factors the institutions come in with are almost always wrong because they’re backwards-looking.
The idea that institutions are coming and they’re going to own crypto is not a new idea. I don’t know if you remember the Ethereum Enterprise Alliance.
Not really. Give me some context. 2017?
Yeah, 2017. Go look up a set of logos. It’s JPMorgan and, I don’t know if BlackRock was in there, but a bunch of financial institutions were part of the so-called Ethereum Enterprise Alliance.
What did they do? I couldn’t even tell you what they did, but there was a story that they were going to create these Ethereum forks that were all going to be for institutional use cases. They were doing the exact same thing they’re doing now: tokenizing bonds and moving settlement of some stupid thing that you’ve never heard of onto a blockchain. Isn’t that amazing? It never went anywhere. None of this stuff goes anywhere now.
The New York Stock Exchange going to 24/7 settlement is good. They didn’t need a blockchain to do that. That’s on them. They’re already stock exchanges. Currencies trade 24/7, right?
Fair.
Yeah. What exactly needs a blockchain to do that now? I’m glad that they’re doing tokenization. I don’t know—we don’t have a lot of details about what exactly they’re doing—but that is totally orthogonal. That’s not the story.
The New York Stock Exchange moving to 24/7 is good. They didn’t need us to do that. They didn’t need anything else to do that, just their own will to do it. The story is: What does crypto enable? It’s way bigger than, “Now it’s 24/7 stock clearing.”
What crypto enables is this global, permissionless, programmable thing. That’s not what the NYSE is doing. The NYSE is not doing global, permissionless, 24/7 programmable finance. If you’re not doing that, you didn’t need crypto.
In the early days of the internet, I remember people were very excited that Time magazine was coming onto the internet.
They were like, “Time magazine is coming onto the internet.”
Right. They had an early partnership with AOL.
People were like, “Oh my God, the internet is going to be owned by Time magazine because they’re so early. Everyone’s just going to go to TimeMagazine.com instead of using the internet. They’re going to run everything because they’re so early.”
And in retrospect, what a stupid take. What were the biggest companies in the world at that time? They were Exxon and Walmart. You might think, “Well, Walmart.com is just going to own everything.” The answer is no. Absolutely not. They’re not adapted for this new world.
Do you really think the biggest winner in crypto is going to be somebody who tokenizes bonds and brings them on-chain? Do you think that’s the biggest problem in the world that crypto is here to solve? Come on. There will be somebody who comes in and tokenizes bonds, in the same way that the internet still needs all the other stuff that predated the internet. We still need ExxonMobil. We still use energy. The internet uses energy.
Where is value going to fundamentally accrue? The answer is: what is scarce is financial institutions. A bunch of financial institutions can tokenize bonds, and they will do it. That’s great. We need the bonds. We need them on-chain. That’s going to be wonderful. That’s not where the money gets made.
All that being said, I understand I’m speaking to an audience that’s feeling very depressed. Spirits are low.
It’s a good take, though. I like the Time magazine analogy. It’s a good take.
The reality is that nobody wants to hear this when prices are moving against you.
Fair.
Prices have moved against us really fucking hard over the last 4 months. It wasn’t that long ago that Bitcoin was at $120,000.
Wasn’t it?
It was, and this thing still felt like a party. The party’s over. Everyone’s going home. A lot of people are like, “Hey, I shouldn’t have been here. I need to call my—”
“I’m violently hungover.”
Yeah. Yeah. Exactly. I understand that. I don’t want to be too Pollyannaish about it.
One of the things I said just a couple of days ago is that one of the advantages of being a VC is that, as a VC, you’re never a forced seller. That’s really easy to underestimate in this industry. Being a forced seller is the number one way you end up not making money in this industry.
Over the long run, if you stick around long enough, you see the ICO bubble collapse, you see the FTX collapse, you see the DeFi bubble collapse, and now you see this post-Trump hangover—whatever you want to call this. I don’t know what this is.
You see the Luna collapse, right?
What’s the one thing these things have in common? The only way you could have lost money over those time periods is if you were forced to sell. Yeah.
If you were not forced to sell, you’re okay in the long run, right? Because the answer is that crypto just keeps growing. It just keeps growing. And that’s the story that I come back to with this whole thing with Chris Dixon: people tried so hard to kill financial crypto. They tried way harder than they did with any of the gaming stuff. Who was going after the gaming companies? No, it was very milquetoast in all the enforcement efforts against gaming.
But anything that touched finance was crazy. We have so many portfolio companies that were getting subpoenas, getting threats, getting shut down, getting debanked. And they persisted anyway. They grew anyway. They had adoption anyway. You cannot kill this stuff because people just want it. It has the right to exist.
Even, you know, go look on-chain at these protocols. People say, “Who’s ever going to use Aave again now that BlackRock or NYSE is launching whatever?” This stuff is still enormous, and it’s continuing to grow. So my answer, very simply, is that patience is the antidote to almost all problems in crypto. It’s an unsatisfying answer, unfortunately, but it’s the answer.
But if it’s the right answer, it’s the right answer. My obvious follow-up to you is: how the hell do you guys deploy $650 million in this market, where it feels like there’s an exodus of talent?
All that anyone wants to talk about is AI. We’re talking about AI. I’ll ask you about AI after. I thought you did a good job on TVPN[?], by the way.
How do you deploy $650 million into crypto companies right now?
Yeah. The answer is very happily because—
And carefully, probably, right?
Everything is on sale. Carefully, carefully, carefully, but very happily. I love investing at a time when everyone’s feeling terrible. People say, “Buy low, sell high.” They say, “You want to be greedy when others are fearful and fearful when others are greedy.” But then, at the same time, they ask, “What do you do with the money?”
The answer is, well, obviously, you buy shit. You go and find great founders who are building stuff that’s working, and you invest in them. I feel like this is the easiest my job has ever been. There’s so much stuff that I’m totally confident is going to grow, and it’s at this intersection I’m talking about of financial crypto.
We just made big investments into Polymarket and Rain. We’re big investors in Ethena. We’re big investors in Mesh, which we announced recently. We led a round at $75 million into them. There’s so much more coming with respect to stablecoin adoption, the intersection of crypto and traditional finance, and also AI.
I think one of the worst takes that we have right now in this industry is, “Oh my God, AI is coming to eat our lunch.” No, dude. AI is just technology. Of course technology is going to be changing everything. We’re not the only technology in the world, and you really should not hope that we are.
If the only thing changing the world is crypto, then humanity is cooked. Crypto is a futuristic technology, meaning we need the future for crypto to be relevant. If the rest of the world is frozen in amber and is not moving, then adoption is going to stay at tiny numbers.
I think all this stuff, seeing the rapid acceleration of AI, is so good for crypto because—and this is maybe the place that we were going to go to—I think AI is actually going to accelerate the adoption of crypto, not hinder it.
A star in AI is Peter Steinberger, who created OpenClaw. He goes on the Lex Fridman podcast, and I feel like it was sort of the blow-off top of this “claim the fees on-chain” meta. He goes on Lex Fridman and, for 10 minutes, just says, “I hate crypto.” Every time you see this funny tweet, it’s like every minute into this take, you think it’s climaxing and he’s done, but he finds a way to crescendo it higher. His sign-off is basically, “I almost deleted the codebase,” which, give me a break.
We have this weird incentive structure right now where anyone that isn’t crypto and is building a technology really hates crypto. Part of the reason is that when you have the most open and permissionless capital market possible, you get a barbell. On one side of the barbell, you get revolutionary developers who come here and change the world, like when Vitalik Buterin creates Ethereum. He makes the decentralized world computer and genuinely changes the world.
On the other side of the barbell, you get the lowest-of-the-low grifter scammers who see big money and think, “I’m going to come extract.” The same thing is probably going to happen in AI, right? You get Sam Altman and Dario Amodei creating ChatGPT and Anthropic, and then there’ll probably be a whole slew of grifters, just like what we had in crypto, who see money and VCs that don’t really know what they’re investing in chasing the next OpenAI, throwing money at whatever. Take what you can get while you can get it and leave.
I don’t even exactly know what my question is on this topic, but how much does the crypto narrative to the outside world hinder our ability to grow? What happens to this deepest level of hyperfinancialization that we refer to as the trenches, especially as AI grows and develops?
Yeah, yeah, yeah. Okay, let me try to chew on that.
So, first and foremost—
Crypto will never not be scammy.
Okay, it will never not be scammy. It’s been scammy as long as I’ve been in this industry, and it has not stopped.
Now, that being said—
The relative ratios of non-scammy crypto—the real stuff, the revolutionary stuff, the really valuable stuff—have been growing relative to the scam. Scams have always been there, right? Coming into the space in 2017, most things that you saw were scams. Do you think it’s actually gotten better, the scam-to-good ratio?
The ratio. Of course. If you’ve been around and you saw what 2017 was like, there were so many scams raising enormous sums of capital. I mean, gigantic Ponzi schemes. We don’t see that anymore. When’s the last time you saw a gigantic Ponzi scheme? I mean an actual Ponzi scheme, not, “Oh, this guy rug-pulled some meme coin or whatever.” I’m talking about a Ponzi scheme.
BitConnect was an absolutely enormous token—an enormous, civilization-level Ponzi scheme.
Yeah, you don’t see that anymore. It’s kind of old-timey now to have these gigantic Ponzi schemes.
That is core to crypto. Just look at how many of these guys show up in the Epstein files. It was literally the worst of the worst building this stuff in the early days. The Bitfinex hacker and all the crazy shit that was going on in early crypto was just another level. It was really the Wild West, when it was just cowboys and Indians. That’s what crypto was like.
So, I think it’s gotten better on a percentage basis, for sure. Now, that being said, the scams in crypto will never go away, right? Why will they never go away?
The answer is because crypto reduces money to being completely free. When money is completely free, the worst of human nature is going to come alongside the best of human nature. It will just happen. There are beautiful things that happen with people donating money to Roman Storm’s defense, to Ukraine, or to whatever. And there are also people coming out and rug-pulling memecoins, harassing developers, or whatever it is.
So I think that will never go away, and I think you’ve got to live with that. Okay? You’ve got to live with that. The idea that the favorability rating of crypto is lowest now is wrong. That’s ahistorical. If you remember what crypto was perceived like after FTX collapsed, that was a lower rung of hell than anything I’ve seen.
By the way, not to interrupt you, but history is being rewritten as if it wasn’t that bad post-FTX and people still believed in crypto. That is shit.
That is absolute shit.
That’s right. So, all that being said, what do I think about the trenchers who are coming up and trying to say, “Why are you fighting this? Why are you throwing us under the bus? We were just trying to help. We just want to be a part of the action”?
I think there’s a natural cycle in crypto where every generation that comes into crypto looking for the new speculative game comes in, masters the game as it currently exists, and then the game gets too efficient. It starts off really fun. It starts off—it’s actually, I used to be a professional poker player.
It’s a lot like poker. When I first started playing poker, poker was an amateur’s game. Most people who played poker online were amateurs. As a result, if you learned a little bit of poker, you could make a lot of money.
And the people who were amateurs were also having a good time because there weren’t that many pros. The pros weren’t that much better than the amateurs, even if they were better. As a result, everybody was having fun. It was genuinely, you know, some days you win, some days you lose, but it’s just kind of a party.
As people got better and better at poker, poker started getting less fun for amateurs because they were just like, “Man, I just keep getting destroyed every time I sit down at the table. These guys are so tight. They’re so mechanical. Every time I sit down, there are five of them that suddenly join to play with me.”
That experience just starts getting worse and worse until those players start leaving. Then what’s left just becomes shark-on-shark action, and they move on to the new game.
And the new game is maybe Pot-Limit Omaha. In Omaha, nobody really knows the game yet. It hasn’t been very well studied. All of a sudden, this game is where all the bad players are playing.
Exactly. This game becomes where the party is.
I think this phenomenon happens in crypto, too. First it was ICOs, then it was DeFi Summer, then the food coins and the farming, and then all that stuff. Then it became the NFT game, and then it became the memecoin game.
The interesting thing about crypto is that it’s not always the same people. Some of the same people are there, but oftentimes it’s actually a new generation that comes in and is doing something that really is not recognizable to the previous generation. The NFT crowd— a lot of them think the memecoin stuff was stupid. They were like, “Oh, you guys are so nihilistic. We were really in it for the culture, and you guys are just in it to try to make money.”
I’m sorry. Give me one second.
Yeah, you’re good. Such a good take. NFT guys are like, “What we were doing was authentic. The memecoin trenchers are nihilistic. Fuck the memecoin trenches.” It’s such a good take.
Yeah. It’s always “kids these days.”
Back in our day, we were doing it right.
That’s right. That’s right. That’s right.
I think when you see the so-called revitalization of the trenches, it’s not going to be the same form factor because memecoins are too efficient. People are too good at it. There’s too much of a supply chain. There are too many professionals. The people who lost their money in the memecoin game are not coming back.
They’re not coming back in the exact same form because they remember. They were like, “Man, I got fucked over. I read about LIBRA. I read about Hayden and all these guys, and it’s just a crooked game. I don’t want to come back.”
But it’ll be something else. It’ll be something that they can now believe is different. This new game is, you know—and I was speculating with somebody that it’s going to have something related to AI. I don’t know what it is, but there’s going to be some way in which you can tell a story that people are willing to believe is fundamentally new and fair and not gamed. There’s nobody behind the desk at the casino rigging it for you.
I have no idea what that will be, but there will be some new speculative game within the next couple of years, and that’s going to revitalize a new generation. It’s going to look different, and it’s going to be a new set of people who come in. That new set of people is probably going to be very, very AI-native.
This form factor is going to make perfect sense to them. It’s not going to be that the trenches as currently composed are going to be like, “Great, now we can make it all back.”
They’re going to be the ones who dismiss this thing as stupid. “Why are you guys doing this?”
Exactly. Exactly. Exactly. That is the instinct you want to be looking for: when a new generation of people comes in doing something that looks really stupid, that everybody you’re following is dismissing. I think it’s just human nature.
What’s funny about the AI crowd—and I want to ask you a couple of questions about it—is that while they think they hate crypto, they’re basically running the NFT founder playbook from 2021: be as loud as possible, have the biggest launch video, tweet all day, get the hype announcement. They’re basically us without the crypto part. They don’t really realize it.
Us with products.
Yeah, us with—no, not even fake products, right? Products aren’t real. No, they are usually like one to one.
I want to ask you a two-parter on the AI topic. I want to ask you about U.S.-China. I’ll ask about that after. But—
Hold on. Before I answer that, I want to push back against one thing there. I think this is also crypto being way too down on itself. Elon Musk still tweets about Dogecoin. Sam Altman launched Worldcoin. If you look at the big labs, literally today EVMbench was just launched by OpenAI—the paper—
Sam Altman doesn’t give a shit about Worldcoin.
Sure. I mean, look, I’m not endorsing Worldcoin. I’m not an investor in Worldcoin. But if you look at the big guys in AI, almost to the last man, they believe in crypto. They’re not coming out here pumping our bags and doing any of that stuff, but the idea that futurists don’t believe in crypto is just wrong. It’s not correct.
Fair. Fair counter. God, Worldcoin. I think I tweeted this, and I think you engaged with my tweet: there’s a parallel universe where Worldcoin is the most important crypto token ever created, but unfortunately, we don’t live in it.
No, we don’t live in it.
We don’t live in that universe, basically.
You know, the narrative—I agree with your take. The game is solved. It’s been solved. We needed a new game. Eventually, it’ll just happen. It probably has something to do with AI.
There’s this narrative that I think is one of the grandest narratives we’ve had in crypto in a long time: agentic commerce. Agents need a Phantom wallet, access to blockchain, and USDC to do anything. I hate when people are like, “If this token fails, crypto is over.” It’s the same idea as, “If Bitcoin loses this level, it’s lights out.” I hate doing that. But this one does feel relatively higher-stakes, depending on whether it plays out, because I feel like the upside is very high—
—and the downside is that crypto gets skipped over in the agent adoption curve.
My question for you is: how do you envision this agentic commerce ecosystem and marketplace playing out as it relates to crypto?
Okay. The first thing here is that it’s a “yes, and.” It’s not, “Well, either credit cards win or USDC wins.” I think that’s obviously foolish.
OpenAI already has commerce built into ChatGPT. You can go into ChatGPT, and you can buy stuff on Shopify. You can buy stuff on Etsy. Interestingly, you can’t buy things on Amazon on ChatGPT because Amazon does not allow them to, and there’s a bigger story there.
Clearly, those purchases happen through credit cards, and the workflows will be exactly the same as the consumer workflows that exist today. All of that stuff is user-approved. It cannot buy anything without you approving it.
All the big labs pretty much guarantee there is no fucking way that they’re going to allow an AI agent to make transactions without you approving them.
Right. It’s such a liability nightmare. Way too risky.
They will not touch it. Even something like OpenClaw—Sam Altman said when they acquired OpenClaw, “We see this as a big part of the future of ChatGPT.” What does that mean? Who knows? But you can get a sense of where this is going: a kind of generalized agent that ingests your Gmail, your calendar, and all this other stuff.
But the reality is, are they going to have a self-driving agent that’s going to write emails for you? That’s probably a liability nightmare. We already have stories about people committing suicide from getting AI psychosis.
Imagine what could happen when an AI gets you fired from your job. The lawsuit mania is going to be crazy. You’re going to see open source pull ahead because open source is always willing to take more risks.
It’s like peptides. People are just willing to inject themselves with anything if it can get them ripped or skinny, way before the FDA says, “Okay, this thing is definitely safe for you to do.”
The same thing is going to be true with respect to commerce. Getting your AI agent to take a credit card—you could do it, but how are you going to pay for something that’s not a U.S. merchant? What’s going to happen when your credit card company starts to realize that you’re charging back stuff that your agent did while you were asleep?
You’re asleep, your agent buys some stupid thing that you didn’t want it to buy, or it got prompt-injected and bought something where you’re like, “Oh, shit, I didn’t want it to buy this.”
Then you do a chargeback, and Visa’s like, “Wait. This kind of looks like an AI agent did this.”
This did it while you were asleep, and it did 50 of these in 2 minutes. Clearly, this is not you. We are going to make it a violation of the terms of service for you to use an agent autonomously, doing things without a clear record of you approving it.
Otherwise, just charge back.
Actually.
Right. Right. Exactly. I think there’s—
There’s just all this stuff that gets built in. When you really think through all the steps here, there are going to be 2 workflows. There’s going to be 1 where a human approves everything, and there’s going to be a second where the agent runs autonomously, right?
If the agent is truly running autonomously, the API for that is no chargebacks and basically buyer beware. What is that? That’s crypto. That’s what stablecoins are: no chargebacks, buyer beware.
And the way that opens up the universe of what your agent can pay for—because in the world of credit cards, it can only pay somebody who can take a credit card, right? That’s not everybody. If your agent wants to pay another agent, if it wants to pay somebody whose identity and location you don’t know, or if it wants to buy a financial asset, all that stuff can happen in crypto, but it can’t necessarily happen using a credit card.
So I think what you’ll see is that we bifurcate. You’ll get your AI agent to approve, “Hey, buy me some toilet paper.” Fine. That’s going to happen with you clicking a button, and it’s going to happen with your credit card.
But there’s going to be a whole different set of people. Again, probably mostly open source and homebrew in the beginning. These are going to be the weirdos who are doing stuff online. And, to be clear, a lot of this stuff is going to be happening outside the US.
So, non-US stuff where it’s like, okay, credit card acceptance is maybe not as high, and there are a lot more international transactions going on. That’s already where you’re seeing B2B stablecoin adoption is very, very high.
Exactly. So, in the same places where crypto adoption is already high, you should also expect those people to have AI agents—
—and those AI agents will be using crypto.
So now, look, this is early days, right? Go look at x402 volumes right now. That’s the payment protocol that’s becoming increasingly dominant, and Coinbase is adopting it. It’s tiny, right? We’re talking a big game about something that will happen years in the future. It’s less than $1 million.
Is it going—
Oh, it’s okay. Okay. Okay.
It’s nothing. These are tiny, right?
I mean, you want to look at the trend line, but mostly you want to look at what they’re doing. That’s the real question. What are they doing? What are the use cases? Right now, they’re all pretty tiny, right?
There are basically some farms inflating the numbers. If you remove those farms, then you’re getting to a number of around $1 million a day, something like this. It’s not significant.
So what we’re talking about right now is that, very clearly, the models are not good enough to do this yet.
If you give your model crypto—if you just get Claude Code or OpenClaw and give it some crypto—you should not expect to see that crypto in a few days. That’s just the reality of where things are right now.
But these are all getting way better, really fast. We just got here a month ago. Realize the accelerating time frames here, and you have to draw the line out. This stuff is only going to get better and better.
And the number 1 thing I guarantee you is that the moment it makes sense, people are going to be giving their agents crypto and getting them to start doing stuff, because it’s human nature. It’s just the span of human ingenuity. People will find ways to automate their own jobs. They’ll find ways to automate procurement. They’ll find ways to automate all this shit.
I think it’s going to be really weird, but AI agents are going to be driving more and more economic value, whether we want them to or not. It’s going to happen under our noses.
That was beautiful. God, you’re so optimistic. It’s so good.
One of the last things I want to ask you, and then I’ll let you go in a second here, is—you know, all I want to talk about is the AI bubble, right? AI bubble, AI bubble in macro.
We did this really fun streaming segment yesterday, and we called it “China Maxing.” We basically looked at the US versus China and how scarily competitive it is on the AI side, especially with China running most of their models open source. DeepSeek wasn’t a 1-off. We have Kimi, we have MiniMax, and then we looked at everything else—every other vertical.
We’re looking at GDP, infrastructure, manufacturing, and the US is basically winning AI, barely, and then getting blown out of the water on just about every other measurable vertical. Shipbuilding, someone in the chat said. Yes, it’s not great.
My question for you is: on this AI intelligence race between China and the US, how do you see this playing out? Specifically, is the US government—AI is a proxy to the military—just gassing this with backstopping OpenAI and Sam Altman and the capex? How do you see this playing out over the next year or couple of years?
Yeah. Okay. I think this is a very mid-grade take. The idea that China and the US are very close on AI—I think that’s clearly not true.
Okay.
The open-source AI models out of China are a few months behind the US. This is true, but it’s true because they’re distilling on US models, which we know very, very—
Allegedly. Come on. We’re not idiots.
The reality is that being behind the frontier has so little value when there’s going to be more and more compounding taking place within these big labs. We basically just got to the point where we’ve exhausted all public data, and more and more of the data being generated within these labs is proprietary. It’s internal. They’re creating the data themselves.
That means that getting your model to be on par with these things is difficult. The other thing, of course, is that there’s all this benchmark gaming.
If you look at Qwen, for example, it’s been proven multiple times over. There are actually a bunch of recent papers that demonstrated this. AIME, which is a big mathematics competition that happens every year, has a lot of competition among the big labs to show that they can do well on it. It’s kind of like the SAT: you want to show that you can do really well on this year’s SAT.
What they show is that Qwen, which is the big Alibaba model and one of the best small-scale models, falls off a cliff on the new AIMEs, but it’s amazing on the old AIMEs. That’s exactly what you expect to see if you’re benchmark-gaming, right?
All the Chinese models benchmark-game. We all know that. The way that you can prove that to yourself is to go use them.
If you go actually use them and try to write production code, they’re just way worse. They’re all way worse. They’re a lot better than they were 6 months ago, and they’re also better than the state of the art was 12 months ago. But that’s true for the labs, too. That’s true for everybody. We’re just learning so much more, and compute costs are driving down really fast.
So I think what you should expect to see is the gap between open source and closed source grows, not shrinks.
Okay.
That’s what I think you’ll see. But I also think it’s hard to measure that gap, because benchmarks are not nearly as effective as they used to be because of all the benchmark-gaming.
Yeah, that’s the nuance, right?
So that’s my claim there.
Anyway, all this is to say: is China formidable? Absolutely. Are they way ahead of anybody else in the world? Absolutely. Europe’s not even on the map. Japan has 1 company that matters. Everybody else is basically at 0.
I think it’s true that China has the most competence to be able to catch up to the frontier and copy what the big labs in the US are doing. That’s the right claim to make.
But here’s the thing about the AI labs in the US: they all leak like sieves. You go to Silicon Valley, and there are just dudes in coffee shops leaking to each other: “Here’s what we’re doing. Here’s what you’re doing. Oh, how did you do this thing?” They just tell each other. They have thousands of employees, so it’s impossible to stop it.
That stuff will stop when the national security stakes increase enough that the US government basically takes over and says these companies are getting nationalized.
What’s the timeline on that, do you think? The AI 2027 article speaks about this, right? Obviously, it’s going there.
Yeah. Yeah. Yeah. I don’t think it’s 2027. I think it’s maybe 2028 or 2029, when you see the amount of economic disruption ratchet up and the amount of military applications of AI basically go vertical.
What did you think when you saw the government come out and say, “We used Claude to capture Maduro”?
Yeah. I mean, they’re now going into this big public fight about it. The Wall Street Journal covered it this morning.
I feel like that was a big deal that they came out and said that publicly. I was like, “What?”
Yeah. It’s not surprising. If you hear what people in the administration say about Anthropic—“There’s a bunch of woke AI doomers”—it’s not surprising that all the culture-war stuff is coming out in this back-and-forth. I think the reality is that the people within the government realize that Claude Code is really good.
There’s a bottoms-up thing of, “Look, we just can’t afford to lose this.”
My guess is that Anthropic caves, not that the government goes and uses some other thing.
They’re not much of a choice.
That’s my guess.
I think so. The reality is that the government is—you know, Anthropic is an enterprise go-to-market, right? The government is the biggest enterprise in the world.
Your biggest buyer.
Exactly. It is so big. It’s the biggest organization that has ever existed in humanity. They need that contract, and I think they’re probably just going to do what it takes to win it.
What do you mean when you say “nationalize the companies”? How does that work?
Basically, if these companies get treated as government contractors, then everybody who works there has to have a security clearance. If you touch anything that’s sensitive, it’s like working on military equipment. These are government contractors, but they’re required to have very high levels of secrecy in order to continue doing what they’re doing. It’s going to look more like Los Alamos than Facebook.
What are the economic implications of that?
The economic implication is that it’s much harder to compete with the labs. They have the biggest contracts, they cannot be removed, and nobody can compete with them anymore. The capex requirements and the regulatory requirements to compete are astronomical. They also have this compounding data advantage, because all the data they’re ingesting from these gigantic contracts is something nobody else can replicate.
Is this the spot that Palantir is in on the defense side?
To some extent. I don’t know how reusable their data is across different contracts. Probably there is some reuse, but I don’t know that much about it, so I’d be speculating. It’s definitely true that, for Palantir, a lot of it is just that they have the defense-contracting relationships. I don’t think it’s necessarily that they have such a tremendous data advantage. They’re just very good at what they do, and the government trusts them.
For the AI labs, I think it’s more that we just cannot have these things leaking to China anymore. I think that’s really what it’s going to come down to. Right now, these things leak to China. Is it a national-security concern? Kind of, right? We talk about it as though it is, but what are they really doing about it?
Yeah.
The answer is nothing, right? But if they leak our military blueprints, or if nuclear scientists defect to China, that’s a—you know, we will potentially bring out bomber jets if that kind of stuff happens. That’s not happening with AI, and it’s very weird that it’s not happening with AI. It will happen. If this stuff is moving GDP by 10% a year, then the US is going to say, “No, no, China, you don’t get to have this. We have this.”
So, if we get there, does that mean we just backstop, infinite spend, and markets just keep bleeding higher?
I think this backstop conversation is very simplistic. What does backstop mean? Does that mean that the government is going to invest in OpenAI? Does that mean that they’re going to buy OpenAI bonds? Does that mean that, if they’re going bankrupt, they’re going to nationalize the company? There are many different dimensions of what a “backstop” means. A backstop could also just mean that they say something publicly if OpenAI is having a bad funding round. What does it mean? I don’t know.
The reality is that OpenAI is so far away from actually being in any kind of financial distress. People are freaking out because the stock market went down by 1%.
2%.
Right. There’s nothing. They’re still raising rounds at higher valuations. NVIDIA didn’t invest $50 billion—oh my God, now it’s all going to collapse. Come on. They’re still raising the money, and they might just pare back their compute commitments. That’s fine. They’re not going under. They’re not going bankrupt. That’s very different from what people are imagining, which is that there’s a gigantic bust.
If there’s actually a gigantic bust, I think there’s almost no question the government will step in to buy the dip. Whether it’s buying bonds, backstopping things, or investing in compute, they will absolutely do that, because the argument they will make—which is very credible—is that this is a national-security concern.
This is the key question.
Yeah, exactly. The argument is, “Look, we need this strategically for the future. Even if the consumer industry isn’t ready for it, even if enterprises aren’t willing to pay for it, we’re going to pay for it. We’ll pre-buy the demand.” I think that is very credible, but that’s not the same thing as saying they’re going to backstop a VC raise.
Okay. That’s the take I wanted to hear. I know you have to go, so I’ll ask you this last question. I know you get asked this a lot, but you’ve come on the show a couple of times and I’ve never asked you about Naval before. Can you give us a memorable Naval moment, or something that you learned from him that’s relevant to where we currently sit in the timeline, in the market, and wherever we are in society right now?
There’s actually something he told me relatively recently. I was reflecting on the exit of Kyle Samani. Kyle Samani, of course, is the founder of Multicoin. He was super OG in the space. He entered crypto the same year that I did, and he’s now moved on to greener pastures. He’s still running Forward Industries, or whatever—which is his Solana bet—but he’s ridden off into the sunset.
I got a little nostalgic about it and wrote this little homage to him. Anyway, I was talking to Naval about this, and I was reflecting on the fact that, when I came into the industry, all these guys were the titans. Polychain, Pantera, Multicoin—these were the big dogs. I thought, “Someday, I want to compete. I don’t know how I’m going to be able to get to the level where I can compete with these guys.”
Kyle was kind of the last guy who was really just bruising every single day the same way that I am. He was out there fighting for his portfolio companies and doing the thing. So many of these other founders of these funds kind of checked out. You don’t really hear from them much anymore, and you don’t really know what they’re doing. I know these guys; they’re great dudes, but they’re not at the same energy level that they used to be.
One thing Naval told me is, “Look, I saw this so many times in so many different industries. The same thing happened with the dot-com era: early on, the smart people win early and check out. What’s left are basically the losers and the lazy people. If you just stick around long enough, you end up winning, because all the really great people retire early.”
It’s a little sordid to think about it that way, but I got to live long enough to see it happen. Increasingly, the guys who I came into this industry alongside—or even after—and thought, “I don’t know how I’m ever going to beat these guys,” will beat themselves. You barely need to do anything. You just need to stick around and have the work ethic and discipline to keep doing what you’re doing.
I think the same thing is true for you. I don’t know what your experience has been, but I’m sure when you came into this industry, it was the exact same. There were all these dudes, BitBoy and all these guys, and what did they do? They imploded, disappeared, fell into the ocean, or got lazy. It’s the same game. It has a different form, but it’s the same game. Self-discipline and consistency, as long as you’re smart and talented, always win. That was something I really took to heart.
That was beautiful. It’s a cockroach mentality. So, you’re not concerned when you see talent leave crypto? You just see opportunity.
I’m obviously concerned. It’s not good to have talent leaving crypto, but the question is where they’re going. If they’re going to fintech, then I’m really concerned.
If they're going to the banks, then I'm in tears. If they're going to AI, I'm like, "Okay, fair enough. They should. AI is really fucking important. You should go work on AI. That's totally fine. There's still a lot of people to work on crypto. We'll be okay. We got a lot of talent."
But it's a big world out there. I don't begrudge anybody going to work on AI because I do agree with them that it is the most important technology of this century. No doubt. But it's accretive to a lot of other stuff, right? Smartphones benefit from AI, crypto benefits from AI, and energy benefits from AI. All this stuff works together. There's no technology in isolation, so I don't get too teary-eyed about it.
When I see the people I came into this industry with, like my peers, when I see them leaving, yeah, it kind of hits me. It's not just that, like, "Oh, man, I love Kyle." It's also seeing somebody you're very close to, and who's a rival of yours, exit stage. That has a certain resonance to me.
But it's normal, man. It's been 10 years. We've been in the space for a long time. I've been grinding, and that's a good tenure for anybody. So I have nothing but respect for people who are leaving this industry. I think it's totally fair. It's a reasonable decision. I'm not going anywhere, and I think this industry has a lot more to go. But it's not the only game in town. That's okay. It doesn't have to be.
When's the next The Chopping Block episode?
I think it comes out tomorrow morning.
When's the last one you posted? I haven't listened to one in a couple of weeks.
We skipped a week. I was in Hong Kong last week.
I was about to say, we didn't have one last week. Okay, this is going to be a good one, then. I'm excited.
It will be a good one. Actually, we talk about Dixon. We talk about—
Unfortunately.
You can't skip it. Of course you do. You're awesome. Is there anything you want to mention or shout out before you leave? I think this was actually a really fun one and timely. We needed it.
Absolutely. Look, I got no shills. I would just say, look, stay in the game. Stay in the game. That's the number one thing. If there's any invocation I can make, it's that crypto is not going anywhere. You didn't make a mistake coming into this industry.
That being said, it doesn't mean you need to close your eyes to everything else happening in the world. You will be better in crypto if you understand the world better. And that means play around with all the AI tools, play around with OpenClaw, and play around with all the AI models. No technology lives in isolation, and crypto is no exception.
Haseeb, you're the GOAT, man. An absolute pleasure, dude. Congratulations again on the