Haseeb Qureshi 與 Kyle Samani 談 AI 風險與加密貨幣金融應用

0:00密碼學危機與 AI 數學突破

0:00早安，各位。很高興在這裡。你們都不需要介紹了，所以我們直接進入主題。這場討論的主題是「結束的開始」。現在又是一個很適合待在加密貨幣圈的時機。我們還在這裡，我們彼此認識超過十年了，也投資過各種不同的項目。我想先從昨天發生的一件事開始。昨天有一則關於 Justin Drake 的密碼學推文，你覺得密碼學真的可能被破解嗎？Haseeb，先請你談談你的看法。這到底是「開始的結束」還是「結束的開始」？Morning, everyone. Guys, great to be here. None of you need an introduction, so we'll get straight to it. The panel is the end of the beginning. It's a great time to be in crypto yet again. We're still here. We've known each other for over 10 years. We've invested in so many different things. I want to start with something that came out yesterday. There's an interesting tweet about Justin Drake about cryptography. Could that be broken? Hasib, maybe starting with you, how do you think about that? Is that truly the beginning of the end or the end of the beginning?

0:37讓我先補充一點背景。昨天 Justin Drake 發推說，他認為產業應該進入「避難模式」。所謂避難模式，如果你昨天有看到，OpenAI 發布了一大批新的數學成果。其中一項，特別是對準 Riemann 假設的解決方案，被很多人認為是近百年來最重大的數學突破之一。這是個巨大的成果。而且據說這只是 OpenAI 三波重大數學進展中的第一波。奇怪的是，這批成果裡完全沒有提到密碼學。他們有大約 700 項不同的成果，卻沒有密碼學相關的內容。密碼學應該是數學中最具經濟意義的領域，但 OpenAI 什麼都沒公開。很多人猜測，包括 Justin Drake 的假設，是 OpenAI 可能刻意保留了密碼學的突破。他們可能發現了密碼協議的漏洞，正在協調負責任的揭露，或者這些發現太具破壞性，他們還在思考如何處理。Yeah, so maybe just to give a little bit of context. So yesterday, Justin Drake tweeted that he thinks it's time for the industry to go into bunker mode. So bunker mode means that, for those of you who saw yesterday, OpenAI published a giant drop of new mathematical results. Some of them, one of them in particular, which was a solution to the quasi-Riemann hypothesis, which many people consider to be one of the biggest mathematical results in a century. Enormous result. And all of this supposedly is only one of the three drops that OpenAI is going to do of a huge amount of new mathematical progress. And the weird thing about this drop is that there's nothing in there about cryptography. They had something like 700 different results, no cryptography. Cryptography is probably the most economically significant part of mathematics, and yet OpenAI didn't release anything about cryptography. Many people conjecture, and this is what Justin Drake was hypothesizing, is that this is because maybe they're holding things back. That they have new cryptographic results. Some of them might be very bad, in that they found vulnerabilities in cryptographic protocols, but they might be coordinating, they might be doing some kind of responsible disclosure, or they just genuinely might be so damaging that they're trying to figure out what to do.

1:50所以他基本上認為，我們現在正在以人類歷史上前所未見的速度「快轉」數學進展，這可能意味著支撐區塊鏈和公開金鑰密碼學的許多假設都岌岌可危。當然，這不代表我們已經確定有什麼被破解，但事實是，我們對於區塊鏈公開金鑰密碼學底層的很多東西，幾乎都沒有真正的數學證明。我們之所以認為它們難以破解，是因為很多非常聰明的人研究了幾十年都沒找到漏洞，這算是很強的證據。但像 ECDSA 這種演算法，我們並沒有證明它一定安全，也沒證明用傳統電腦無法破解。我們很有信心，但其實不能百分百確定。如果哪天有 AI 加速的數學家發現傳統電腦能破解 ECDSA，那我們就大麻煩了。所以 Justin Drake 認為我們應該進入避難模式，也就是不要再用大額資產的地址簽署交易。如果你從不簽署交易，基本上你的資產是安全的，因為雜湊函數比公開金鑰安全得多。只要你把錢轉到一個地址且不移動，保護你的就是雜湊函數而不是公開金鑰。And so he basically claims, look, we're now speedrunning mathematics. At a pace that we've never seen in human history, and that may mean that a lot of the cryptographic assumptions underlying blockchains and public key cryptography might be in jeopardy. Now, that doesn't mean we know for certain that anything is the case, but we do know for a fact that we don't have proofs for almost anything that underlies our public key cryptography in blockchains. We're pretty sure they're hard because, you know, the usual hardness assumptions in cryptography are that a lot of really, really smart people have looked at this for decades and they haven't found anything. That's what we think of as, you know, very strong evidence that these cryptographic techniques are sound. But we have no proofs that ECDSA is sound and that it's cryptographically hard, meaning that you can't break it with a classical computer. We're pretty sure it is, but we don't know. If, God forbid, some kind of AI-accelerated mathematician figures out that there is actually some way for a classical computer to break an ECDSA key, we're in a lot of trouble. And so his claim is that we need to go into bunker mode, which means stop signing transactions on addresses that are holding a lot of money. If you never sign a transaction, almost certainly your stuff is safe because we know that hash functions are much more secure. And if you just send money to an address and you never move it, you're protected by a hash function as opposed to an exposed public key.

3:18所以問題來了，我們怎麼看這件事？我昨晚看到這則推文後也發了一則推，覺得這有點像密碼學界的「末日情緒」。我認為這是合理的。就像大家討論 AI 帶來的生存風險時，大家說的機率大概是 5%、10%。ECDSA 被破解幾乎不太可能發生，但我們還是得認真看待這個風險。密碼學本來就會隨時間演進，像 15 年前我們用的雜湊函數，現在都不用了，因為真的被發現有漏洞，像 MD5、SHA-2。密碼學一直都在遷移，這也是為什麼 Ethereum 現在不用 SHA-2 而用 SHA-3，因為 SHA-3 更安全。如果 AI 真的加速一切，十年、二十年、五十年的數學研究在半年內完成，那我們確實該更深入思考，在這個數學突飛猛進的關鍵時期，我們到底該承擔多少風險。So the question is, what do we think about this? I tweeted last night upon reading this that it's a little bit of the doomer impulse that's now come into cryptography. I think it's reasonable. Look, the probability—it's kind of like when people talk about existential risk from AI. The probabilities people are talking about are like, you know, 5%, 10%. It's almost certainly not going to happen that ECDSA is going to be broken. But we have to take seriously that, look, cryptography changes over time. You know, hash functions that we used 15 years ago, we don't use anymore because actually we have found vulnerabilities. Against MD5, against SHA-2. And so there's always this migration that's happening in cryptography. It's why today, you know, Ethereum does not use SHA-2 anymore; it uses SHA-3 because it's considered to be more secure. If, if in fact AI is accelerating everything, if, like, we have 10 years or 20 years or 50 years of math research that's happening in the span of six months. Then it probably is reasonable to start asking deeper questions about how much risk we should be taking during this critical period when math is accelerating so much faster than it has before.

4:27加密產業進化與金融本質

4:27Kyle，換你了。你一直很強調區塊鏈在金融應用上的價值，尤其是在其他投資人嘗試過各種方向後，你多年來始終堅持這個論點。你怎麼看「結束的開始」？你認為加密產業結束了什麼階段，現在又進入了什麼新階段？Kyle, maybe over to you, because you've been very vocal about, like, blockchains are, you know, very good for financial applications in a world where other investors really have tried other things, and you've really stuck to that over so many years. I want to get your perspective on, you know, if this is truly the end of the beginning, like what has ended in crypto as you think about it, and what are we entering now?

4:52我確實把很多錢投入非金融領域。有些項目還算成功，有些真的有成果，但大多數都沒有。Web3 曾經是最流行的詞，但現在很明顯，這些其實就是金融系統，並沒有超越金融本質。我們只是在現代化金融，這很明顯。當然，我們也發現了新的資本形成方式。我最喜歡的例子是 DePIN，雖然沒有我預期的那麼大，但還是有一些成功案例。除此之外，也有一些有趣的社交應用，比如 Pump、FOMO、Axiom 等等。當我想到「結束的開始」這個主題時，我覺得這其實是大家認清現實，我們不再亂丟點子，Web3 已經結束了，我們知道這就是金融。Yeah, so I definitely sprayed a lot of money into non-finance things. Some of them kind of worked, some did work, but most didn't. Web3 being the word du jour. At this point, it's been very clear that these are financial systems and not more than that. And, you know, we're just modernizing finance. That's fairly obvious. And then we've kind of found some interesting new forms of capital formation. You know, my favorite example that I'm partial to is DePIN, which like—Didn't quite get as big as I was hoping it would get, but, like, there are some success stories there. You know, beyond that, then we have just, like, interesting social things happening, right? Pump, FOMO, Axiom, all these things. When I think, you know, the point of the panelists at the end of the beginning, I think it's kind of like just a recognition that We're not throwing spaghetti at the wall like Web3 is dead. We understand it's finance.

5:572020 年時，我們都以為會有 40 種、甚至 400 種新的標準化風險交易合約。結果發現，市場其實只要現貨、永續合約、選擇權、借貸、CDS 這幾種。全球幾乎所有風險轉移都靠這八大金融合約。對我來說，「結束的開始」就是我們終於知道自己在做什麼，不再浪費時間在其他方向。加密產業現在發展很快，我想繼續討論這個話題。你離開 Multicoin 後，這應該是加密創投圈最成功的經歷之一。你現在的新角色，怎麼讓你實踐自己的投資論點？You know, in 2020, we all thought, hey, maybe there's going to be 40 or 400 new standardized contracts for trading risk. And turns out, no, we want spot, perps, options, borrow lend, CDS, like. The, like, eight major financial contracts that govern almost all risk transfer in the world are what the world wants. So I think the end of the beginning to me just means like, hey, we know what we're doing, and we're going to stop wasting time on all these other things. A lot is happening in crypto. I want to stick on this thread. You know, you left Multicoin, probably one of the most successful runs in—Crypto venture at large, how does your new role allow you to kind of express your thesis?

6:32可以聊聊 Multicoin 嗎？我很想聽。不過這場合好像不太適合。那你自己有什麼改變？我現在在 Solana 上創業，細節還不能透露太多。我很幸運能早期參與 Solana，不只是 Solana 核心，還有 Jito、Kamino、marginfi 等生態系團隊，還有 Pyth、各大交易公司等。作為創投，這些團隊需要資金、資源、策略協助，現在他們都成功了，也都賺錢了。作為 VC，我已經無法再大幅推動 Solana 成為全球無許可金融層的進程。所以我現在思考的是，怎麼「完成這個任務」？我認為現在以經營者角色，比 VC 更能推動這個目標。這也是我新創業的出發點。除此之外，我也在私募股權、併購等領域尋找機會，透過 Forward 和天使投資人身分推動這個使命。Can we talk about the multi? I mean, I would love to hear it. I don't think this is the right forum for that. But what has changed for you, I guess? You know, like I'm building a new startup on Solana. I can't talk about it too much, but—You know, I feel fortunate I was there early Solana, and not just Solana core, but all the, like, important ecosystem teams around Solana, Jito and Kamino and marginfi and a bunch of other teams necessary to make this thing all work, Pyth and all the trading firms and all these things. And, you know, like, that job is done as a venture capitalist, like. These guys needed to get off the ground. They needed money. They needed support. They needed strategic help, whatever. They're all successful and making money and stuff. As a VC, I'm, like, no longer substantially moving the needle on, like, making Solana become the global permissionless financial layer to power global finance. So a lot of how I think about my time now. It's like, well, what can I do to, like, complete the job or finish the job? And I think I can actually do that in an operating role now much more than I can as a VC. So that's my new startup, certainly you can think of it through that lens. And then I think there's opportunities in the kind of, like, the private equity side of the world, takeover side of the world, that I can also use to, like, accelerate the mission. So via Forward and via as an angel, I've been doing some other stuff like that too.

8:19投資觀念轉變與 AI 新機會

8:19很好。Haseeb，換你。Dragonfly 你們大概在市場谷底時募了新基金，六、九個月前吧？你們這幾年投資了很多不同領域。你對加密產業有什麼觀念上的重大轉變？有什麼是五年前的你絕對不會相信自己現在會這樣想的？That's great. Haseeb, maybe turning it over to you. I mean, Dragonfly, you guys raised a fund probably at the bottom of the market, you know, what was it, six, nine months ago? Yep. Again, thematically, you guys have invested in a lot of things over the years, like—What is something that you've really changed your mind on as it relates to crypto, maybe the opportunity set or what you're really focused on, that the Haseeb of five years ago wouldn't believe you talking about this today?

8:51我覺得 2020、2021 年是我們所有人加密瘋狂的巔峰。我們都深信不疑，完全投入。回顧 2016、2017 年，我記得 Kyle，我第一次見你是在 Expo 辦公室。那時你們的基金才剛起步，還在發推文、寫關於 XRP 分叉的文章。那時候一切都很小眾、很瘋狂，沒什麼人看好。我們都是怪咖，沒人覺得我們做的事有前途。其實你現在還是怪咖啦。哈哈，我還是怪咖。現在這些人都變得很酷、很有份量，但當時我們都冒著很大的職涯風險。其實我們本來可以去很有名的產業，但我們選擇當加密怪咖。Yeah, so I think, like 2020, 2021 was peak crypto psychosis for all of us. Where we were just totally convinced, hook, line, and sinker, that—I think it's important to rewind the clock a little bit to, like, 2016, 2017. I remember, Kyle, I first met you, I think in the Expo or the Expo office. Yes. The Nerdal Capital guys. Yes. And at that time, you were still—it was a fledgling fund. You guys had just gotten off the ground. You were, like, tweeting about publishing a bunch of—You know, posts about, I think it was like XRP fork or something. There was like all this crazy stuff, and none of it was working. It was all tiny. And, like, we were all weirdos, okay? We were weirdos. Crypto was weird. Nobody thought what we were doing— I see, but I hate to tell you that you're still weird. I was gonna say. Okay, I'm still weird. These guys are cool and, you know, respectable now. But, like, at that time. All of us were taking huge amounts of career risk. Like, we were all people who could have had, you know, could have been in natural prestigious industries, but instead we were crypto weirdos.

9:542020 年，突然全世界都對我們大喊：「你們完全正確！」我們做的每件事都對，加密貨幣會吞噬一切。全世界都同意，甚至比我們想像的還大，會顛覆文化、每個名人都會玩 NFT、上電視。我們真的很想相信這一切，因為誰不想呢？其實在大 VC 裡，我算是最抗拒這種說法的人。我一直不相信加密會顛覆遊戲產業，也不覺得 NFT 不是泡沫。這些東西爆紅時，我和團隊、其他 VC 爭論了好幾個月。最後我也覺得，也許我只是太老、不夠潮，應該接受 NFT 會吞噬世界、大家都會活在區塊鏈上、DAO 會接管一切。結果這些都沒發生。DAO 一團亂，加密遊戲都很爛，NFT 雖然還在，但只是個奇怪的小角落。And 2020, all of a sudden, the entire world screamed at us, You were fucking right. Everything you decided to do was absolutely right. Crypto is going to eat everything. Everybody in the world now agrees. And you know what? It's going to be even bigger than you thought. In fact, it's going to take over culture. Every celebrity in the world, NFTs is going to be on TV. Like, you were more right than you ever thought you were, and it's going to eat every single industry. And we really wanted to believe it. We really wanted to believe it because, of course, like, why would we not want that to be true? And so, you know, I was probably among most of the big VCs, I was the most resistant to the idea. That crypto is going to take over gaming, that NFTs weren't a bubble, that, like, and I fought it for months when this stuff really started exploding. I would get into big arguments with both my team as well as a bunch of other VCs. And eventually I too was like, you know what, I'm probably just wrong. I'm probably just, like, not young and cool enough to get it. And I just got to get with the picture that NFTs are going to eat the world and, you know, we're all going to live on the blockchain and DAOs are going to take over everything. And turns out, none of those things happened. DAOs are a mess. Crypto games are all terrible. And NFTs, I mean, they kind of exist, but they're kind of this, you know, weird corner of the—

11:07我們現在都叫它們 RWA 了，Kyle。RWA 算是 NFT 的繼任者。投資其實很講究時機，什麼時候是對、什麼時候是錯，這些實驗有時要花幾十年才看得出來。如果你看科技演進，有時「太早」和「錯誤」根本沒差別。有些實驗要花幾十年才會成熟。We call them RWAs now, Kyle. RWAs are now the successor, I guess. So much in investing is timing, right? What can be true, what can be considered wrong, and—Yeah, there's no difference between too early and wrong. And some of these experiments take multiple decades, so to speak, if you look at the evolution of technology.

11:29AI 投資與加密產業新常態

11:29我們現在更緊密了。我剛剛在後台聽你說，你們基金大約有 15% 現在投資 AI。你怎麼看這件事？這會不會讓人產生存在危機？我們怎麼看加密的機會，和 AI 這種技術相比，還有足夠吸引力嗎？我們都參加過很多討論，大家都說這是最有趣的產業，也許這裡才是我們的優勢。但現在有人把火箭送上太空，有人用 AI 幫 ALS 病患恢復語言能力，這些都因 AI 加速。你會不會照鏡子問自己：「我還想留在這產業嗎？」Yeah, we're more connected. I heard you say backstage, roughly 15% of your fund is now investing in AI. How do you think about it? Is there a conversation there around just existential crisis as—How we think about crypto, the opportunity set, is this sufficiently interesting as a piece of technology versus something like AI that, you know, I think we've all been in panels where we've said this is the most interesting thing that we could be doing, and maybe it's this is where I have an edge. But it's, you know, there's people putting rockets. In space and regaining speech for ALS patients, a lot of this accelerated by AI. Does that, you look at yourself in the mirror, you say, shit, is this the industry I still want to be a part of?

12:11對我來說，答案非常明確，就是「是」。我仍然認為加密是過去幾十年最科幻的事物之一。很明顯，AI 是過去一百年最科幻的技術，未來還會繼續改變世界。我剛剛談密碼學危機，就是因為 AI 的進展。當然，AI 已經滲透到所有科技領域，現在大家投資 AI 實驗室，但這種純 AI 投資的時代很快會結束，就像加密投資一樣。現在純 Layer-1、zk 網路這種基礎建設已經建得差不多了，接下來要做的是擴展和改善。當然，量子電腦等新挑戰還會來，但這些改進多半會在現有基礎上進行，而不是從零開始創業。So, I mean, for me, the answer is unequivocally yes. I still think that crypto is one of the most sci-fi things that has happened in this, in the last couple of decades. It's clear that AI is the most sci-fi thing. That has happened in the last century, and it's going to continue to change the world. And we were just, you know, the thing I was just talking about with cryptography is precisely because of what's happening in AI. Now that said, the reality is that AI isn't everything technology, and the way in which AI surfaces, okay, there's investments in labs and so on, but, like, we're probably going to be exiting that period relatively soon where there are AI investments in the same way that, you know, crypto investments. Are becoming less and less of what we are investing in at any given time, because, like, there's just not that much room for pure play layer ones or zk networks or this kind of thing. The reality is that we've built most of the infra. And now the infra does need to scale. There's a lot of, you know, we're talking right now about, you know, obviously quantum is coming, and all of these things will need to be improved, but they'll probably be improved in place rather than improved. by a new startup that starts, you know, from zero.

13:16所以我認為加密產業也在發生類似的事。加密正像自來水一樣滲透到各個領域，每家金融機構、每個新創金融科技公司，現在的簡報裡幾乎都會提到穩定幣。這不是因為他們想靠穩定幣募更多錢，而是因為這樣真的能帶來成本或執行優勢。同樣地，現在新創公司如果沒提到 AI，大家會覺得你到底在做什麼？新創有機會從 AI 出發，沒有舊公司的包袱，但這不代表它就是 AI 公司，只是「公司」而已。我們看到加密也在發生這種現象，現在我們投資的項目，已經不再只有加密原生 VC 能投，其他投資人也能參與、也懂得評估。以前三、四年前，加密新創的股東名單全是加密 VC，現在不是了。很多加密好案子，其他人也能投，也有能力判斷價值。So I think for crypto, what we're also starting to see is that crypto is becoming a little bit like this too, is that crypto is just getting into the water supply and it's going everywhere, where every single financial institution, every single fintech, like you look at a new fintech pitch and almost every single one has stablecoins somewhere in there. And it's not because they think they're going to raise more money with stablecoins. It's because that is actually what you do now in order to get a cost advantage or an execution advantage with a fintech. In the same way, if you're building any new startup, there better be some talk of AI in there. Otherwise, like, what are you doing? If you're a brand new startup, you have the ability to build AI first and not have all the baggage of an old company. But that's not an AI company. That's just a company. You know, so I think we're seeing that happening to crypto as well, that more and more of the stuff that we invest in, we are not the only people who can invest in that company, right? You know, back in the day, like three, four years ago, every single person on a crypto company's cap table was a crypto-native VC. Today, that's actually not true. Most of the good deals in crypto, other people can invest in it as well and are equipped to underwrite it.

14:38加密貨幣終局的指標與大眾認知

14:38有加密貨幣投資人，也有一般投資人，因為這些東西對一般投資人來說很容易理解。但加密貨幣和我們討論的所有概念，都是他們作為挑戰者的優勢之一。假設我們十年後還在這裡，希望真的是這樣。你怎麼判斷這是終局的開始？不是說加密貨幣會消亡，而是說加密貨幣真的已經嵌入、融入，成為金融未來的主流。我們之前在後台聊過，你很投入這一塊。你怎麼知道加密貨幣已經比現在更進步了？There's crypto investors, but there's also generalists because these things are very legible to regular investors. But crypto and all the concepts that we're talking about are part of that advantage that they have as Challengers. Say we're here 10 years from now, hopefully that's the case. How will you know that this is the beginning of the end? Like, what is the end state for you that you could say, not in the end of, like, crypto is going to die, but more so that crypto is truly kind of Embedded, immersed, become truly the future of finance. You know, we talked about Backstage. You know, you're very involved there. Like, how will you know that crypto is, you know, much farther along than it is today?

15:27我認為最簡單的檢驗標準，就是當大眾不再覺得加密貨幣是詐騙。這就是最簡單的檢驗。我覺得這個時刻大概還有五年，可能五到十年之間。要達到這個目標需要什麼？像今天早上，三星宣布要把 Solana 錢包內建到 8200 萬支手機裡，軟體更新一週後就會上線。我猜其他 Android 廠商也都在想類似的事情，Apple 可能還要一兩年才跟上。現在所有主要的金融機構都在參與，像 CME 也來了，還有 Hyperliquid、Backpack、Coinbase 等等。整個金融體系的現代化正在發生，這個過程大概還需要五年左右。I think probably the simple litmus test for that would be when the general public doesn't think it's a scam. Like, that's the simple litmus test. I think that point's probably five years away, somewhere between five and ten. What needs to happen for that to— Oh, I mean, just like this morning, right, Samsung announced they're building Solana wallets into 82 million phones. The software update goes live in a week. I'm assuming the other Android manufacturers are all thinking about similar things. Apple's probably a year or two behind, right? Obviously, we've got all of the major, you know, CME is here, literally a token. We've got Hyperliquid, Backpack, Coinbase, all these guys, right? Like, so all of this modernization of the financial system is happening. That process takes probably, you know, five-ish more years, roughly. Yeah.

16:20鏈上證券與大眾採用的障礙

16:20那我追問一下，過去兩年你認為加密貨幣領域最有趣、最加速大眾認知或實際應用的事件是什麼？你這些年看過很多事情。我最興奮的是鏈上證券，我本身也是 Backpack 的董事會成員，當然對 Backpack Securities 很有偏好。現在大概有 1.5 億個 Solana 錢包，像 Phantom、MetaMask、Solflare、Backpack、Binance、OKX 等等。很多人原本無法或不願意透過其他金融服務買美股，但過去三四個月，他們現在可以買到大約 60 到 70 檔 Backpack 證券，這個數字很快還會大幅增加。我覺得這很令人興奮，因為你可以投資 SpaceX、Amazon 等等你想要的標的。I guess maybe as a follow-up to that, from your perspective, what is the most interesting thing that has happened in crypto maybe over the last two years that has really accelerated either the perception of crypto or just the actual traction that is really surprising? You've seen a lot of stuff over the years. I'll say the thing I'm the most excited about is Securities on-chain. And I'm on the board of Backpack. I'm obviously very partial towards Backpack Securities. You know, right, you've got, call it 150 million Solana wallets out there probably: Phantom, MetaMask, Solflare, Backpack, Binance, OKX, etc. You know. A lot of those people either couldn't or wouldn't be able to buy U.S. equities via whatever other financial services they may have access to. In the last three or four months, they have access to now something like 60 or 70 Backpack securities, and that number is going to increase pretty dramatically here in the not too distant future. And I think that's very exciting that, like— You can unlock access to SpaceX, Amazon, whatever, thing of choice.

17:37我一直不喜歡比特幣，現在還是不喜歡，因為它沒有經濟生產力。我喜歡有生產力的資產。整體經濟，尤其是美國經濟，在 AI 時代下，產生了史上最有生產力的公司。我希望大家能接觸到這些資產。我不知道加密貨幣形象什麼時候會改變。我確實擔心加密貨幣的負面標籤還會持續一段時間，因為過去兩三四年，大家對加密貨幣的情緒很強烈。尤其是年紀較大的人，特別是在美國，會很難放下這種成見。這很大程度上是因為在美國和其他富裕國家，大多數人根本不用加密貨幣。如果有用，通常只是放在 Coinbase 這類帳戶裡而已。I've always hated Bitcoin, still hate Bitcoin because it's not economically productive. Yeah. And, you know, like, I like productive assets. And then, you know, the economy at large, and really specifically the American economy, especially in the AI era, is producing the most economically productive companies of all time. So, like, I want people to access them. I don't know how long it's going to take for that perception of crypto to change. I think, I do worry that there's a stank on crypto that's going to continue for quite some time because people are, you know, the emotional valence that has attached to crypto over the last, call it, two, three, four years. is really strong. And I think for, especially for older folks, I think it's going to be very difficult for them to let go of it, especially in the U.S. And in large part, this is also because in the U.S. and in a lot of the really rich countries, most people don't use crypto. You know, if they do use crypto, it's to own some in a, you know, Coinbase account or something like that, but, you know.

18:42美國人其實不會用穩定幣。雖然 99.8% 的穩定幣都是美元計價，但主要使用者其實都在海外。所以對他們來說，根本不需要了解加密貨幣。就像大多數人根本不知道 CME 是什麼，雖然 CME 幾乎壟斷了衍生品交易，而衍生品又是資產交易的主流。現實是，大多數人根本不需要知道這些事，也沒那麼重要。我覺得對比特幣、ETH 和一些大型資產來說，這些形象還是重要的，因為零售投資人還是主要持有者。但對於加密貨幣、鏈上資產、穩定幣等等，我覺得大多數人怎麼想其實沒那麼重要，就像大家不會在意 DTCC 或 CME，甚至大多數人討厭銀行，但還是會用銀行。Americans don't use stablecoins, obviously. I mean, 99.8% of stablecoins are U.S. dollars, but they're not used by Americans. They're used overwhelmingly overseas. And so as a result, it's kind of a—they don't really need to know. They don't really need to care about, you know, what crypto, in the same way that, like, you know, most people don't know what CME even is, even though it has a near monopoly on the trading of derivatives, which is the majority of how assets are traded. In financial markets. So the reality is most people don't need really to know, and it kind of, I don't know that it matters so much. I think for Bitcoin it matters, and for, you know, ETH and some of the mega cap assets it matters because obviously retail predominantly are the owners at the margin of most of these things. But for crypto, for, you know, tokenized assets, for stablecoins and so on, I kind of, I don't know that it's going to matter what most people think about it in the same way. Doesn't matter what most people think about DTCC. Doesn't matter what most people think about CME. Most people hate banks. They still use banks, you know.

19:39所以我們心裡要把這兩件事分開。現在大家討厭 AI 公司，討厭 AI 執行長，可能比任何其他產業都更討厭。但 ChatGPT 卻是史上成長最快的商業應用，App Store 排行榜也常常被 AI 應用佔據。消費者總是有這種矛盾心態，這就是現代科技的現實。And so I think we probably have to separate those two in our minds about, you know, right now people hate AI companies. They hate AI CEOs, I think more than like any other category of anything. And yet ChatGPT is the fastest growing commercial application. You've got, you know, this. The app store charts are constantly topped by AI applications. So there's always this two-mindedness to consumers, and I think that's just part of the reality of technology today. Yeah.

20:08加密貨幣將成為無感基礎設施

20:08另一個層面是，隨著加密貨幣逐漸淡出人們的討論，技術分發和推廣還有很多工作要做，也許我們也該停止一直用「加密貨幣」這個詞。我覺得未來大家會不再討論 AI，不是因為我們把它改名叫 SI，而是因為它會像網際網路一樣融入生活。大家現在也不會說「我用網際網路」，而是直接說「我用手機上的這個東西」。甚至大家也不太討論「App」這個詞，而是直接說用哪個服務。同樣地，未來大家也不會再提加密貨幣，「crypto」這個詞只會讓人想起早期的詐騙、爆雷、政治醜聞等等。但大家會說：「我匯錢到國外很便宜」、「我用這個 App 做這件事很快，完全不用擔心」。我覺得五到十年後，大概就是這種感覺。I guess the other element of it is that, you know, I think as crypto fades into the background, there will always be a lot of work for us to do in Improving the technology from a distribution perspective, and probably also stop talking about it as crypto. I think crypto will be one of these things that, like, I suspect that people will stop talking about AI, not because we rebranded it to SI, but because, like, it's just going to fade into the background. The way people don't really talk about, you know, the World Wide Web. You know, they just talk about, oh, they use this thing on my phone. You know, people don't really talk that much about apps anymore. They just talk about what the app is. And I think in the same way, people probably will just not talk about crypto. Crypto will just be this kind of ugly word. It reminds us of the early days when there were a lot of scams and blowups and political scandals and blah, blah, blah. But they'll just be like, yeah, when I send money overseas, it's really cheap. And when I, you know, use this app to do this thing, it's really fast and I never have to think about it. And I think that's probably more what it's going to feel like in five to ten years. Yeah.

21:1124/7 鏈上股票交易的現實與瓶頸

21:11Kyle，回到你剛剛說的，你認為大部分核心基礎設施都已經建好。深入一點來看，像鏈上股票這一塊，我們都認為很有潛力，真正要打入主流。現在到了嗎？我們會走向 24/7、365 天交易嗎？要達到這個目標還缺什麼？速度會有多快？瓶頸在哪裡？其實現在在 Solana、Robinhood Chain 上，大家已經可以 24 小時交易股票。但現實是，流動性很差，沒有人會想在 Amazon 下 50 萬美元的單卻滑價 100 個基點，這完全不合理。Kyle, just going back to something you said earlier around you think most of the core infrastructure has been built. Just double-clicking on that, I mean, tokenized stocks, huge part of where I think we all agree is a lot of potential talk about truly, really reaching mainstream. Are we there yet? Like, are we going to move to 24/7, 365? And what needs to happen to get there? How quickly do you think we get there? What are the bottlenecks? I mean, look, like on Solana today, Robinhood chain today, people are trading equities 24/7. Now the reality is that it's, like, pretty crappy because the liquidity is not there. You know, like no one expects to trade Amazon and to have 100 basis points of slippage on a $500,000 order. That doesn't make any sense.

21:58我認為只有一條路，就是要把流動性橋接到 NYSE 和 Nasdaq。這樣你下 1,000 萬美元的 Amazon 單，滑價只會有 2、5 個基點。只要做到這點，很快就會自然擴展到 24/7 且有足夠流動性的市場。但這一塊一定要搭傳統金融的便車。要在鏈上原生打造 Citadel、Jane Street、HRT 那種流動性，根本不切實際。即使是在 Binance、Coinbase、Backpack 這些平台也不可能。一定要靠這些傳統巨頭。只要這件事發生，24/7 交易的成長速度就會出現跳躍式變化。I actually think there's only one path to get there, and that path is you have to bridge liquidity to NYSE and NASDAQ so that if you place a $10 million Amazon order, like it should be, you know, two bips, five bips, whatever that might be. And then once you do that, then I think. It very quickly naturally spills over into 24/7 with still substantial liquidity. But you have to piggyback off of TradFi on this one. The notion that we're going to recreate, you know, Citadel, Jane Street, HRT-level liquidity natively on any of these chains is pretty silly. Or even, for that matter, on Binance or Coinbase or Backpack, that's not going to happen. You got to piggyback off of those guys. And I think once that happens, the rate of growth of the 24/7 trading, that's kind of a step function change.

22:54穩定幣、預測市場與 AI 代理人的新機會

22:54看來你很看好 RFQ。我確實很喜歡 RFQ，也很喜歡 Backpack Securities。Haseeb，換你來談談你怎麼看加密貨幣的機會。以你們這種規模的基金來說，過去投資回報多來自基礎設施。如果這些都建好了，未來你最期待什麼？你剛剛提到 AI 是很大的一塊，像 Venice 等等，那你怎麼看這些機會？這一輪基金和以往有什麼不同？Sounds like you're bullish RFQ then. I definitely like RFQs. Okay. I like that Backpack Securities too. Fair enough. Haseeb, maybe over to you. Just when you think about the opportunity set in crypto. Deploying a fund of your size, if a lot of this stuff is, I think a lot of the returns for investors historically have come from infrastructure. If a lot of that has been built, when you look towards the future of the next stage of crypto, like what are you most excited about? I know you mentioned, you know, AI is a big part of that, stuff like Venice, but how do you think about the opportunity set? What parts of the layer? What is different, perhaps, in this fund relative to all the rest? Yeah.

23:35我們投資的主要類別之一，就是穩定幣和支付的成長。像 Rain、Mesh 這些公司，這條線整個產業都在往上衝，任何參與其中的人都看到數字爆炸成長。這是一個很大的類別。我覺得外匯（FX）也是一個正在成長的新類別，因為穩定幣讓這件事變得可能。第二個類別是預測市場，Polymarket 和 Kalshi 是目前最突出的贏家，但市場還有很大空間。這兩家在美國競爭，但美國只是其中一個市場，全球還有很大成長空間。預測市場這個概念很明顯會在其他地方複製，只是各地法規還有很多障礙，要做很多地面推廣。但大家都想押注文化、政治、重大議題，也想從市場獲得專家和名嘴給不了的資訊。我們還在全球推廣的早期階段，這是我們很關注的方向。So if you look at the big categories in which we've deployed dollars, one of the really large and obvious categories has been the growth of stablecoins and payments. So, you know, companies like Rain, like Mesh, you know, increasingly this is the thing that you can just see the line is going up and to the right for the entire industry. Anybody who's in the flow of this is just seeing numbers are exploding. So that's one category. And I think also FX is this growing category as well that's starting to be enabled by using stablecoins. Second category, I'd say prediction markets. Obviously, Polymarket and Kalshi have been the standout winners, but I think it's still a lot of white space. Polymarket and Kalshi are going— You know, toe to toe in the U.S., and even in the U.S., they haven't saturated the market. There's still a lot of growth among both of them. But the U.S. is only one market, and it's a big world. And I think it's very obvious that the ideas that work in the U.S. are also going to work elsewhere. The idea of prediction markets are just right. It's, like, very clearly an idea that is going to replicate in other places. Now there's regulatory— Barriers. There's going to be a lot of work to, like, build the, do the ground game in these different places. But people want to bet on culture. They want to bet on politics. They want to bet on the big questions of the day. And they also want to look to markets to give them information that experts and talking heads can't. And so I think we're still in the early stages of that global go-to-market, and that's a place where we're paying a lot of attention of people taking different angles of attack. On that problem globally.

25:08第三個重點是加密貨幣和 AI 的交集。我昨天有講一場關於 AI 代理人和加密貨幣的主題。年初大家還在討論 AI 代理人會用加密貨幣，我也是其中一個提出這想法的人。現在我們真的有 AI 代理人了，像 Muse、Instinct、Hermes Agent 等等，而且發展非常快、非常主流。但很明顯我們錯了，他們其實用的是信用卡，這樣也沒問題，效果很好。幾乎每個國家的傳統支付業者都意識到這很重要，已經想辦法把這塊做好，表現也很不錯。我認為 AI 代理人會在需要用加密貨幣的地方用加密貨幣，比如跨境匯款，因為這是最快、最便宜的方式，很多外匯通道也比銀行便宜。但 AI 代理人一定要用 USDC 或穩定幣這個想法是錯的，我們要認清這點。And then the third thing I'd say, a lot happening at the intersection of crypto and AI. We're still, you know, I gave a talk yesterday about AI agents and crypto. People used to talk about, you know, even just beginning of this year, the idea that AI agents are going to use crypto. I was one of the people talking about that. We now have AI agents. We've got Muse and Instinct and Hermes Agent and all this stuff, and it's now going extremely mainstream very rapidly. And very clearly, we were wrong. They're not using crypto, they're using credit cards, and it's totally fine. It's working very well, and in almost every single country, the traditional payments players have realized that this is really big. They've basically cleared the boards and figured out how are we going to do this in a way that is robust, and they've done a pretty good job. Now, I do think AI agents will use crypto where humans use crypto. If you want to move money cross-border as an AI agent, you will do it the same way we do it, which is the fastest way to do it is using crypto, and you're going to get better pricing on a lot of FX corridors doing it that way than doing it with a bank. But— This idea that AI agents just have to use USDC or you have to use stablecoins, this was wrong, and we should realize that this is wrong.

26:23AI 改變加密貨幣 UX 與價值流動

26:23我認為 AI 會徹底改變加密貨幣的介面。現在用加密貨幣很容易被駭、出錯、被詐騙或釣魚。AI 會成為你和你自己的愚蠢之間的智慧層，把很多加密貨幣 UX 的難題變得簡單。未來你不用再自己研究怎麼跨鏈、哪裡收益高、哪些協議被駭過、Twitter 上有沒有什麼壞消息、Stream Finance 是不是突然爆雷等等，AI 都會幫你搞定。每個人都會有自己的「加密小幫手」，你只要下指令，它就幫你處理所有麻煩事。這會徹底改變我們使用加密貨幣的體驗，也會改變大家對於鏈的黏著度和使用者鎖定的想像。像 FOMO、NEAR Intents 已經讓大家不再只活在單一鏈上。對 FOMO 使用者來說，不管迷因幣在 Robinhood、Base、BNB Chain，反正按個鍵就能買，這種體驗會普及到整個加密貨幣圈，也會改變價值流動的假設。我認為很多新創會在這個轉型過程中扮演關鍵角色。So I do think that AI is going to change the interface of crypto. Is that the way in which we use crypto and all the footguns that attend to crypto, in that it's easy to get hacked, it's easy to make mistakes, it's easy to end up getting scammed or phished or whatever. AI being this intelligence layer that's going to sit in between you and your own stupidity, and making a lot of the UX challenges of crypto much simpler, that is something that I think is going to totally change the way in which we interact with crypto. The point-and-clicking experience of, okay, I'm going to go shop around for the best way to bridge assets over here, and I've got to figure out, you know, where is the best place to get yield and— Which protocols have been hacked, and I got to read the Twitter and make sure that nothing bad happened this week and that there's no weird Stream Finance, this thing is like randomly in debt and I didn't know about it. All of that stuff, AI will do it for you. Like there's no, it's like everyone will have their own crypto guy, and their crypto guy will do all the annoying stuff for you and you just bark orders at it. That's going to change a lot about how we experience crypto. It's going to change a lot about the sense that we have that chains. Have users with lock-in, with commitments to that chain. You already see it with FOMO. You see it with Near Intents, is that the sense that, you know, people live on a chain and they'll use the apps on that chain, that experience is already gone for people who are living on FOMO. For people living on FOMO, if your meme coin's on Robinhood, if it's on Base, if it's on BNB Chain, who cares, right? You just click the button and you buy the asset. That is coming to all of crypto, and I think it is going to really change a lot of the assumptions we have about where value flows. And I suspect many startups are going to end up being very pivotal in how that transition takes place.

28:02慢速金融與產業未來展望

28:02Kyle，最後一題，你現在最興奮的是什麼？你一直都很敢講，也不怕跟主流不同。你現在有沒有什麼最非共識的觀點或投資想法？加密貨幣最熱門的應用一直都是短線、快節奏的，像各種交易終端、Axiom、FOMO、Pump 等等，甚至現在很多新產品也是給短線、動能交易者用的，大家都在看 RSI、技術指標。我認為未來有機會用加密貨幣基礎設施打造「慢速金融」的生意，而不是「快速金融」。我對這個方向很興奮，我現在做的事也跟這個有關。Kyle, last question to you. What are you most excited about? I mean, you've always been unafraid to be vocal and perhaps non-contentious. What is perhaps your most non-consensus view at the moment that you're either investing in or idea? You know, crypto is always, a lot of the most popular crypto apps have always been kind of very short horizon in nature. You see all the trading terminals and then the Axioms and the Fomos and the Pumps, but then— Even like a lot of the newer ones that are built on Hyperliquid or elsewhere tend to be like, I'm a swing trader, momentum trader. I'm looking at RSIs and these other kinds of technical indicators. And I think there's going to be an opportunity to build businesses that leverage crypto rails focused on slow-twitch finance instead of fast-twitch finance. So I'm pretty excited about that and what I'm building is in that. In that direction. Great.

28:59各位先生們，希望十年後還能在這裡見到你們，到時候我們討論的就不再是「終局的開始」，而是產業的持續發展。謝謝大家。Gentlemen, hope to see you here in 10 years, and we should be talking hopefully not about the beginning of the end, but just more of a continuation of the industry. So thanks, thanks very much.