Jennifer Sanasie and Renato Mariotti sit down with Haseeb Qureshi, Managing Partner at crypto venture firm Dragonfly at the Avalanche Summit. They discuss the collision of AI and crypto. Qureshi lays out three paths for AI — chaos, centralization, or alignment — and argues government restrictions will only accelerate once a real catastrophe hits, like an AI agent breaching a power or water plant. He explains why the CLARITY Act really failed, pointing to the Iran war and Trump's slipping approval as the cause beneath the cause, and where his crypto predictions stand now — from a DeFi insider-t...
What Haseeb Qureshi said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Haseeb Qureshi, managing partner of Dragonfly, discussed the politicization of AI frontier pacing, seeing three possible futures: chaotic uncontrolled models, centralized lock-downs, or aligned open models, with the latter preferred but requiring solved fundamental problems. He mentioned Dragonfly's investment in Venice.AI, a crypto wallet-based access to unrestricted AI. On the failed Clarity Act, he attributed its defeat to ethics provisions aimed at Trump, the Iran war's impact on Trump's approval, and three Republicans voting against it, foreseeing a possible renegotiation in 2027-2028. He predicted government restrictions on large AI companies after a non-existential but serious catastrophe, comparing to social media regulation. He partially vindicated his DeFi insider trading prediction, citing Robinhood indictments and ongoing investigations, expecting a bigger case by year-end. On Bitcoin, he updated his previous call to $100k, suggesting a rally likely after Anthropic IPO employee unlock in Q1-Q2 2027.
Key takeaways
Clarity Act reconstitution possible in 2027-2028, but overwhelmingly unlikely this year.
Bitcoin has a 'very good shot' to 100K, but rally likely after Anthropic IPO employee unlock in Q1-Q2 2027.
DeFi insider trading prediction partially vindicated; expects something bigger by end of year.
There's going to be some kind of AI agent gets into a power plant, gets into, you know, a water plant, gets into something that really causes actual damage. It's not going to be existential, but it will be serious enough that people will start to say, 'Oh we do not have control of this. We do not have this handled.'
Haseeb Qureshi joins us. He is managing partner at crypto venture firm Dragonfly where he invests at the intersection of crypto markets and increasingly AI. Hello.
I've been following you on X. I mean I follow you on X. I've been following the conversation the last couple of days and some of your commentary around where we are with AI and the things that we need to be aware of has been interesting to me. So I want to start the conversation there. Just talk to us a little bit through an investor's lens about what we should be looking at when it comes to the development of AI.
Well, we're in this weird moment right now. Obviously there's this negotiation about a slowdown or a pacing the frontier as it's being called. It seems like all of a sudden this has gotten politicized. So now it's almost a little bit like masks where what was once a very technocratic conversation has now become a political conversation. Trump and Jensen Huang are very clearly like this is a hoax. We should not be focusing on this and said it's full steam ahead. The people at the labs, the scientists overwhelmingly they're like hey we're realizing that we don't really know what these models are doing. We find it very hard to monitor them. They seem to be doing things that we don't want them to be doing or we didn't program them to do or train them to do. So we're in this very weird moment right now where it's hard to project what 3, 4 years in the future is going to look like.
There's kind of two paths. One path is basically something that ends up being quite chaotic where models just start really being kind of let it rip and we don't really control them and we don't really have good guard rails on the kind of damage that they're doing out there even regardless of our intent. Second path is that we basically get enormous amounts of centralization where the labs get locked down, GPUs get locked down and governments basically tighten the guardrails especially if there's some kind of catastrophic event that scares everyone and freaks them out. Third path is basically that we get alignment under control. We have pretty good confidence of how these things work and probably in that world you actually see less centralization in the labs, more open source, more open models. That's the world I think everybody prefers. It's a world with a lot of competition and a lot of openness. But right now, we have to solve a lot of fundamental problems before I think that world is going to be super plausible.
It's not really clear today what the economics longterm of AI are going to look like. There's so much stuff that's making this stuff cheaper and making it more, you know, uniformly accessible. We're investors in Venice.AI, which has been one of the big winners in the crypto cycle over the last month. They do a lot to make private unrestricted AI available to anybody even without having an account. You can just log in with a crypto wallet and pay with stable coins and get access to frontier models from anywhere in the world. But this is becoming now a political fight. It's not just technology. It's the whole world is having a conversation about it.
So, it's interesting. There's kind of the cause and the cause underneath the cause. The cause obviously has been that Trump and the ethics provision within Clarity was the big stumbling block for Democrats. Right now, the things that Trump did to warrant this ethics provision being this big political fight were things that happened last year. So, it was World Liberty Financial, it was Trump token, and obviously these two things going down and him making a lot of money. Nevertheless, are a big part of the reason why it plays politically to go after Trump using this as ammunition. And the Clarity Act kind of became his political football.
In reality, Democrats didn't really get what they wanted, right? So to the extent, you know, the banks that wanted to fight back on some of the stable coin provisions, they didn't get what they wanted, right? The loophole in Genius Act is still there. So the banks lost. Democrats wanted some restrictions on Trump. There are no restrictions on Trump. Clarity didn't pass. So therefore, they lost. Everybody loses, right? So this is not a good outcome for pretty much anyone. So I do think there's a good chance that clarity will get reconstituted in some form. There were a group of six Democrats today that came out and made a statement that they still want to see some form of a renegotiation of clarity that maybe 2027, 2028 can come back to the table this year. Overwhelmingly likely it's not going to happen.
But I do think like the real thing I think that ultimately caused this, remember three Republicans voted against clarity. So it wasn't as though it was just Democrat. Obviously, most Democrats were against it, but I think the cause underneath the cause was the Iran war. If Trump had not started the Iran conflict, his approval rating would be much higher. It wouldn't be as animating of an issue to Democrats to go after Trump specifically rather than to go after the general Democratic talking points and he would have a tighter control over Republicans and he'd be able to make some concessions and play more, you know, political football. The fact that Trump has lost so much support even within the Republican party is a big part of the reason why he couldn't push through what otherwise in February was a relatively uncontroversial part of his campaign. Most Democrats were on board with some version of the bill, but increasingly Trump losing that political center on so many things has ultimately caused I think this to be part of the collateral damage.
15 more exchanges in this transcript
Sign in free to read the rest of this interview. No card required.
Qureshi, H. (2026, September 24). Why an AI "Catastrophe" Could Trigger a Regulatory Crackdown [Interview transcript]. CoinDesk. CEOInterviews.AI. https://ceointerviews.ai/interview/2949763/
MLA
Haseeb Qureshi. "Why an AI "Catastrophe" Could Trigger a Regulatory Crackdown." CoinDesk, 24 Sep. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/2949763/.
BibTeX
@misc{qureshi2026_2949763,
author = {Haseeb Qureshi},
title = {Why an AI "Catastrophe" Could Trigger a Regulatory Crackdown},
howpublished = {Interview transcript, CoinDesk. CEOInterviews.AI},
year = {2026},
month = {sep},
url = {https://ceointerviews.ai/interview/2949763/},
note = {Speaker-attributed transcript with timestamps}
}
