[REDACTED] 2024 | Distributed Rebellion: A Thesis on Crypto x AI From Delphi Labs

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   thank you really excited to be here today I'm going to talk about our thesis for crypto AI we called it the distributed rebellion and that'll make a little bit more sense to you as I get into it but first I want to introduce you to kind of what we've been up to at Deli at this intersection of crypto and AI at Deli digital our research company we've published four really comprehensive reports covering the entire landscape of decentralized AI we've open- sourced these so they're

2. 0:29

   aable to anyone highly recommend you go check them out we'll continue to cover this sector very deeply at Deli Ventures we've been investing in this space for quite a bit of time uh we're up to about 20 Investments and we were one of the earlier funds to start underwriting in this sector and we continue to actively deploy capital and Lead rounds and then just recently we concluded our crypto AI

3. 0:55

   month we had about 15 events with a lot of the thought leaders in the space covering a whole bunch of the relevant topics those were all recorded they're available online again something you can check out I guess just to sum it up we had a lot of conviction in crypto aai very early on and that conviction has only since grown and then at Deli Labs we're really excited to partner with near on this crypto AI accelerator cohort we've got eight great teams that we're working with right now um you'll hear about many

4. 1:28

   of them throughout this week and couldn't be more excited we're going to have a demo day next month that showcases all these projects uh again very excited for what we have going on and then before we started this accelerator cohort we are very thesis driven at what we do and where we spend our time at Deli Labs so we took our time to think about crypto and AI from a builder perspective and that's where we developed our thesis the distributed

5. 1:57

   Rebellion so to just start off with a little bit of context on where we're coming from um we want to have a a view of where we think the world is and where it's going now some of you you may recognize this chart it's from the paper situational awareness and it just plots the rate of innovation of AI as measured in orders of magnitude improvements and the author thinks that we'll get to AGI

6. 2:21

   in four to five years and then Super intelligence shortly after now whether you agree with that timeline or not it's kind of hard to argue the pce of innovation that's happening in AI right now I think we see this on our timelines every day with lots of new developments and so um kind of with that all I think this quote by the Google CEO really sums it up it's that the risk of

7. 2:58

   underinvestigated well underway so the question that we all have now is which path does this take us on are we headed towards this utopian path on the heels of a few big tech companies or do they lead us down this darker path now the way that we see it is that the utopian path is a world where we have many models small and large Niche and generalized decentralized and centralized they're all coexisting and the point is that gives us Choice the darker path and I don't think

8. 3:38

   I'm offending anyone here in this room is a world where you have a couple of big tech companies that have these centralized God models and you think about that you think about the amount of influence that they would be able to exert if this happens the ability to control the content and the creative output from these models that's a pretty scary proposition but if you're a builder it's more than just that if you're working in

9. 4:07

   in AI um and that's your your world then you have to deal with a bunch of risks so for one you've got rug risk at any given time the owners of these models they could change they could limit they could restrict your access it's really hard to build a sustainable business model with that underlying risk factor you have cost performance trade-offs a lot of use cases for AI may not make

10. 4:32

    sense financially to be run on these big expensive models and so you're just limiting the potential here you've also got vertical integration what's to stop the big companies from eating up more of the stack and you know ultimately they're going to pursue a strategy of Maximum value creation and we've seen this happen over time with other examples and then of course you have privacy I don't think I need to expand

11. 4:56

    on that too much here so we see world with all of these many models is making so much more sense than relying on a couple of handful of centralized God models then we have to ask ourselves this question of where does crypto fit into all this crypto AI has been criticized a little bit for um in some cases forcing a token where maybe it doesn't need to be and that's fair so let's take a step back and think about how crypto fits in from a first principles perspective and you start

12. 5:27

    with well what do we want Builders and as users we want as I said earlier we want choice we want to be able to pick and choose what services we want to use to meet whatever needs that we have and so to deliver on that choice well what do we need we need infrastructure data routing orchestration payments privacy all of these things this is where crypto really thrives it's bringing these

13. 5:53

    Services together in a decentralized way with crypto incentive mechanisms so crypto is a perfect fit for AI when it comes to the decentralized aspects of this many models world that we want to see and that leads us to this decentralized AI stack now I know there's some logos on this page this isn't meant to be a comprehensive view of the crypto AI space we think the one that topology Ventures puts out is really good so check that one out this

14. 6:25

    is just meant to be illustrative and if we break it down very simply into four layers at the bottom you have infrastructure this is where we've seen a lot of the early projects bloom in crypto AI it's where a lot of the capital has flown and that makes a lot of sense if you just look at Nvidia today the world's most valuable company capturing a lot of the value um that's kind of where it all starts in an early sector Innovation is just easier to underwrite As you move up

15. 6:51

    the stack into Data that's where we things see things getting pretty exciting you can think of data as like an iceberg where the part of the ice that's above water that's the publicly available data and the models today have consumed much of what exists out there the much bigger portion that's underwater that's private data that's data that can be unlocked through crypto incentive mechanisms and in fact three of our eight cohort companies are working on various Data Solutions that we are pretty excited about and then

16. 7:22

    moving up to stack at the model layer you have some really interesting Innovations happening here you have um you have you have things like enshrining models directly into blockchains you have model optimizations this is what is I think really more groundbreaking and where we're going to see a lot of new unlocks that open up the app layer and at the app layer that's where over time

17. 7:47

    we see the most value uring we see the most exciting things that are going to be happening and what makes the app layer for crypto AI really interesting is that you have such a large add ible Market because you have all the people all of us out here as users of these applications and then you have all of the AI agents as potential users and so all of this put together is what we think is the big unlock and what we have to build and so I'll just kind of wrap

18. 8:18

    things up with a few closing thoughts the first is that it's really hard to overstate how impactful AI will be we also can see though that a world where AI dominated by a few centralized models um either run by companies or or worse by the state that's a dystopian future that we really don't want to see um what's so great and what's so important about crypto AI is that it has the opportunity to give us an alternative and actually deliver

19. 8:50

    genuinely better Solutions the good news is here we're still very early the challenge that we all have is that there is just so much to build so everything I've shared here today it's available in much more detail if you read our paper I think there's some hard copies available around here and then it's on our pin tweet at the deli Labs X profile if you're an investor or a builder and you want to learn more about what we do uh please reach out or come find me thank you very much