The Convo: Unpacking the Solana Outage, Crypto & AI, and Blast Vision

Transcript

the highest likelihood of any of it is that we still haven't seen the winner that like we don't know the name of the actual you know underlying Financial substrate that will guide the world in the future that that still has not been created I thought it was Black Rock chain what's up everyone before we jump into the episode little plug for digital assets Summit coming up in London March 18th to 20th tickets are pacing so far ahead of schedule that we had to decrease the discount codes so instead of Empire 20 it is now Empire 10 head over to the website digital asset Summit D London March 18th to 20th use code Empire 10 and get 10% off your ticket

see you in London hey everyone just a reminder that nothing said on this episode is a recommendation to buy or sell any Securities or tokens this podcast is for informational purposes only and any views expressed by anyone on the show are solely our own personal opinions and not Financial advice myself Santiago or Ryan May hold tokens personally or their Associated funds may have exposure to tokens or equities mentioned hey everyone welcome back to the convo with Ryan Zer Santiago and myself excited to talk today we're going to cover salana outage we're going to talk about blast we're going to talk crypto Ai and we're going to talk defining themes for 2024 so we'll get spicy to start off Ryan let's talk about sana's latest

outage so it it's almost like you know in The Simpsons opening Montage where there's that thing of of like how many days without an accident and and Solana almost passed the one year without an accident but didn't get there and now it goes back to zero days um and the big question is does it matter right does does it matter to the Solana Community doesn't matter to the broader Community but me looking at it I would I would say for the institutional like high transaction volume clients that Solana is trying to attract it matters a lot um it clearly didn't matter to their core

Community this week because the price stayed relatively flat um throughout this and you know leadership seemed to brush it off uh um as as we saw Anatoli put out a tweet saying you know he had gone to sleep before it happened woke up after it was solved and you know he seemed unbothered by by the whole thing but you know I'm wondering from you guys what is your take you know does this matter as as investors as community members and what are the implications of course it matters of course you know it's like that Meme of like everyone's like cracking their neck and you know let's go you know the like salon's had an incredible amount of attention and run up and you know this

is a moment to really uh dance on the grave for so to speak uh there is a big distinction so like if we look historically this was a liveness failure my understanding is not to bore people to death was it it's it was a known issue they were working on a patch it was related to like the validator kind of U code base and um so in in this instance yeah it was close to the year um I believe this what this outage lasted 4 hours 46 minutes um um and relative to Prior outages the one before that was I

believe you know February 25th of of 2023 that outage lasted 18 hours um so you know a lot more um and historically look I I am not the most technically competent to speak to this so I will um go by what I've learned from the salana dev community and try to process everything that went on over the last couple of days and look I don't want to brush it off or minimize it I think I certainly agree with you like liveness failures are no fun it is a

challenge uh it is well known that you know salana as an ecosystem project is much younger as they were trying to push out a lot of the code from the salana labs Foundation I believe it's Labs who did most of the coding uh what you see now you know it's even Anatolia would concede they had to make certain compromises one they're really bootstrapped and two they just didn't have enough resources right and they're sprinting now that's like I agree with you that's not a sufficient um I don't think it's meant to be an excuse it's more meant to be be just an acknowledgement of the state of salana it's a much younger project um

they are they acknowledge these issues and they're working on them right I mean you're going to have fired dancer uh later this year which for anyone that's not familiar fired dancer is another client so that would add more redundancy to the chain and that's huge right ethereum I believe has Ryan you know this better than I three or four clients uh four yeah four exactly so um having more client diversity is essential uh fire dancer is developed by jump um there is an excellent blog post by the helas team that goes into a lot of detail it's not just that you're

going to have a second client but a lot of the upgrades uh both in terms of performance and reliability uh are meaningful and so um yeah it's just part of the you know there's a reason why salana trades at a fraction of what ethereum trades at sure but I mean salana trades at 56 you know versus ethereum's 220 and um you know so so so so let's call let's call that a quarter and the the big question that that comes to mind is you know the fact that ethereum didn't have you know major liveness issues four

years into live chain um this is core value proposition to Solana like imagine what it costs a visa to have four hours and 46 minutes of of outage time and how they must be taking a pause with with all of this um and you know we're relatively large deployers in the salon defi ecosystem where we see a lot of very raw um you know products and and codebase and it does beg the question clearly was undervalued at some point over the course of the last year but has it gone over its skis versus where it's where it is technically right

what do you mean by raw uh so for example like def5 pools don't pull directly from rpcs and you have to run command line to deploy a lot of capital and you know we're talking about a second client from a relatively centralized Source in in literally year four so like there's a lot of you know there's a lot of extra software to be built out from here um for it to to be in a comp category with ethereum right like for it to be a quarter of ethereum's value one would think that you know the D5 ecosystem should be relatively intuitive like D5 and Solana

for retail right now really challenging right like if you don't have good engineering resources best of luck you know to all players in the game uh and and that's the question that I I look at is is has it it again clearly was undervalued but has it run too much versus where it is in its moment is it just ahead of its moment in valuation at this point at one quarter of ethereum or roughly you know yeah I think so the way I'm thinking about it is I expected salana to have more Growing Pains but I think the less time of this outage is obviously better right that's like an improvement I think the thing that would

really screw me up is if fire dancer was released and was not performing that's what I think would be obviously very bad like so I think the market is pricing in a very strong fire dancer release significant upgrades you know people are okay with lack of client diversity at this point and are overweighting the performance from fir dancer which is great but I think if that comes out and there are issues there that's going to be way more disastrous than the labs client having issues does that make sense or so success in fired dancer is priced in and that is compensatory against the outages uh look I don't know exact first

of all I don't think you ever know what's priced in uh just to get the Bitcoin happenings um we're 70 days out is it priced in historically it's never been priced in so like something as predictable as that is not priced in what makes you think that fire dancer is priced in I don't think I I guess if you were to survey most people in the market don't have a true appreciation of what fire dancer is uh or Ana which is a third um Dev shop building um you know another client I I don't think um I think um I mean Ryan your your point around how you know is salana current

valuation uh relative to ethereum Accurate uh I would argue it's not and it's undervalued uh I don't want to minimize this outage I mean I think I was kind of expecting it to happen sooner given the incredible amount of activity that you've seen in a very short order like there's just been let's put it in perspective over the last couple of months the number of activity across defi um and other applications like drip tensor has been incredible um I would have expected like I would have expected if anything the network to to go down during the Jupiter

claim which was uh you know uh put a lot of I would have thought would have put a ton of strain on the network and in fact the network performed really smoothly uh priority fees kicking in is something that is pretty encouraging from my standpoint I think I think the issues with salono are well known documented and go listen to a lot of PODS that anatol's been on recently including mine and fire uh uh light speed and well includ including this one in a couple weeks yeah yeah I think they we'll get a chance ask him directly yeah but I would have thought that the network if anything would have gone down

sooner particularly during either the goo or most recently the Jupiter airdrop claim but when I say what gives me the most amount of conviction yeah look I hear you Ryan salana is you know to say that look it's it's a baby baby chain is not really right I me it's 50 billion worth of a baby chain that doesn't constitute a baby chain and it's been a couple years out but the reality of things is I almost think that in reality for me at least salana is a very different chain um it's just gotten the quality of projects I hear you out still not

perfect um I think the quality of products on the salana ecosystem are superior to many of what you see not not not comparing just to ethereum just you know are products that I would send um a uninitiated person in crypto to try out first particularly tenser and Jupiter but I'm not here to debate the retail side I'm hear more to debate more with you guys the the fundamental question is is it acceptable I think it's unrealistic to assume that any chain is not going to have issues right there's always going to be bugs you've had bugs of Bitcoin pretty serious when you've had bugs in ethereum um I mean this is part of the game right um so

Santiago so one one thing that's different versus the last outage is there's obviously way more usage on salana I think I mean just I think so now versus in the past right so the Ripple effects of an outage could be like Quantified it's larger like I don't think we even have like an instant report out yet but do you have any idea of how this affected salana defi trading you know the whole nine yards there like I don't have a postmortem I think the impact um I was looking at a couple of uh people reached out the Jupiter team say hey well what's going to happen right uh some of these positions like is there bad debt I don't think it was yeah I haven't seen a postmortem is a short

answer to that and you're right the biggest risk during these um outages is particularly dii right you have like Oracle like these PRI this C like centralized exchanges keep processing these trades and so there's a price there's an oracle there and so once the network is back online you might be uh it might be instantly liquidated so um yeah I haven't seen a postmortem but I suspect it's not as big because there wasn't that much of movement as r alluded to earlier in prices different story if like salana would have been down 10% in Prior just this just happened or other defi assets

would have been down like you know quite a lot but I don't think this was the case in this instance maybe because most of the world was sleeping like I guess like you know like in the west coast like just went to bed and wake up and nothing if nothing can happen def5 remains strong I mean we when when we were talking in November it was around that 3 00 million uh range and and like it had come down really precipitously but now we're talking about you know 1.5 billion I actually think it's going to be closer to two or three billion than maybe a month because of all of the interesting stuff going in that ecosystem and to your point there are

some you know there is some Brest to work out but there's a lot of excitement and like what is the what are s the one or one to three things that cause you to conclude that not only is it not over at skis in spite of the challenges but it's undervalued versus its peers I mean to take it a different way I think the thing that I'm struggling with with eth is just the obvious question of long-term value flows in a fractured environment like you have eth a settlement you have igen and Celestia as da layers and a couple others coming online then you have the l2s like the modular stack is awesome and seeing like the Celestia tweets from Nick white on

all the money that rollups would save by using them is phenomenal but that's all money that's not going to uh e L1 so I am struggling a bit to understand where the best place to play eth is long term um and you know the da layers have their own concerns on on value capture themselves because all the money they save is is clearly not money they they are making um so sort of one issue that I'm struggling with on e um which is kind of ironic because on Salah you get composability by default um and fpes and now we have this uncomposed uh value capture issue which I'm struggling with but yeah I definitely did not answer your question ran I took it a different

way but maybe yeah you answered it via the counterfactual and that you're saying you know maybe relatively speaking eth is just over at skis because of of these really significant architectural challenges es so that that does answer it I that is a way to the to to why one would be outlier bullish salana right now yeah and I mean the I mean the amount of eth Locked Up is also phenomenal right like the amount in Blast the amount in what is it 25% of all eth is staked across I layer l2s and a bunch of other places yeah and we're I mean we're investors in igen L I'm an angel like I I love seeing the the traction there I I think that I mean there's a lot of

different reaking options you can do ethery you can do Renzo puffer there there's a lot of options and it makes sense right I layer has caps you can go through these protocols earn extra to earn their points or tokens and then also get some prata access so I think the the adoption that IG layer has on the supply side is phenomenal and definitely blew my expectations out of the water the questions I have on that side are obviously rehypothecation risk so you SL e somewhere how does that affect the rest of the system and then to the demand side like what is actually being built Beyond Igan da I think is um my two largest things I'm tracking right now um but yeah it kind of just goes

back to the E side by the way Tommy uh check out the Q&A that uh shiram and the Ian layer team posted recently around this rehypothecation and a bunch of other questions it was quite useful I think my big oh look I'm not to get to on a tangent here but uh yeah just go read it because I think it the the risk is not so much in rehypothecation at the IG layer layer at least certainly on like who you like the resters but uh it's really around like this idea that the chain and the scalability road map of ethereum is being I guess delegated to a whole set of other parties and particularly my

biggest issue with when I first heard the pitch from Calvin and siram I guess I'm maybe butchering his name is that they're going to have this like Security Council to determine the security parameters initially we're going to select this few people to set these guard rails and when I hear that it makes it gives me pause I'm sure there there are competent people to make this determination but a lot of it is just it's difficult and I I absolutely would want a centralized Council making those decisions to start I definitely do not want a community of addons like deciding risk engine I'm not I'm not suggesting that an unon would do

it I'm just saying the point around I guess like I much rather see trust minimized systems and the original conception of ethereum was just trust minimized like you don't need to trust anyone right and now you you're seeing certainly like I think the scalability roam up of ethereum is something that perhaps we should debate at some point I I don't think it's impossible I think uh composability at one point if your guys were Devon and Kyoto was like the word right you're going to break composability I think break in composability is not the right mindset I think the internet when I go talk to core

Engineers way back in the day I think that a lot of what you think about composability not not everything needs to have atomicity in it's a spectrum and I think you can fix that with better connectivity between chains I think it's solvable um I don't want to come across because most recently I've been of course very vocal about salana most of my portfolio is still ethereum Centric folks like I hope it's I can't unwind say it again I I will say it again most of my historical and current portfolio because I invest in early stage projects are skewed heavily heavily on the ethereum camp but well but that doesn't mean that

I I'm not like I think I I've just gotten frustrated with the lack of visibility into how ethereum really becomes a chain that can attract like mainstream adoption and I hear you on the Celestia pitch I hear you on like modularity and certainly it gives us more breathing room and I think maybe that's that's the maybe that is the bull case for ethereum which is the human capital the developer side which you know it's like high school you at the last minute figure [ __ ] out and hand in the homework and you f and Humanity has a way of fixing things in the last

minute and I'm never going to short that side because ethereum dominates but I just see a very clear path for salana to when I see how it has been able to handle a surge in activity I think it would be unfair to look at this network outage which was known issue which is recently this this upgrade um called the like the Berkeley packet filter it it was like a recent Improvement and they knew the issue they were fixing it and it just sort of the network went down but I think it would be a bit unfair to look at this and say oh salana is just a centralized VC coin

Sam from prison is probably you know or the I seeing memes of like the Nintendo 64 blowing in the cassette uh I I think that misses the point and there's still a lot of people that are off sides a lot of people that haven't been paying enough attention to salana I think it is a very still a very non-consensus trade and crypto is still very much not consensus so doubly non consensus interesting uh well that that's great feedback um with that let's let's switch gears and and dive into the crossover of of crypto and AI one one last thing Ryan because last pod I think it's funny because we were talking about the salop

phone and you know you really dismissed that and now you're gonna have egg on my face on the salon three that not not not put you on the spot we've it's been I don't think like like it sounds like you're putting him on the spot listen I the question was more directed around Ryan you're someone that has paid a lot of attention you all are playing in Salon it sounds like um we everyone here has invested in projects that just didn't get off the ground or haven't gone off the ground have you change your mind on something in particular about salana as of late given everything that's been going on both good and bad and neutral but have you Revisited your

thesis on salana um absolutely I mean I think you you you come back to it consistently and from the time of the scen Saft investment up to now it has been this sort of like give and take of okay this feels over at skis and then okay you know actually they're doing some really interesting things this is a genuine community that that that's one of the things that was very obvious of 2023 that you you had to respect the fact that it's a really legitimate community of like you know Network devs and people taking a very different approach to

the you know the almost religiosity of Bitcoin or or or or or derivative of that with ethereum where kind of decentralization Trumps all and and and you must sacrifice so much more for that because that that trumps all um their their different approach is is really compelling and I think they they brought new energy into the space and there's some really interesting things the the phone um I I I I certainly respect the fact that you can pay for the phone with aird drops I think that's a great idea I as a as a venture capitalist would would question the um the wisdom of going and fighting the

holy war against an apple and a Google in in phones on your own while trying to to to run a chain and so you know there are red flags and I think that's fair and I deploy Capital mostly in in in yielding but also but also speculatively in in the ecosystem right now but do so very tepidly and and do so questioning myself at every uh at every corner because I don't you guys must go through the same thing but the more time that we spend in this game the more somewhat unsure of of myself I am and the more I'm questioning my own my

own hypothesis and just like trying to to stick to a to a framework and and so I don't propos to have the the solution on on Solana right now um I would I POS in this exact moment I take a little bit of the other side and I say man a quarter to a fifth of ethereum In This Moment feels expensive because of my perspective from yielding and the quality of the yielding ecosystem in in evm versus quality of the yielding system in this moment in Solana but at the same time they evolve much faster than ethereum because they don't have all this architectural tech technical debt what's up everyone quick

break from the show to talk about digital assets Summit this is going to be block works' eighth Das we've been hosting this since 2019 it's our longest running conference series coming up March 18th to 20th in Sunny London Town Spring in London you don't want to miss it it's a crazy lineup definitely our craziest D yet just added Michael sonen Shine from grayscale the head of digital asset research at Franklin Templeton we've got drw we've got KKR we've got Marshall was tme group we have Bloomberg we have Visa we have wintermute we have Vanek we have FIA we've got the framework folks joining us 1kx coin shares JP Morgan Standard Chartered qcp Morgan Creek zanti is coming of course Ledger CEO and chairman Pascal will be there it is going to be a phenomenal

event you don't want to miss it tickets are pacing so far ahe of schedule that unfortunately we can't offer 20% off anymore there's a new code it's empire 10 Empire 10 is going to get you 10% off see in London what uh maybe perhaps if you guys could entertain a question because I would be curious to understand what are you most excited about from the ethereum scalability road map is it parallelization through something like monad is it Z VMS is it modularity like Celestia type of stuff really compressing down the costs all the above a lot has happened I guess and is increasingly

happening it's hard to keep up but I'm curious um if you guys are what is it that you're most excited about from ethereum side um so the the thing that I'm most excited about is also the thing that I think is the biggest issue so salana narrative very simple um you know fastest Global saate machine speed Al light yada yada um I'll let someone on salana correct me on that on Twitter after this goes live um the ethereum side I think the thing I'm most excited about is also the problem in that there are so many teams building I don't think any of us and I've tried to find a guest

to do this could conceptualize everything that's going on in the short amount of time or reasoning that that makes sense like I don't think you can ever put the dozens of L2 and infr teams now building around ethereum in a concise way that makes the narrative super sticky and understood so it's positive in that there's all these teams building it's negative in that the narrative is not a catchy oneliner um but I don't know any other way that it's going to to play out other than the things built on top get mega popular that drives value back to eth as a settlement layer and that's sort of the

narrative um but well you that's n for the value cruel but not so much on like how do you get billions of users yeah well it is though because like the the Hydra of apps built on top of eth on l2s will directly Target users it doesn't have to be on the L1 yeah yeah I'm I'm pushing the fees yeah I oh sorry I I think I think we've we have probably we've clearly not seen the emergent one that will capture like the millions of retail users the thing that I'm excited about it it with respect to to ethereum scalability is capital efficiency and capital sink

holes so Capital efficiency is you just what you guys were talking about reh hypothecation um you know new ways to to to like yield and use that Capital such that it's you know compounding andring back to eth as a base layer and then eth as a base layer I am resigned to the fact that I mean if you own less than $1,000 and an e address get it out in the next X number of months because eventually it's going to be dust right like like it will be it will have a certain cost um and and that's because of the second thing the capital sink holes where there are all these ways to

to to use capital and and you know efficiently reh hypothecate Capital such that you know it sucks up a lot of eth and so there isn't that much circulating eth and a as a result you get um you know you get appreciation and uh I think that generally that will drive interest because you know because of the financial incentivization people are making money and it's scalable because of that you one of my core thesis right now is that uh the financial use case for the foreseeable future is the killer app of crypto and that's okay we don't actually

necessarily need Social and need a whole bunch of other things all we need to do is be able to give decentralized loans and hopefully do decentralized Capital formation uh AKA Ico 2.0 on the horizon and that's fine and if we can just do that that will attract millions of users because if you give a man an incentive and a lever he can move the world he'll get he'll figure out Wich L2 and he'll figure out all the mess of at all because he's financially incentivized to do so because this is where the capital is Flowing because of capital efficiency and capital uh s HS Sor right that was a bit longwind that's that's no it's not longwinded at all it's great both of you

guys have like touched on like I agree the financial gimmicks of reaking reh hypothecating like yeah like go [ __ ] Farm yams like it was fun I get it but none of you guys have touched on fees and when I think of to your point Ryan what about the user that has less than a thousand bucks to speculate impossible impossible L1 main chain and L2 me who knows you still got a bridge to got a bridge to it and it is challenge it is you know I would encourage a listeners you guys for me one of the more eye openening

experiences in this journey was I've constantly like re underwriting my thesis talk to some a project like drip like compressed nfts the ability to do things at like meaningfully orders of magnitude cheaper than even an l2s you know You' be like ah yeah so what like l2s are pretty cheap and I get it today but you know look so much in investing's understanding the present I think l2s today can support more way more users than what they have but there is I think a a limitation to that and I don't see that big of a problem with salana there's other issues

for sure they're going to have to implement an EIP 1559 fees you know not perfect but they're like on the margin is what I'm trying to convey I see more fundamental issues more leap of faith um on the ethereum scalability road map not the that it won't be figured out I just think it carries actually more risk than perhaps people are willing to acknowledge today whereas inertia yeah and you're betting on fantastic developers that time and time again have risen to the occasion and I think quality of devs is Superior in ethereum or you know it just attracts a lot of attention because the

liquidity has been there historically I'm just not so sure that that will continue to be true going forward if you're a builder I ask every portfolio company that is deciding on which chain to build I'm yeah I understand like I'm an investor in many l2s that we're going to talk about seed investor in scroll blast arbitrum uh Mont like a lot why because I just want to see these things work but today I just genuinely asked the question like what can you get out of all of these that you will not be like that you're not getting like why

not just building salana with fire dancer performance wise and maybe Ryan yeah it's uh hey well the chain goes out and there's Network outages and that's fair criticism and we can leave it at that you know we shall see well Santi to your to your point I mean one thing cus at at Deli or head of research I always brought up was that like the design space of a chain with such low fees has never truly been like exercised to its full potential right like there hasn't been years and years of which to create apps that could take advantage of that low fee Market um and I think that's like a a really good bull case for for

salana as well and up Pi up what you guys are talking about yeah and this there will be many on my view and Ryan I think we can all agree and I know we're supposed to disagree but you know it's a it's a multi-chain world in my mind not all applications require composability not applications require the same level of security trust assumptions but you know the highest likelihood of any of it is that we still haven't seen the winner that like we don't know the name of the actual you know underlying Financial substrate that will guide the world in the future that that still has not been created I thought it was Black Rock

chain oh black Rock's busy buying up all the single family homes and and pricing single mothers out of their houses well that's what the Instagram memes are doing Blackstone or Black Rock you're talking about sorry yeah talking about Blackstone real I'm talking about black sorry come on man get out it it was a late one I get I get one mess up yeah we we can see we can see that Vagas that Vagas landscape in the background why so quiet so actually quiet let's let's let's hear from you I want to I want to hear the the Tommy Shany um viewpoint on the crossover

crypto and AI uh and where you know where you're excited uh about these these two important Technologies which there has been this ridiculous false dichotomy created of like ah people in crypto should leave and like go to Ai and it's like like actually the two are incredibly enabling for each other and so I want to hear more of how you think through that all right I'll try I'll try and keep it quick now I'm on the spot um so I guess the my interest got peaked with mid Journey right I think that's you know what over 18 months ago I gave that a shot and was just blown away by what you could make you know type in a prompt get a picture super cool right

that significantly accelerated with chashi PT launching as you know who else sort of went into like a slight existential crisis like you know are we in a simulation does this all work yada yada um ended up realizing that large language models like chat gbt are are obviously not Arnold schagger Terminators right they're really good at finishing our sentences right they're don't have any creativity thought value system any anything like that uh it's almost like uh confusing the the user on how good they are um so they're trained on the the the world's internet goes into a system F Tunes AKA changes a billion zero and ones to certain numbers

um and when you write a prompt it it goes through that system so that was started my journey with llms and Chachi BT um but then I further got solidified that crypto and AI is perfect uh given the open AI drama um the concept that you have a board of what looked like high school teachers I think one was controlling the most powerful foundational model on Earth um is just extremely laughable um and the idea that that could change hands uh back to Sam mman and crew over a weekend uh is ridiculous um so imagine not just chat GPT that you have now that you're entering chats into but the chat GPT App Store there's three million shat gbts

built on that foundational model if you slightly change that model um you know let's change it a little bit so that we hate Santiago right or let's change it a little bit so we like salana that reverberates through every application built on top of chat GPT and literally changes the the flow of society um so at a foundational level I think that the most powerful open most powerful models should be open source should be transparent um and should live in a crypto likee World um they should not be centralized like open Ai and we should be able to see them um so that's sort of my core thesis for why Ai and crypto go together the other kind of tidbits are that I think in the

future on the the long Arc to AGI AGI will be using crypto um they're never going to go into bank branch they're never going to trust a Charles Schwab account they're going to use permissionless money to build uh their own purposes right another angle is that um all the latent Hardware we have around the world you know Macs ps5s Etc should be used to train these models there is a cost that people will pay to do that the Jensen folks talk about it as just increasing the amount of energy in the world right um having all these extra gpus to train models or to do inference and host those models against uh to run prompts I think is is massive that's like IET aosh Jensen you know they're all doing that um the other side

of it you know one level up the stack is just uh smarter apps so you know you swap liquidity provision is really hard um let's create an offchain AI model that is large and can handle that effectively and when you want to rebalance your position you know now you could ping that model and move things around so applications get really smart um another angle is the user side right llms make it really easy to use crypto so instead of uh you know Bridge my eth to state eth to put it into a pool to lever up yada yada just type your intent in a large language model see the simulated result and it's sent um so yeah there there's a lot of ways this

can go um I have pros and cons for bit tenser and other projects but I'll uh I'll chat I'll stop there for for right now yeah I've um I share a lot of the you so I a lot of my education and AI come from specific folks like Daniel modulus some an investor in and they're doing U uh verification uh you know if you being able to prove that the model is actually doing and performing in the way that you told it or or or telling it to and that becomes quite valuable um you know a model could go Rogue so to speak um and having the guarantee that it's

not is pretty valuable um and I am you know I've looked at to your point Tommy I've looked at the decentralized kind of compute like a deep pin angle like you know for training purposes um like the Aaj and bit tensor of the world um I'm curious to get your thoughts around like two things one are you more excited about inference or training um of these models um and second you know talking going back to the initial

discussion I think it's unrealistic to deploy an llm on ethereum just just can't handle it just he had to bring it back to Salon ride it was five minutes too long I'm not I'm not saying it has to be I'm not saying it has to be salana I'm just saying C had a hilarious tweet in like I think it was sort of mid 2017 at where he said people who are talking about AI on top of blockchain why do you want to live your life in this way you know like um yeah and in the context of

those BL chains like it was a ridiculous conversation right to talk learning using you know Gollum or something like that on top of ethereum mayet it just like like to his point like why why do you even want to live your life Life In This Way um that's fair you know one of the there are obviously some like lwh hanging fruit that that can be compelling like are you guys looking at um at any startups uh focusing on uh attribution of uh of data like data inputs into into models so for example you know it it would seem like a lot logical use of

crypto Technologies is getting that decentralized attribution so that value can be can be shared across all of the necessary tokens or or data points that compose to make an output so if I if I go and say you know what were the top 10 um news headlines of the day that you can use it you know a a cryptographic proof that shows that they pulled one headline from The New York Times and one from The Economist and one from here and there and everywhere and then those entities can be paid in a secure trustless and even private fashion for the the data that they supply to the the model that creates the output and that

that seems to solve one of the big problems that very clearly AI has this tsunami problem on on the horizon of the fact that they're using all of this intellectual property and not paying for it and they got to find a way to pay for it and it would seem that crypto is the most efficient path to getting these disperate parties to to align on a payment mechanism um is that something that you guys are looking at well didn't uh what's his name he there was a decentralized hedge fund model right you would upload your model there is you would upload your model and if your model outperformed like you get oh yeah yeah numer um yeah is it happen to be

num happen to be a investor however so so with a you know day one you know 2015 investment the interesting thing about the numerai story is that and this is my observation from the outside because they have to to homomorphically encrypt the data before they share it with the tournament participants there is an argument that I have made observationally that you may have greater loss in the quality of the data and the inference that you can derive from it because it's hidden behind homomorphic encryption than you gain by having the wisdom of the crowd and thus the it

becomes really difficult because you have that veil of the homomorphic encryption to outperform say you know like The Medallion fund or or or whoever else on on Wall Street because they don't have to worry about that part of the privacy of the underlying data um I do say that having with a tip of the hat that Richard kbe and his team have performed exceedingly well and very consistently upmarket down Market numerai continues to do really well um and but like it it doesn't have those like um you know those massive outlier

gain years like a citadel will have and I would argue that that is probably the missing link is that you have to hide at all behind homomorphic wait a minute but uh the lizard brain of mine my basic understanding of homomorphic encryption is you take any set of data and if you don't trust you don't so if you're uh you know an analyst you don't want other people to look at your model because hey it's Alpha and but you're going to submit it no one's going to be able to see the model just a result right but the model that you can train an llm with homomorphic encrypted data and the LM itself becomes Superior so say that like 10 analysts upload their models it's

homomorphically encrypted you can still the whole point of that is to train a model with all these data sets and only like and you care about like the result the model becomes smarter adapts takes all these inputs but but the whole notion is that the people behind it can't look at each other's models and copy that but the LM itself becomes smarter as a result because it's ingesting more data the quality of data should be higher so how I'm lost in your the point you made which is the wisdom of the crowds like should beor let's say let's take a really trivial case right let's say we're

trading a bunch of stocks and these stocks have a number of data sets that's like weather and you know p and um uh you know some like economic data and things like that and then you you hash that underlying data the the argument there is that in hashing that underlying data so that look behind homomorphic encryption so that it can be shared with these groups and then they can evaluate the trends between those data points in order to create models or the inference between you know PE and weather and how those

those interact and then the model will create a more intelligent output because of that um the the the sort of loss of data quality by going and and and holding it behind a behind a a hash of a homomorphic enpon prevents the model from getting that really outlier Insight right because it's you know it like it it's just an inference and um and so even if it's like let's say it's like 5% loss of data right you're talking about something that is harder than the Olympics right let's say it's a 1% loss of data you're done against a

citadel over right like that that one % is their their lunch money right and so you know like can the question is can you make that 1% up or that 5% or whatever that small loss of data quality up by using the wisdom of the crowd across a number of models um and I think that remains to be seen now again numerai does pretty well um it's not Citadel though it's not millenni um but it does pretty well uh and I would argue that the reason the thing that holds it back from being that like dominant Global Capital pool is that like small bit of of of data

quality loss um through the encryption process to your point is how do you choose what botel is being used for what inference right like there's going to be billions of models right exactly like they don't know that it's weather data they have to infer that it's weather data right exactly like something with bit tensor is like a good example like like let's they have you know 30 subnets every subnet does something different model creators upload their models to each Subnet in an effort to provide the best value right to the to the inference provider right so if you're doing image Generation all three of us upload our fine-tuned image

models mine's adime Ryan yours is hyper realistic Santiago has salana on top of every picture for his like you know how do we gauge which inference is the best one right and I think deciding which model is going to be used in each situation is going to be really hard um and I don't think bit tenser saw that or anyone else doesn't the market solve itself like the market solves this problem very efficiently right but how though if you if you have a million images image models created you have to run each inference against each of those models and get feedback from every customer to then rate them yeah it's it's very hard

and economic challenge right because not everyone's going to have the ability to do that at scale and well resourced players this is why technology is a centralizing force the only shot that we have right because this would dictate you play this out sufficently over number of iterations the well capitalized players might not have the bone Superior model initially but they just become the the facto standard because they were able to play this game farther out than a less capitalized player crypto comes along and I think this is from a more philosophical standpoint why crypto is essential in this transformation that

we're going to see in AI is because technology is an incredibly centralizing force and the only shot that we have is to to prevent AI to be incredibly even more centralizing than what we've seen today is by creating more sophisticated coordination mechanisms with better incentives that make it the superior alternative now what does that actually mean in practice I have theories but I don't think it's a bit tenser in AOS as they look today candidly or even I like it's it's G to take a lot of work because you can just go to you can use the the like Time and Time Ryan you

remember these [ __ ] [ __ ] back in 2017 like let's build a decentralized uber I'm like why that's the last thing I want to [ __ ] use who do I call Airbnb decentralized Airbnb who do I call like how do I resolve this we need to up our game like we need to up our game in crypto the stakes are getting higher the opportunities are higher blockchains are data Rich they will produce pristine tamperproof data that is incredibly valuable will be over time we haven't figured out the product side of the house the incentives side of the house the economic

cryptoeconomic side of the house we need more talent coming into these sectors to really be able to have a shot of competing against we two players like mid journey and and and and so many of the other players that allow you to like spin up or rent out quickly a GPU doesn't have to be decentralized by the way like it doesn't the the the more compelling are Tommy in your list of things that I definitely agree with his AGI like AI agents will use stable coins just because they don't have any other option and that's to your point Ryan maybe that's good enough like there still I would love to see stable coin demand go up 100x yeah I want to I want to share my

the Counterpoint to my own thesis right so my own thesis is that AI should live transparently we should all see the parameters y y right the Counterpoint to that is exactly what modulus Giza and our own Porco inference Labs is delivering which is if you're spending an obscene amount of money training a model for a specific use case you're paying for specific data specific training or some algorithm that's god tier you know you have this huge Sun Capital cost why would you ever give that away for free right you're going to keep that model off chain you're not going to want to reveal the parameters because you have this Edge you spent all

this money why would you give it away for free and you're going to let crypto applications and users inference that model without revealing any of the weights or parameters while it lives off chain because it also is too large to fit on chain anyway right and someone like inference Labs does that right like you have the model you have a hash and an identifier and you use that model if the model parameters change the hash changes you know something's changed you're not going to use it but uh it still lives off chain so that is like the Counterpoint to my own crypto AI thesis a bit but curious what you guys think about it that well that dovetails with a an observation that I make about AI broadly the the sort of generalist mechanic venture capitalist

in me is very cautious with respect to AI as a as a BC category um because it may be the first emerging industry where the incumbents like acrew the most value right you know um Microsoft didn't win search Google didn't win social uh Facebook didn't win crypto you know you can go down down the list usually the upstarts are the dominant um players in in a novel industry but in this case there may just be be so much um you know compute and data concentrated in a few centralized forces

that this time around that story doesn't play out the same way and that's why most of my AI play has been Magnificent Seven thus far and not in startups same here even though I've invested in share number of portfolio companies but I think um well and Tommy you you've hit it out of the park with mid Journey um congrats on that no thanks it wasn't I mean it was a it was through a fund that we we invested in so shout out to to them uh but um where where I am um I'm looking at a company now that uh I've actually this is one of the best Founders I've talked to recently we were supposed to have a 40 minute call we ended up having like much longer discussion back to my original point of

where I think AI like crypto is particularly well suited for training and reinforcement learning uh I think that's where you could probably uh create a more um Superior system relative to web to and the whole notion is you can train these like models with human feedback you can optimize um that and the rate of improvement is faster um you can the question is can you do it efficiently and I think crypto so far has been really interesting experiments to create systems that um really improve

coordination and incentives um and so I think that's an area where probably is where you have a better shot at winning over an incumbent with uh again reinforcement learning with human feedback and you know it's no different than perhaps uh you think about like PL earn for example which most people discredit now these days the question is is it sustainable you're doing out these incentives via token reward mechanism is it sustainable and that's an issue that gaming at least the initial games if you want to call it that have struggled with at some

point the game economy collapses um and in this case is actually I I think it collapses because you the network effects are not properly understood the network effects I think of machine learning like uh this are you know the idea is Ryan of us three are contributing to a particular model and we're being incentivized to do so I have an incentive to provide feedback the machine becomes better Tommy you your turn comes you also have an incentive to um you know um give feedback and then when it's your turn Ryan the model is just so much better and when and when

my and once you get that flywheel going you could argue that you don't necessarily need that much Financial incentives to to keep that Loop of feedback going around and because it has very powerful stronger Network effects than your axi Infinity um and I think a web2 company is going to struggle to incentivize a vastly distributed and um harder to coordinate uh distributed and decentralized group of of people and for training to your point wisdom of the crowds fails why there's two parameters that you need for

wisdom of the crowds to be true otherwise it's a bunch of idiots that produce terrible results and that is sufficiently large n so the pool is large and diversity and I think with crypto you can do that very efficiently better than perhaps a web 2 company can the the one push back I don't totally understand is that AI is a centralizing course right you have all the best hardware all the best data you train the best model but the Counterpoint is if Ryan were to create a fine-tuned llm he takes

an open source model like llama trains it on all of his DMS it becomes Ryan right that model will always be Ryan better than fat gbt will be Ryan because Ryan's model is fine-tuned it's specific it's it's perfect right and we'll have we'll have it on the Pod someday right but the large models have to generalize about the entire world so sure the model gets smarter as you get more info but it's getting info from everyone and has to cater to the largest target audience which is everything all the time you know who creates a better model of Ryan than Ryan who has more data than Ryan than anyone else oh Google or Facebook not

Facebook you Ryan but I'm talking you know what I mean but that's it Ryan we're creating you we don't want to tell you that lens the the the second order implication of that is that Google captures a lot of the value of these individual models because they already contain the data Trove on you and then they also have the compute War chest to to go and do the work and it's like Ryan I I tweeted this out a couple weeks ago I and I think it's a contrarian view I think Google beats open AI Facebook I was loading up on Facebook so because most people are like crushing it because of the metaverse servers I'm like these guys are going to spin up some of the

best model out there they I think they have the most h100s Facebook Ryan I I definitely am in your camp like look at who has the most h100s it is the top the biggest tech companies right and then furthermore who has the best data they're not sharing it with anyone Amazon Facebook Google Twitter like uh Pharma yeah good luck trying to be spinning up a specialized LM for insurance Fara you don't have any of the [ __ ] data how are you going to build a model it's all in the data this is where the AI Hy is going to come crashing by the way there's so many there's so many Venture

companies that are going up this I've looked at a a a a sector that I like but I have been really cautious to invest against my own will is biotech everything looks like a Christmas tree everything looks shiny it's going to cure cancer and it just doesn't because the the the the um startups don't have good data and the big the the the big far big Pharma has data on all these clinical trials that you've never even heard of they're going to then go back to that shelf and train it and like run through and you know do protein unfolding yada y y and come up with the next vaccine mRNA or come up with the next you know cancer

treatment or whatever it's going to probably be big Pharma yeah whenever anybody says oh aren't there a lot of scams in crypto you just say oh you've clearly never spent time in bi Ryan what's the difference between a scam and a um not a scam in the Venture world yeah and and and a what's the difference between a scam and a preclinical trial um company in biotech nothing wait but you have Henry ProMed us isn't that supposed to be legit well anyways let's switch gears uh we I think we got blast wait wait wait before

we do that uh parting thoughts AI how bullish are you in a scale one to 10 Tommy Ryan and then I'll give my answer and and and when I say bullish I I back it up with dollars like percentage of your book sure yeah yeah uh so I'm very bullish I think crypto AI is the next the next narrative for sure and I think it's a long-term narrative um and we're we're very active here we're going super deep with Founders we're putting our money where our mouth is and we have been for a year or you know our first investment in AI was alter State machine and then recently it was test machine over 15 months ago um and to date we've done I onet inference Labs um we've done upshot uh T Fu

jepetto Mythos Ventures Angel and my peach so if you're building on the intersection of crypto and AI with would love to talk to you cool I would say um you know the allegory or maybe the lesson from the Advent of chat GPT it's it's interesting how the world was like oh my God this is going to change everything like overnight like jobs are like going to be wiped out on mass and then now a year and a half into it so to speak it kind of changed a lot less than I think most of us predicted but then at the same time over a long period of time it will do all of those

things and um and it's something that I'm to to Tommy's point I'm taking Long View on so we've made some Investments you know some that that that Tommy said um you know the first one being ASM obviously um and origin and a few and a few others you know actually going back to to numerai in in 2015 but um it's something that I am also more cautious and and sort of like less High conviction in early stage than I have been in the past because of this really significant challenge that I see with

compute M data that is owned by you know the Magnificent 7 and so we're I would say we're we're relatively tepid right now yeah I am probably somewhere in the middle like I I think uh on my non- crypto portfolio I've meaningfully allocated to names like none of this financial advice but you know like The Usual Suspects uh I think Facebook was a clear winner that a lot of people were overlooking they just I'm looking at who has real data because I just think that's the biggest moot um on the crypto side I think um I'm increasingly cautious

to have taken my time to form a thesis largely leaning on people that are way smarter um go listen to the podcast with Casey and Daniel uh and so to to really just understand like okay what are these use cases what is actually doable who's actually going to be willing to pay for this stuff and um so yeah I made like 10 or 15 Investments one of which is a fun Tommy and I share with um because I just think much like Venture I'm not an expert on AI but I think there are people that have a very deep Network within the AI community and a lot of people for instance I want to

be I want to be investing in the manager that left um open AI knows all the developers that are going to leave at some point and build their own companies it's not going to be me who captures that Alpha it's going to be someone that is very deep in that community and I'm happy to pay manager 220 for that cool that's that's a great strategy yeah I one one last note the the thing that I like about the fund that Santiago mentioned it's it's mythos.png proprietary business

Integrations um you know the AI agent for that's integrated with Microsoft Excel clearly has a moe versus the one that you have to download as a plugin so I really like his thinking and always like to share that and I would argue most VCS that are pouring billions of dollars into startups that are trying to get on the waiting list for h100s the nastic probably outperforms those VCS nine out of 10 times right because all the efficiency gains that traditional businesses are going to get through AI or over a 5 10 year period if you just play that out low cost low fee ride the NASDAQ for the next 10 years I think that's probably as simple as it sounds but I think that is probably a better bet than uh most VC Investments

right now that are just purely like funding nvidia's linning nvidia's coffers Theo I need this a deep level I need it see how like I feel like both of you guys are you know it's okay it's all right so we love you no no we just not I'm not questioning that we're not cyber bullying you it's been a while since someone actually I got a lot of uh I love well anyways BL um yeah very controversial eh you know mo I think it started with admittedly a launch that um didn't go probably as planned um and I think the team owned up

to it and I'm specifically saying it would have been far better to communicate the risk to communicate how it was architected to communicate what it was and what it wasn't at the time meaning you were depositing in a contract just a multi that is going to take in eth and deposit it to Lio or maker and capture yield until the protocol was built I don't think there were forthcoming with that I don't think I don't think they were explicit about that and I think um the community really crucified them for it and that's crypto you you make a wrong turn you're going to get called out on it and I think

we've gotten better candidly as a community to really call out [ __ ] and yeah I think they they acknowledge that um where I think a lot of people have been a bit blindsided if you will or just myopic is not revisiting thesis um I happen to be of course I'm an early investor I've known the team way before they even started their first project well the most people know them for which is uh blur which is the nft marketplace that you know is dominant in ethereum are we are we talking about that in past tense by the way is Dom is to this day like yeah just one market share like there's just a great

our Network newsletter coming out this week that talked marketplaces and the evolution of how tensor beat out magic Eden and salana and then uh blur beat out like um open SE just incredible and it starts with great product and so I think to me that was my thesis simplistically I think these guys know how to build a great product um and I was involved in the launch mechanics or whatever um but I do think uh a lot of portfolio companies have reached out out saying hey we want to be connected to the team how can we participate uh we want to deploy in the ecosystem of course they're looking at over a billion dollars of staked eth uh

now it's up to closer to two anyways um yeah it's it's between one and two between one and two yeah who would have thought with all that you know liquidity there' be interest from developers um yeah so I I don't I I I've never faded Founders that know how to build great product it how how is that split now for them like did they go and split the team and and it's like okay now half are Building blast and the other half are doing blur still and like that kind of feels disjointed in in a bit like how how does that fit together for you great question it's something that was something I asked at the Alton set um they hired

additional resources that's why they raised money um tan of course known as Pac-Man I guess is split I think he I would encourage people to go read his tweet when this came out publicly uh and it's something I talked to him about was like Hey how do you think people are going to perceive this like you're there's very clear rationale why they're building the cell2 is because as the amount of traction that they got demanded that they build an L2 and I think makes a lot of sense when you get that amount of traction it's like dydx at some point you're fa the issue of can I build an L2 or can I own my destiny in a greater capacity dox went the app chain route and I

understand that um you have more agency you have more control over certain parameters your validator said your incentives um you're not competing for a single threaded like block space fine in the case of uh blur U I think they got an incredible amount of traction uh and they saw an opportunity to build their own L2 with a more customized parameters to just make it a more unified and candidly improve the um improve the product the core offering which today is is is blur so that was a rationale I I believe it of course the resources from the top are split but I think there's a very clear Direction and where they're going it would be no different than the dydx

team going out and you know building their own app chain y I mean so what's the what's the naive like understanding for why blast is different as in L2 outside of the oneliner which Tian always goes back to oneliner like for blur was we're building a Marketplace for Traders nft Marketplace for Traders that's it open SE neglecting them in the case of blast it's a yield like a native yield bearing protocol and when you think about blast a lot of it is in these marketplac is if you have the deepest liquidity you're going to have the best execution in order to attract

liquidity there is a trade-off if that if that e deposited waiting to hit a particular threshold to sweep a floor of a particular you know nft collection it needs to be earning some like there's an opportunity cost increasingly so to your point around these other opportunities of reaking of staking and then reaking your eth and so if you don't have that natively built at the protocol level becomes difficult increasingly so to attract Traders and that is if you want to call that a oneliner really long uh oneliner but that that would be it is is to compress the opportunity cost of depositing e to

build deep pools of liquidity to improve the marketplace experience at the protocol level and that's that's really interesting that cuts to the capital efficiency right Brian you understand this better if you improve that then you're in the game for sure I mean and the like absolute Central thesis that we talk about around the office uh this year is that any L1 or L2 that isn't propagating their own Capital efficiency through a vibrant defi Community uh is not going to have a general Community there's no path to

to retail use or to just having a a uh like a a real ecosystem if you're not directly incentivizing a defi Community to be the backbone of that ecosystem and I I I confess I like blast way better than blur I think I think blur will like a number of things in you you know that we see in the space will have its day in the Sun and then let's see if if if nft become the domain of Traders or not um but one one could argue that volume is down so much and that it's really just like high value nfts that are are

actually dominating the landscape at this point so it's becoming more digital art and and that could be the constituent to to to Really lean into um that is not being adequately attended but blast is really cool because he's taking a direct approach to how can we build Capital efficiency in at the protocol there um and and I think you know if I was talking with Pac-Man right now I would say it's fine for you to split your time but split at 8020 uh don't split at 5050 yeah I mean a blur is kind of a ship it's sailing andic you know the amount of work that's required at blur at the Blast Level is much higher right you're going to you're

thinking about spinning up an entire ecosystem attract developers coordinate all of that takes time and effort and so I I'll report back on this guys I'm a judge at the first hackathon we're getting a ton of inbound of projects that are looking to deploy both new teams but also existing teams that are in other ecosystems and are trying to see if there's an opportunity uh and so I think it's going to be it's rather interesting um I think there are ways to bootstrap ecosystems arbit is probably the most in my mind and I'm biased of course has been probably the most successful L2 to to capture a very organic for the long-term kind of ecosystem of Builders and projects probably the most legitimate out

there without paying for crazy uh and giving crazy grants I said organic but see that this is the thing is is this game is just beginning in the L1 and L L to direct incentivization whether it's grants or subsidizing defi ecosystems or deploying that Capital the big story I think of 2024 and maybe over the next two years but certainly this year is how these very well- capitalized ecosystems and Foundations deploy that Capital to get what they're after which is users eyeballs Capital um you know tvl

activity so on and so forth uh and we're we're in inning one of that for sure but anyways that's sort of a defining thesis for me 2024 what is a what is a a defining one or one or two thesis um points for you guys in uh the year ahead well I mean I've it's no surpris I've tweeted about this a lot at length so not the poor listeners are you guys but um I'm quite excited I feel that I'm earlier in crypto today than I was 12 years ago because finally like we're entering that stage of you've been in a 12 years of infrastructure buildout that has rendered a lot of

business models impossible to scale historically and it's very exciting to revisit that and encouraged by the amount of activity that you're seeing in higher performance lower lower latency and higher througho chains particularly Salon in this case it is encouraging from my standpoint not because I'm ideologically a maximalist I I don't I don't [ __ ] care give me a chain that is meets the thresholds of security decentralization and performance all and because I want to see these chains we all want it like I want to see the day where crypto is uh translating its core properties in a

benefit to the consumer and and I think we're finally going to be there uh I think the quality of Builders at the app layer is improving dramatically uh anyone that is like not using protocols like on chain is pontificating stuck in Old narratives that UI ux is an issue and all this crap I think chain abstraction is going to be real something that has probably not paid much attention to is the cosmos ecosystem you have IBC solving the connectivity layer across chains Cosmos Builders have had to and I've heard this from Sam from skip Cosmos Builders from day one have had to

deal with the uh with chain abstraction and I think that positions them very well to scale out of Cosmos and then Implement IBC Across the Universe of blockchains um and and secure these you know the without compromising on security you want to connect all these different chains and so I think that's incredibly exciting soly I think we're entering historically it's paid off to invest in infrastructure over anything else in crypto vast majority of dollars have gone there I may be calling it too early I'm certainly not implementing this but

I do think that there will come a time where more dollars are going to flow and returns are going to be far superior at the app layer um Maybe not maybe far superior but I think on a proportional basis I I on the proportional basis I I think it has to right otherwise if it doesn't I I agree the the the fat protocol thesis is is sort of I ascribe to that notion I think if the token serves his purpose like you're locking it you're consuming it like value should flow down but on a relative basis of what historically has been returned if you're an infrastructure investor to an Apper investor I think the proportion of

dollars should meaningfully increase at the app layer it should and that will be a Hallmark for entering um you know a the kind of the application Phase of this industry which would dictate more mainstream adoption so I'm actually quite excited like encouraging to see Activity one of the areas we're seeing it in nft trading is kind of like will be a thing it's it's such a killer product stable coin demand is going to be a killer product anything financialization driven is going to be killer product but also gaming real gaming so that's it for me great Tommy yeah I have one question

s what do you think about the mix of apps and then app chains like because there's a world where the apps and the infra are one in the same investment when the app has its own change thex yeah exactly so it's sort of like you get the best of both worlds in that way yeah I mean topic for another discussion uh how many more successful breakthrough applications are going to go the route of building their own L2 blur blast uh dydx went their own way maker tinkering with this uh maybe this is a theme that we're we're going to see we saw it in 2017 right a lot of really smart developers

realized that they would make more money if they built their own competing L1 the incentives are just stacked maybe that's what we're going to see more and more of breakthrough applications deciding to build their own either L2 or app chain I mean Cosmos is is due for uh Resurgence I think Celestia this there was it Dimension injective like I wouldn't fade Cosmos I would certainly would fate IBC so what should we chat on on next time guys what should we leave listeners with ran you're usually pretty good at this let's hear let's hear your you know I'd like to hear one of year core thesis

of of uh of 2024 if you can if you can squeeze it out before you before you pass out and and pain in your hang over got a couple couple minutes left in me the struggle's real um just keeping you propped up over there yeah it's it's the Red Bull and the chair basically um crypto AI for sure um already covered that so I don't want to go into that in too much depth um pretty excited for the app chain thesis to to actually play out um making it really easy to create app chains I think is new design space like with Anisha and others I think um I want to see the Salon thesis play out for

that area under for lowcost transactions what apps are built there um pretty excited about the gaming side as well so a pretty big hodge podge but I'd say generally crypto AI is one of the themes I'm most interested in the user sovereignty side so running a light note on your phone with Celestia using thorchain using chain flip like building up the onchain future hypnosis is is my second theme that I really like um and then I think seeing how the app chain pieces for E fragmented value flow plays out is something I want to figure out soon all right I got an idea for next spot most

overhyped narratives and this was inspired because I did a poll on Twitter and end of 2022 it said which which out performs and it was like Bitcoin uh ethereum NASDAQ or like other I didn't even list salana in there uh funny enough like 70% selected ethereum and it was like the one that underperformed the most I ran it again yesterday I've gained more salana followers so now salana was that what ethereum was at the end of 2022 I don't want to jinx it but I think it's a good setup to uh think about non-consensus areas just things that are like overly crowded and getting ahead of

skis or something awesome yeah let's dive in I think Ryan's gonna have some Ryan's gonna have some fun stuff for us too I think next episode but oh let's see he hasn't told me coming soon he's been teasing all right well hopefully hopefully well that that then puts a hammer on like close it soon because hopefully it doesn't take four months as we took the corn right you have till next podcast you until next podcast which is in two weeks three weeks my closing ability is definitely better than our ability to get the next episode here here can't think of a better way to end this pod then all right we'll get done see you guys next year thanks guys [Music]

cheers