So I Just Made 2025 predictions

Okay. So I just made my 2025 predictions, and somebody was asking me to see if I could record a little mini podcast over it. So here's my attempt to try to boil it down for people who hate reading, which I'm sure includes you. I'll try to be quick, so the goal is to get all this done in 10 minutes. There's gonna be 6 sections, l ones, l twos, token launches, stablecoins, regulations, AI agents, which is gonna be a big one, and then crypto AI.

So in L1s and L2s, first, the distinction between L1s and L2s is gonna collapse. Users aren't gonna perceive the differences between L1s and L2s if they ever did, which is questionable. The blockchain landscape, including L1s and L2s, is super overcrowded, and it's due for a shakeout. So I think you're gonna see consolidation, which will be less about what matters with respect to technical superiority, but it'll be more about having a unique niche a unique identity and building stickiness through a strong go to market. Now on VMs, what I think you'll see is that, SVM and MooV are gonna do well, but EVM market share is actually going to grow in 2025, despite the fact that it's already the biggest.

I think this is gonna be driven by base, by Monad, and by BaraChain. It's not gonna be primarily because of compatibility anymore. It'll be because the EVM in Solidity just has way more training data, and LLMs are gonna be writing most of the application code in 2025. So having a deep library of battle tested cryptography contracts will be a separator. Because LLMs, right now, they suck at writing low level code.

They're very good at writing blue code, very good at, munching together a lot of high level, concepts. But the details, they still suck at, and I think they'll be sucking at it for for a while hence. So DevEx and foot guns are just gonna matter less when training data and solid libraries are going to be ubiquitous in the EVM ecosystem. And I think in the LM era of development, that's going to dominate. Now that being said, Solana is going to continue to pressure more blockchains to optimize for low latency.

So I think we're gonna move away from the TPS wars, which has been the primary metric people obsessed about for the last, call it, you know, 5 years. I think it's gonna move more toward an obsession with latency. So infra like double 0, as well as super low latency l twos like mega ETH, are gonna push user expectations toward web 2 style responsiveness. So expect more optimistic UIs, precomps, intense email onboarding in browser wallets, and progressive security. Privy has done a big a lot of work here in advancing the meta alongside, of course, Hyper Liquid.

Now Hyper Liquid has demonstrated that specialized chains can work when they're laser focused on a specific application and prioritize UX and easy bridging. So more projects, I think, will follow this model, and the dream of 1 chain to rule them all is dead. Okay. So that's for l ones and l twos. 2nd is token launches.

What's gonna change in token launches? First, the meta of everybody doing huge airdrops via points programs is over. We're moving to a 2 track world. Track 1 is for projects that have a clear North Star metric. So if you're in exchange, you're a lending protocol, you've got a clear KPI, and you're gonna distribute points purely off of those KPIs.

You're not gonna care if you're being farmed or you're being gained. You're effectively distributing tokens as a rebate or discount on the core activity of your protocol, and the farmers are basically your users anyway. You're actually trying to convert them. Now track 2 is products that don't have a clear North Star metric. So these are things like l ones, l twos.

These are gonna move toward crowd sales. They may do smaller airdrops to reward social contributions and things like that, but but the majority of tokens are gonna get distributed via crowdfunding. Airdropping for vanity metrics is dead. Those aren't going to users. They're now going to industrialized farmers, and they don't work.

Lastly, meme coins, I think, are gonna be losing market share to AI agent coins. I consider this to be a migration from financial nihilism to financial overoptimism. And, yes, that is a term I just coined. Okay. 3rd up is stablecoins.

Stablecoin usage is going to explode, particularly among S and B, small, medium sized businesses, not just for trading and speculation. Real businesses are gonna start using on chain dollars for instant settlement. You're already starting to see this. There's been a lot of growth in 2024. 2025, it's gonna hit an inflection point.

Now banks are taking notice. Expect to see the announcements of bank issued stablecoins toward the end of 2025. They're not gonna be left behind. They're gonna need a little bit of regulatory clarity to actually be able to start running toward the the front door. But especially with Letnick as secretary of commerce, I expect this to get pretty serious.

And Tether, despite all of this, will probably remain number 1. Expect Athena to gobble up even more capital, especially as treasury yields decline over the coming year. When the opportunity cost of capital declines, it's gonna make basis trace basis trade yields even more attractive. Okay. Next up, regulation.

On regulation, stablecoin legislation is gonna pass in the US, but the broader market infrastructure overhaul, 5 21, is gonna be delayed. Stablecoin adoption is gonna accelerate, while Wall Street adoption, things like asset tokenization, other stratify integrations, I expect them to lag behind and be a little bit slower. Under Trump, you will start to see fortune 100 companies become more willing to offer crypto to consumers. We're in a new era now. The debanking narrative is behind us, and and all the stuff about the Libra, I think, is gonna energize a lot of these companies to become more aggressive.

You'll see tech companies and startups showing higher risk appetite relative to old school companies. But I think up until, you know, shortly after Trump's inauguration, there's gonna be this perception that there's a little bit of a regulatory jubilee that maybe nobody's watching. And until clear rules and enforcement priorities are set, people are gonna take a lot more risk. So expect during this window a lot more aggressive expansion into crypto for a lot of the web 2, super platforms. Okay.

Now next up, AI agents. This part is gonna be meaty, so listen up. This AI agent trend, this this craze that you've probably been seeing of all these AI agent tokens going crazy, I think this is gonna continue through 2025. It's not going anywhere. But it will die off eventually.

This is not the long term disruption to watch out for from AI, but it will be Crypto Twitter's fixation because it's just the most social. These things are not really agents. They're chatbots with meme coins attached. They're barely agentic at all besides just posting on Twitter. You know, agents must take actions.

What's the action that these things are taking besides talking on social media? Now the the other thing is that these current AI agents, they're mostly Wizard of Oz agents, meaning that they're humans behind the scenes ensuring that the AI doesn't go off the rails. This won't change anytime soon because the current generation of agents are too janky. Even fortune 100 companies aren't using agents abroad yet. Plus, current agents can easily manipulate it into saying crazy things that are gonna damage their brands, or they can be jailbroken to steal all their money.

So see Frasier for what an actual autonomous AI looks like. If your favorite AI is not getting jailbroken, it's because it's a wizard of Oz AI. It's not actually operating autonomously. That being said, all that being said, I do think this trend is gonna accelerate. Chatbots can replace a lot of influencers because chatbots, you know, they never sleep.

They're always on message. They're less greedy than human influencers. Plus, the majority of influencers aren't very original anyway, so they're not that hard to replace. And real time information aggregation or amplification, it's a thing a lot of influencers do. It can be easily replaced by an algorithm even today.

So if you look at AI SPT, it's kind of already better than a lot of influencers today. Now right now, these chatbots, they're fascinating to us because they're so novel. It's it's like seeing an elephant paint. The first time you see an elephant paint, you're like, wow. That's amazing.

You don't really care if the elephant is not very good. It's just spectacular to see that. But the 100th time, the novelty starts to wear thin. And I think that's gonna start to happen as these chatbots plateau. So you can see that already with AIXBT.

You know, it's pretty good at generating aggregating data about different projects and generating some, you know, cursory insights. It's not something you're necessarily gonna be trading on, but it's like, oh, okay. You know, here's some new thing that happened and that's being contextualized for me. By next year and the next generation of agents and foundation models, maybe AXPT will hallucinate a little bit less. Maybe it'll go a little bit deeper.

Maybe it'll have a little bit smarter takes. But how much are you going to even notice? It'll probably feel the same to most people. And so I I think this novelty and this market eagerness is gonna continue throughout 2025. Crypto takes a while to get bored of the shiny thing.

But by 2026, I think there's gonna be a reversal. I think these chatbots will become so ubiquitous that people are gonna get turned off by them. Sentiments can reverse. Seeing stories of their favorite influencers, human influencers, losing their livelihoods is gonna kindle a kind of class consciousness. Users will start discriminating in favor of human influencers even if their content isn't as good, if it's less consistent.

And in response to this pro human bias that they start seeing among users, chatbots will start hiding that they are AIs, and they'll start trying to pass as humans in order to capture more of the attention market. They're no longer gonna be able to monetize through meme points like they can today. Instead, future chatbots are gonna be monetizing the same way that human influencers do, through sponsorships, affiliate links, and pumping tokens they own. And these and and what you'll see is that influencers, both real influencers and AI influencers, are gonna be routinely accused of being chatbots. And you're gonna see scandals where influencers get unmasked as AIs.

It's it's gonna get really fucking weird. But there's a darker side. Right? Remember, LLMs are currently great at being word sells, but they're not great at the other stuff yet. So what are the best ways to make money if you're a word sell in crypto?

First answer is probably, yeah, be an influencer. But the second way is to be a scammer, and it's a pretty close second. So I think you're gonna start to see autonomous scam bots proliferate. These are gonna explode. It's gonna be comparable to what ransomware and cryptojacking became post 2017.

I I think this is gonna become, to be honest, a real social problem, and crypto is, of course, the easiest way to scam people. But while chatbots are likely to remain the center of attention in 2025 because they're so social and because they're so visible, the long term disruption from AI is not gonna be at the social layer. And, no, it's also not gonna be in trading. AIs are not gonna give everyone their own trading agent or little mini hedge fund. Yes.

AIs are gonna scale everyone, but they will scale people proportionally to their capital data and infrastructure. You should therefore expect AI to supercharge preexisting trading firms who have capital scale and have data scale. In other words, trading firms are gonna become even better at making all the money. But it's it's also gonna collapse the hierarchy among trading firms. Most of them are gonna become comparably good because everyone's gonna have access to these 150 IQ quants who live in the cloud.

Over time, these AIs are gonna make all markets extremely efficient, even smaller niche markets, which is gonna leave little edge left for normal day traders, even if you have your little homebrew assistant AI. The value of original research should plummet. That being said, the increased competition and liquidity should be a boon to the rest of us who are just injecting noise into the market with our trading. Also means poly market liquidity on almost everything. But if the big story is not chatbots and the big story is not trading bots, what else is there?

Now here's my core thesis, which for some reason very few people are talking about, which is that the truly impactful AI agents are going to be software engineering agents. Okay. Now why is this such a big deal? Ask yourself this. What is the primary input to our industry?

What is that cost that is preventing there from being more applications, more wallets, more infrastructure, more everything? The answer is software. Software is the primary input in our industry. If AI agents cause the cost of of software to plummet, that is going to change everything. In a post AI era, instead of having to raise 1,000,000 of dollars for a seed round, you'll be able to launch an application with $10,000 of AI cloud compute.

Self finance projects like Hyper Liquid and Jupyter, they'll go from being the exception to being the norm. The amount of applications and experimentation on chain is going to explode. For For an industry that's driven by software, it's a deflationary shock, and this is going to lead, in my estimation, to an on chain renaissance. Right? The implications on this insecurity are also really significant.

So AI powered static analysis and monitoring are gonna become ubiquitous. Security is gonna be available to everyone. Doesn't matter how good or bad the the particular software is. It's gonna be really cheap for anything to become secure. These AIs that are gonna do security are gonna be fine tuned on Solidity, EVM, Rust code bases, trained on vast databases of security audits and attack vectors.

They're gonna be reinforcement learning trained in simulated adversarial blockchain environments. You know, I'm more and more convinced that AI tools actually favor defenders over attackers when it comes to security. So you're gonna have AIs red teaming contracts while other AIs will be hardening them, formally verifying their properties, and honing their skills at incident response and remediation. Now in the meantime, many people are gonna be trading AI flavored meme coins, and that's fine. I don't begrudge that.

But real agents are gonna have a lot more impact than just tweeting and pumping their own tokens. Okay. That was a long last section, crypto AI. Above I do you know, in the section before, I detailed the impact of AI on crypto, which is the primary direction of influence, but crypto will also have an impact on AI. So you're gonna see truly autonomous agents that will use crypto to pay each other.

And this is gonna be especially true when they're permissive stablecoin regulations because you start to see even large companies that run AI agents using stablecoins for agent to agent payments because it'll just be easier to spin up than bank accounts. They're instantaneous. And, of course, stablecoins and and crypto wallets are trivial for computers to operate relative to bank accounts. You're also gonna see bigger and larger scale experimentation around decentralized training and inference. A new generation of promising projects like Exo, News Research, PrimeIntellect, they're gonna pave the way for real alternatives to centralized training in company owned models like we have today.

NEAR Protocol is also going all in on trying to create a full stack, incredibly neutral permissionless AI stack. I think this is gonna matter as you start to see compression where the low end of models or the medium end of models start to really be competitive with the high end of models, which is already starting to happen today, especially when the the scaling for these things is going towards inference time compute rather than towards pretraining. The last place where crypto and AI is gonna intersect is gonna be in UX. So post AI wallets will be completely transformed. An AI powered wallet will be able to take care of all the stupid stuff that humans have to manage right now.

So think about, you know, bridging, optimizing your trading route, minimizing fees, and how much is this does this front end charge more money than that front end? Papering over interoperability issues, or you have some front end bug and you can't figure out what's wrong, or steering you clear of scams or rug pulls or just, you know, why is it that you're not going and checking the Twitter every time you interact with something to see if if they announce it? Like, hey, don't interact with this website. Because you're lazy. You don't have time.

But an AI is never gonna be lazy. It's always gonna be willing to do it. You won't be juggling multiple different wallets or changing RPCs or rebalancing your stable coins. The AI should be able to handle all of that for you. It doesn't actually even require that much intelligence.

Now it's likely gonna take a year or so, maybe up until 2026, for it to be reliable enough to really transform the UX of crypto. But when this arrives, the question you should be asking yourself is what does this do to blockchain network effects? What happens when a user stops caring or even experiencing what chain an application lives on? Now all this stuff is still young, and I'm I'm hopeful it's gonna take off from here. But in in the long run, in, let's say, you know, mid 2026, I expect that most of the value in the intersection of crypto and AI is gonna live in this kind of intersection, not in the AI meme coins.

So that's all I got for predictions. I hope that was helpful. I try to make this relatively short, capping it at less than 15 minutes. But if you listen to that, congratulations. And, I hope I'm gonna be right about at least a few of these things so I don't look like a complete idiot.

Happy New Year.