๐๐ฒ ๐๐๐ฌ๐๐๐ โ ๐๐๐ง๐๐ ๐ข๐ง๐ ๐ฉ๐๐ซ๐ญ๐ง๐๐ซ ๐จ๐ ๐๐ซ๐๐ ๐จ๐ง๐๐ฅ๐ฒ ๐๐๐ฉ๐ข๐ญ๐๐ฅ (๐๐)
๐๐ฒ ๐๐๐๐ ๐๐ซ๐ฒ๐ฉ๐ญ๐จ ๐๐ซ๐๐๐ข๐๐ญ๐ข๐จ๐ง๐ฌ
Iโm both going to appear to be a prophet or an fool over these predictions, however one factor is for certain: Iโm going to piss off lots of people with baggage.
Breaking this up into six sections: my predictions for L1s/L2s, token launches, stablecoins, regulation, โAI Brokersโ (oh boy), and crypto x AI.
~9 minute learn!
๐. ๐๐๐ฌ/๐๐๐ฌ
– The excellence between L1s and L2s is collapsing. Customers not understand the variations between L1s and L2s (did they ever?). The blockchain panorama, L1s and L2s mixed, is overcrowded and due for a shakeout. The consolidation might be much less about technical superiority โ will probably be about having a novel area of interest and constructing stickiness via GTM.
– Regardless of the power of SVM and Transfer, EVM market share will truly develop in 2025. This development might be pushed by @base, @monad_xyz, and @Berachain
. This might be not due to compatibility anymore โ itโll be as a result of EVM/Solidity simply has far more coaching information, and LLMs might be writing a lot of the utility code in 2025. Already having a deep library of battle-tested cryptography contracts can even be a separator, as a result of LLMs suck at writing low-level code. DevEx and footguns will matter lower than coaching information and stable libraries within the LLM period of improvement.
– Solana will stress extra blockchains to optimize for low latency. We’ll transfer from TPS wars to latency wars โ infra like @doublezero and tremendous low-latency L2s like @megaeth_labs
will push person expectations towards web2 responsiveness. Anticipate extra embrace of optimistic UIs, preconfirmations, intents, electronic mail onboarding, in-browser wallets, and progressive safety. Shoutout to @privy for advancing the meta right here.
– @HyperliquidX has demonstrated that specialised chains can work after theyโre laser-focused on a selected utility and prioritize UX and simple bridging. Extra initiatives will comply with this mannequin. The outdated dream of 1 chain to rule all of them is useless.
๐. ๐๐จ๐ค๐๐ง ๐๐๐ฎ๐ง๐๐ก๐๐ฌ
– The present meta of everybody doing large airdrops by way of factors applications is over. We’re transferring to a two-track world.
– Monitor one: if a venture has a transparent north star metric, like an alternate or a lending protocol, they may distribute tokens purely off factors. They won’t care if they’re farmed or gamed โ they’re successfully distributing the token as a rebate/low cost on the core KPI of the protocol, and the farmers are your precise customers anyway.
– Monitor two: initiatives with out clear north star metrics (like L1s and L2s) will transfer towards crowdsales. They might do smaller airdrops to reward social contributions, however the majority of tokens will get distributed by way of crowdfunding. Airdropping for self-importance metrics is useless. These arenโt actually going to customers, theyโre going to industrialized farmers.
– Memecoins will proceed to lose market share to โAI agentโ cash. I contemplate this a migration from monetary nihilism to monetary over-optimism. (Yep Iโm coining that.)
๐. ๐๐ญ๐๐๐ฅ๐๐๐จ๐ข๐ง๐ฌ
– Stablecoin utilization will explode, notably amongst SMBs. Not simply buying and selling and hypothesis โ actual companies will begin utilizing on-chain {dollars} for fast settlement.
– Banks are noticing: count on to see bulletins of bank-issued stablecoins towards the tip of 2025. They won’t wish to be left behind. However particularly with Lutnick as Secretary of Commerce, Tether will stay #1.
– Anticipate @ethena_labs to gobble up much more capital, particularly as treasury yields proceed to say no over the approaching yr. When the chance price of capital declines, it makes foundation commerce yields much more enticing.
๐. ๐๐๐ ๐ฎ๐ฅ๐๐ญ๐ข๐จ๐ง
– Stablecoin laws passes within the US, whereas the broader market infrastructure overhaul (FIT21) will get delayed. Stablecoin adoption accelerates whereas Wall Road adoption, asset tokenization, and different TradFi integration will lag behind.
– Below Trump, Fortune 100 corporations will grow to be extra prepared to supply crypto to shoppers, with tech corporations and startups exhibiting greater danger urge for food. Trumpโs inauguration will create a perceived regulatory jubilee till clear guidelines and enforcement priorities are set. Throughout this window, count on to see aggressive growth of crypto integration into Web2 platforms.
๐. ๐๐ ๐๐ ๐๐ง๐ญ๐ฌ (๐ญ๐ก๐ข๐ฌ ๐ข๐ฌ ๐ญ๐ก๐ ๐ฅ๐จ๐ง๐ ๐๐ฌ๐ญ ๐ฌ๐๐๐ญ๐ข๐จ๐ง ๐๐๐๐๐ฎ๐ฌ๐ ๐ฆ๐ฒ ๐ญ๐ก๐จ๐ฎ๐ ๐ก๐ญ๐ฌ ๐ก๐๐ซ๐ ๐๐ซ๐ ๐ฅ๐ข๐ค๐๐ฅ๐ฒ ๐๐จ๐ง๐ญ๐ซ๐จ๐ฏ๐๐ซ๐ฌ๐ข๐๐ฅ โ ๐ซ๐๐๐ ๐ญ๐จ ๐ญ๐ก๐ ๐๐ง๐!)
– The โAI agentโ craze will proceed in all probability all through 2025. However it would die off ultimately. This isn’t the long-term disruption to be careful for from AI, however will probably be CTโs fixation as a result of it’s the most social.
– This stuff aren’t actually brokers. These are chatbots with memecoins hooked up; they’re barely agentic in any respect in addition to posting on Twitter. Present โAI brokersโ are additionally principally โWizard of Ozโ brokers โ there are people behind the scenes guaranteeing the AI doesnโt go off the rails. This receivedโt change any time quickly as a result of present brokers are too janky (even Fortune 100 corporations aren’t utilizing brokers in prod but). Present brokers can simply be manipulated into saying loopy issues that harm their manufacturers, or will be jailbroken to steal all of their assets. See @freysa_ai for what an precise autonomous AI appears like โ in case your favourite AI is just not getting jailbroken, itโs as a result of itโs a Wizard of Oz AI.
– That mentioned, I believe this development will speed up. Chatbots can certainly exchange plenty of influencers as a result of chatbots by no means sleep, theyโre all the time on-message, and so theyโre much less grasping than human influencers. Plus nearly all of influencers arenโt very authentic anyway. Actual-time data aggregation/amplification will be simply changed by an algorithm even at present (see @aixbt_agent).
– Proper now these chatbots are fascinating to us as a result of they’re so novel. Itโs like seeing an elephant paint. The primary time you see it, you donโt actually care that the portray is just not excellent โ itโs spectacular to see. However the one thousandth time, the novelty wears skinny. I imagine that may begin to occur as these chatbots plateau.
– You possibly can see that at present with aixbt โ itโs already fairly good at aggregating information about completely different initiatives. By subsequent yr and the subsequent era of brokers, perhaps aixbt will hallucinate rather less, go a bit deeper, have a bit smarter takes. However how a lot will you even discover? Itโll in all probability really feel the identical to most individuals.
– I believe this novelty and market eagerness continues all through 2025. Crypto takes some time to get tired of the shiny factor. However by 2026 I believe there might be a sudden reversal. The chatbots will grow to be so ubiquitous that folks will get turned off by them. Sentiment will reverse. Seeing tales of their favourite human KOLs shedding their livelihoods will kindle a sort of class consciousness. Customers will begin discriminating in favor of human KOLs, even when their content material is much less constant.
– In response to this pro-human bias, chatbots will begin hiding that they’re AIs, making an attempt to cross as people so as to seize extra of the eye market. As an alternative of monetizing via memecoins like at present, future chatbots will monetize the identical method human KOLs do โ via sponsorships, affiliate hyperlinks, and pumping tokens they personal. KOLs might be routinely accused of being chatbots, and you will notice AI-unmasking scandals. This may all get bizarre.
– However thereโs a darker facet but. Bear in mind, LLMs are at the moment nice wordcels, however not nice on the different stuff but. What are the very best methods to become profitable as a wordcel in crypto? First is being an influencer, certain, however shut second is being a scammer. You’ll begin seeing autonomous scambots proliferate. These will explode, corresponding to what ransomware and cryptojacking turned put up 2017. Anticipate this to grow to be an actual social downside.
– However whereas chatbots are prone to stay the focus in 2025, the long-term disruption from AI is not going to be on the social layer.
– And no, itโs not going to be in buying and selling both. AIs is not going to give everybody their very own โbuying and selling agentโ or miniature hedge fund. Sure, AIs will scale everybody, however they may scale individuals proportionally to their capital, information, and infrastructure. You must due to this fact count on AI to supercharge preexisting buying and selling corporations who’ve capital scale and information scale. In different phrases, buying and selling corporations will grow to be even higher at making the entire cash. It’s going to additionally collapse the hierarchy amongst buying and selling corporations (most of them will grow to be comparably good, since everybody may have entry to 150 IQ quants within the cloud).
– Over time, AIs will make markets extraordinarily environment friendly โ even smaller, area of interest markets โ which is able to go away little edge left for regular merchants, even with their little homebrew assistant AIs. The worth of authentic analysis will plummet. That mentioned, the elevated competitors and liquidity must be a boon to the remainder of us who’re injecting noise into the market. (It’s going to additionally imply @Polymarket liquidity on every thing!)
– So if the massive story is just not chatbots and never buying and selling bots, what else is there? Right hereโs my core thesis, which for some purpose nearly no person is speaking about: the actually impactful AI brokers might be software program engineering brokers.
– Why is that this such an enormous deal? Ask your self this: what’s the major enter to our business? What’s the expensive enter stopping there from being extra purposes, extra wallets, higher infrastructure, higher every thing? The reply is software program. If AI brokers trigger the worth of software program to break down, that may change every thing.
– In a post-AI period, as an alternative of getting to boost hundreds of thousands of {dollars} for a seed spherical, it is possible for you to to launch an utility with $10K of AI cloud compute. Self-financed initiatives like Hyperliquid and Jupiter will go from the exception to the norm. The quantity of purposes and experimentation on-chain will completely explode. For an business that’s pushed by software program, this deflationary shock goes to result in an on-chain renaissance.
– The implications of this on safety are profound. AI-powered static evaluation and monitoring will grow to be ubiquitous, making safety extra accessible to everybody. These AIs might be fine-tuned on EVM/Solidity or Rust codebases, skilled on huge databases of safety audits and assault vectors. Theyโll be RLโd in simulated adversarial blockchain environments. Iโm more and more satisfied that AI instruments in the end favor defenders over attackers with regards to safety. You’ll have AIs continuously red-teaming contracts, whereas different AIs might be hardening them, formally verifying their properties, and honing their abilities at incident response and remediation.
– Within the meantime, certain, commerce AI-flavored memecoins. However actual brokers are going to have much more affect than tweeting and pumping their very own tokens.
๐. ๐๐๐ญ๐ฎ๐๐ฅ ๐๐ซ๐ฒ๐ฉ๐ญ๐จ ๐ฑ ๐๐
– Above I detailed the affect of AI on crypto (which is the first path of affect), however crypto can even have an effect on AI.
– Really autonomous brokers will use crypto to pay one another. This might be very true as soon as there are permissive stablecoin rules โ youโll begin seeing even giant corporations that run AI brokers utilizing stablecoins for agent-to-agent funds as a result of theyโll be a lot simpler to spin up than financial institution accounts.
– We can even see extra and greater scale experimentation round decentralized coaching and inference. A brand new era of promising initiatives like @exolabs, @NousResearch, and @PrimeIntellect will pave the way in which for actual options to centralized coaching and company-owned fashions. @NEARProtocol can be going all-in on making an attempt to create a full-stack credibly impartial and permissionless AI stack.
– The opposite place the place crypto and AI will intersect is UX. Put up-AI wallets might be fully reworked โ an AI powered pockets ought to be capable of care for bridging, optimize commerce routes, decrease charges for you, paper over interoperability points or frontend bugs, and steer you away from apparent scams or rugpulls. You receivedโt be juggling between a number of completely different wallets and altering RPCs or rebalancing your stablecoins โ the AI will deal with all of it for you. This seemingly takes till 2026 to grow to be dependable sufficient to remodel cryptoโs UX. However when this arrives, what does this do to blockchain community results? What occurs when customers cease caring โ and even experiencing โ which chain an utility lives on?
– This area remains to be younger, however Iโm hopeful weโll see issues take off right here quickly. In the long term (say mid 2026) I count on this might be the place a lot of the market cap of โAI x cryptoโ lives.