Web3 Unhashed Podcast

2025 saw a structural reset on the NFT market, with flights to fungible liquidity, the decoupling of IP success from asset prices, and a consolidation of volume into revenue-generating verticals. Here were the major NFT & gaming trends in 2025 🧵

“Parallel execution isn’t a nice-to-have — it’s what makes on-chain apps usable for normal people.” @sachimihara is joined by @jayendra_jog (Founder of @SeiNetwork) to outline which Ethereum-native apps stand to gain the most from a parallel stack—and why higher throughput is ultimately about usability, not benchmarks. Jay points to trading and DeFi as the clearest winners. If an 18-year-old wants to buy $20 of a meme coin but has to pay $15 in fees, the experience collapses—low-TPS environments don’t just slow users down, they price them out entirely.

“Watching Robinhood ‘turn off buys’ during GameStop made one thing painfully clear: the rails still have gatekeepers.” @kenzimerkle catches up with @jayendra_jog (Founder of @SeiNetwork) to break down how Jay’s pre-crypto journey—joining Robinhood in 2017 when it was still scrappy in Palo Alto, seeing it scale 10x and go public—ultimately pushed him toward building in Web3.

“Virtual machines are like cities — once they’re established, they become magnets.” @dikshalyer sits down with @jayendra_jog (Founder of @SeiNetwork) to unpack Jay’s macro thesis: why “OG” hubs—whether cities like NYC and SF or dominant virtual machines—stay dominant even when their flaws are obvious. Jay explains the intuition behind the analogy from personal experience living in New York and San Francisco: congestion, safety issues, homelessness, and other structural problems don’t automatically push people away—because the real gravity comes from where the activity, networks, and opportunity already are.

“Bitcoin was the first distributed systems paper I read with an economic layer built into it — and that changed everything.” @kenzimerkle catches up with @averyching, Co-Founder & CTO of @Aptos, to trace his journey from high-performance computing and supercomputers, to scaling data infrastructure at Meta, to discovering Bitcoin and realizing that crypto was distributed systems with incentives natively embedded — the insight that ultimately led him to co-found Aptos Labs.

“We focused on four areas: finance, gaming, social, and entertainment — but DeFi on @Aptos has seen the biggest pickup.” @sachimihara sits down with @averyching to break down Aptos’ real-world use cases and why DeFi has become the standout: Move’s safety plus deep composability that lets products snap together into bigger protocols — and an ecosystem that’s finally hitting momentum.

“What inspires you to get up and build every day? For me, it’s pushing Web3 forward — making blockchain a true public utility for everyone.” @dikshalyer sits down with @AveryChing (Co-Founder & CTO of @Aptos) to talk about what drives him: building the next era of the internet where blockchain brings ownership back to users and enables permissionless, trustless transactions that connect people globally.

New episode out today featuring @AveryChing — Co-Founder & CTO of @Aptos. A deep dive into the intersection of crypto and AI: Aptos’ fundraising journey, how it stacks up against other Layer 1s like Solana and Ethereum, and what’s next for the Move programming language. Avery shares his outlook on decentralized use cases, long-term goals for Aptos, and how his decade of experience scaling distributed systems at Meta — including work on the Diem blockchain — shapes his vision for the future of Web3 infrastructure.

Small Models Could Crack the Private AI Problem A new study reveals small models with curated "Skills" can outperform frontier models, potentially unlocking fully homomorphic encryption for practical use and finally delivering mathematically guaranteed privacy. Here's the breakdown and what can be delivered👇 ~~ Analysis by @punk5268 ~~ The Privacy Spectrum: Trust, Hardware, Math Current privacy approaches remain trust-based, either through trusting providers not to log queries, or trusting hardware that researchers have repeatedly compromised. These improve on baseline cloud inference but still require trust. Cryptography. Fully homomorphic encryption (FHE) enables computation on encrypted data without decryption—the only mathematically guaranteed solution. The model processes encrypted inputs and returns encrypted results. Your data never exists unencrypted outside your device. The Obstacle: FHE Remains Computationally Intensive The bottleneck. According to @zama , current FHE performance is only feasible with small, old models like GPT-2 scale. Full encryption for frontier models remains years away. But that assumes you need frontier models for frontier-quality results. The Breakthrough: Small Models With Skills Beat Big Models Without Them This month, SkillsBench tested whether smaller models augmented with curated "Skills" (structured step-by-step guides popularized by Claude) could match larger ones. Results: - Claude Haiku 4.5 (small) + Skills: 27.7% pass rate - Claude Opus 4.5 (large), no Skills: 22.0% pass rate The small model with focused guidance outperformed the big model running without them. Healthcare tasks saw the largest improvement at +51.9 percentage points. Two critical findings emerged. First, 2-3 focused Skills outperformed comprehensive documentation—quality over quantity. Second, when researchers prompted models to generate their own skills, performance plummeted. Models can consume curated expertise but cannot create it. Human curation remains essential. Why This Matters for FHE The implication. Healthcare, finance, and legal—the domains most needing privacy—are exactly where Skills deliver the biggest gains. If small models with curated knowledge can punch above their weight, the bar for "good enough" drops. We don't need FHE to scale to frontier models; we need models small enough to be FHE-friendly yet capable enough to be useful. The gap. Zama demonstrated FHE on 2019-era models while SkillsBench tested on Haiku 4.5, released last year. Six years of model evolution—including the advent of ChatGPT—separate what FHE can handle from what SkillsBench proved. Skills augmentation promises to bridge this gap, but success remains unproven. The timeline compresses, but not to tomorrow. An Interim Bridge Local inference. The models powering Apple Intelligence, Gemini Nano, and open-source equivalents already run locally. While "local" by default does not mean "private," proper vetting could change that. Here, ERC-8004 provides a critical bridge. Even if a model is open source, this reputation layer verifies it isn't phoning home to a central provider. Combined with Skills to close the capability gap, local models could offer practical privacy today. What This Could Look Like For users worried about AI's creeping access to personal data, Ethereum-aligned AI offers concrete alternatives to "trust us" or "opt out." For developers, capable private models—whether through FHE cryptography or vetted locality—are closer than many believe. If you're building toward FHE-wrapped inference, Skills-augmented small models, or reputation layers for local AI, the infrastructure is assembling rap

a sharp and honest discussion on builder mentality incentive design and why the best lesson in web3 is still to keep going no matter what

we put together a really strong episode here kenzi and michael go deep on what makes web3 feel fundamentally different and why the builder mindset matters more than ever

“Biggest learning building in Web3? First principle is still: don’t die — don’t quit, keep finding a way.” @kenzimerkle sits down with @michaelh_0g to reflect on the shift from Web2 to Web3: why Web2 often moves from user-aligned to extractive incentives, why Web3 feels fundamentally different, and how embracing that mindset made building feel fun again.

“DA layers differ on three axes: performance, programmability, and AI-first design — because on-chain AI can’t live in a ‘megabytes-per-second’ world.” @sachimihara sits down with @michaelh_0g (Founder of @0G_labs) to compare against Celestia, Avail, and EigenDA: why throughput must jump orders of magnitude, how to avoid the broadcast bottleneck, and why a decentralized storage network is key for ultra-fast data ingestion + retrieval.

“Bored at a new school after moving from Berlin to Silicon Valley, I started hanging out at my dad’s SAP Lab — fast internet, endless reading, and falling in love with tech.” Our host @kenzimerkle talks with @michaelh_0g (@0G_labs) about his origin story: early curiosity, the path into Web3, and how it all led to starting his company.

“Back in 2016–17, crypto felt like an idea factory — hundreds of experiments, zero permission, pure creativity.” Our host @dikshalyer sits down with @michaelh_0g, Founder of @0G_labs, to talk about the most fun part of building in Web3: an experimentation-first culture, and how tokenization unlocks new ways to sustain projects — including open-source — beyond the standard Web2 business model.

New episode this week, featuring @michaelh_0g, founder of @0G_labs. A deep dive into AI x crypto: modular AI chains, decentralized AI infrastructure, community-owned data, and why modularity is critical for scaling AI on-chain. Also: Michael’s founder journey, leadership through meditation and spiritual practice, and the real-world path to building the first modular AI chain.

How NFT "Mechs" Are Supercharging RAILGUN's Privacy RAILGUN crossed $4.5B in shielded volume in 2025, yet complex DeFi transactions remain out of reach for its privacy suite. Gnosis Guild's NFT-based "Mechs" are set to unlock non-atomic transactions while keeping balances fully private. Here's the technical breakdown 👇 ~~ Analysis @punk5268 ~~ RAILGUN 101 @RAILGUN_Project is an onchain privacy suite deployed on Ethereum, Arbitrum, and Polygon. It uses zero-knowledge cryptography to shield ERC-20s and NFTs from public view. Users receive a 0zk address, a private wallet where shielded assets become indistinguishable from each other. Amid crypto's privacy renaissance, RAILGUN has emerged as a key protocol, crossing $4.5B in total shielded volume in 2025. A Mech can execute any onchain action, from basic swaps to complex DeFi positions. Ownership is flexible, controllable by specific NFT or ERC-20 holders. This flexibility provides the breakthrough RAILGUN needs for its privacy suite. The RAILGUN integration Current 0zk addresses handle atomic DeFi transactions (single-block actions like simple swaps). However, they cannot manage non-atomic activities like borrowing, staking, or multi-sig transactions. Under RAILGUN Connect, the protocol's universal private DeFi connector, a Mech sits between your 0zk address and target protocols. The flow works as follows: deposit unshielded tokens into your Mech, shield the Mech into a 0zk address, then execute calls via Zodiac Pilot (a Gnosis Guild tool) to interact with DeFi protocols. Protocols see only a normal account conducting standard transactions while your actual balance remains invisible onchain. What to watch The RAILGUN x Mechs integration remains in development but nears production readiness. Gnosis Guild proposed the integration to RAILGUN governance in April 2025. $RAIL governors ratified it shortly after, with development ongoing since. In January 2026, Gnosis's @auryn_macmillan demoed RAILGUN Connect using the @CoWSwap frontend to execute a private swap on @0xPolygon. No timeline exists for full deployment, but the prototype suggests launch is imminent. Post-launch, expect broader adoption. The Ethereum Foundation's upcoming Kohaku privacy wallet is integrating RAILGUN's tech, creating an avenue for private non-atomic DeFi transactions. Others will follow. ~~ RAILGUN 101 @RAILGUN_Project is an onchain privacy suite deployed on Ethereum, Arbitrum, and Polygon. It uses zero-knowledge cryptography to shield ERC-20s and NFTs from public view. Users receive a 0zk address, a private wallet where shielded assets become indistinguishable from each other. Amid crypto's privacy renaissance, RAILGUN has emerged as a key protocol, crossing $4.5B in total shielded volume in 2025. Rise of the Mechs A Mech can execute any onchain action, from basic swaps to complex DeFi positions. Ownership is flexible, controllable by specific NFT or ERC-20 holders. This flexibility provides the breakthrough RAILGUN needs for its privacy suite. The RAILGUN integration Current 0zk addresses handle atomic DeFi transactions (single-block actions like simple swaps). However, they cannot manage non-atomic activities like borrowing, staking, or multi-sig transactions. Under RAILGUN Connect, the protocol's universal private DeFi connector, a Mech sits between your 0zk address and target protocols. The flow works as follows: deposit unshielded tokens into your Mech, shield the Mech into a 0zk address, then execute calls via Zodiac Pilot (a Gnosis Guild tool) to interact with DeFi protocols. Protocols see only a normal account conducting standard transactions while your actual balance remains invisible onchain. What to watch The RAILGUN x Mechs integration remains in development but nears production readiness. Gnosis Guild proposed the integration to RAILGUN governance in April 2025. $RAIL governors ratified it shortly after, with development ongoing since. In January 2026, Gnosis's @auryn_macmillan demoed RAILGUN Connect using the @CoWSwap frontend to execute a private swap on @0xPolygon. No timeline exists for full deployment, but the prototype suggests launch is imminent. Post-launch, expect broader adoption. The Ethereum Foundation's upcoming Kohaku privacy wallet is integrating RAILGUN's tech, creating an avenue for private non-atomic DeFi transactions. Others will follow.

The NFT Invasion Nobody Noticed While headlines obsess over JPEG speculation, NFTs have quietly become the backbone of modern DeFi. Uniswap positions, Polymarket bets, and Sablier payment streams all run on ERC-721s. The infrastructure takeover is already happening 👇 ~~ Analysis by @dikshalyer ~~ Why Everything Will Be an NFT "Everything will be an NFT." So tweeted @pbrody, Chair of the Enterprise Ethereum Alliance. As an NFT wonk, I concur. Ontologically, NFTs are ownable handles for unique onchain state. Their flexibility and programmability position them to permeate every area of human life. The Infrastructure Reality These protocols already use NFTs as core infrastructure, though none market themselves as "NFT projects": > @Polymarket. Underpins Yes/No prediction market outcomes with conditional ERC-1155s. > @Uniswap. V3 and V4 liquidity positions are minted and managed through ERC-721s. > @LiquityProtocol "Trove" borrow positions transfer as ERC-721s, enabling onchain debt markets. > @eth_strategy. Issues long bonds composed of American call options represented as ERC-721s. > @flaunchgg. Lets coin launchers access trading fee revenues via ERC-721 Royalty NFTs. > @Sablier. Represents Flow and Lockup-style real-time payment streams as ERC-721s. > @ensdomains. Domain ownership and transferability function via ERC-721s. > @Courtyard_io. Tokenizes physical collectibles like Pokémon cards as ERC-721s to streamline secondary markets. > FundingWorks by @token_works. Vests raised ETH through soulbound NFTs and issues patron NFTs burnable to reclaim unspent funds. > @RAILGUN_Project. Deploys ERC-6551 NFT smart wallets inside shielded addresses for private DeFi access. > @thewarren_app. Supports deploying multi-file websites onchain with no external dependencies via NFT containers. > zOrg by @z0r0zzz. Mints membership badges as NFTs for its onchain DeFi corporation. For these protocols, the NFT is simply the most elegant tool for the job—an ownable, programmable handle for unique state. As Brody suggested, this quiet proliferation across finance and business will only accelerate, even as the NFT label fades into infrastructure. Image

The Ethereum L2 Squeeze Much will come in the wake of Vitalik's declaration that "the original vision of L2s... no longer makes sense." With the L1 scaling and blockspace now an abundant commodity, L2s and Alt-EVM chains must differentiate or get squeezed out 👇 ~~ Analysis by @kenzimerkle ~~ Last month, Vitalik Buterin sparked debate with a blog post stating that "the original vision of L2s and their role in Ethereum no longer makes sense." Days later, he sharpened the message: > "If you make an EVM chain without an optimistic bridge to Ethereum (aka an alt L1), that's even worse. We don't friggin need more copypasta EVM chains, and we definitely don't need even more L1s. L1 is scaling and is going to bring lots of EVM blockspace." As L1 scales, it becomes cheaper and more capable. Mainnet is becoming the blockchain that L2s aspired to become, without the complexities of L2 designs. This forces L2s and Alt-EVM L1s to differentiate or get squeezed out. The Squeeze The pressure has been building for months. L2 and Alt-EVM L1 tokens are down 80-90% from highs, with adoption plateauing once airdrops ended. Last week Base announced it's leaving Optimism's Superchain, taking 97% of the collective's real economic value with it. The rationale: ship faster, reduce dependencies, and keep fees in-house. Beyond recalibration, chains face revenue pressures as the industry matures. Blockspace is no longer scarce. Too many chains compete for users, turning differentiation into a commodity. Meanwhile, revenue-generating chains like Hyperliquid set a new standard, proving sustainable economics matter more than narrative. The "gas fee only" model is breaking down. Chains must find a niche justifying their existence off Mainnet and generate revenue to sustain themselves. How Chains Are Responding While Vitalik's post served as a messaging wakeup call, months of rough metrics had already led EVM L1s and Ethereum L2s to seek deeper differentiation. > Polygon: The Payments Stack. Even before Vitalik's post, Polygon pivoted to become a "revenue-generating blockchain company." In January, 0xPolygon Labs announced $250M in acquisitions: Coinme (payments firm with money transmitter licenses) and Sequence (wallet infrastructure). These anchor the forthcoming "Open Money Stack," a framework for regulated stablecoin payments launching later this year. Stablecoins see the most real-world adoption globally. USDC in Polymarket drives significant Polygon activity. Stablecoin transactions on Polygon outpace all other L2s, gaining speed from these acquisitions, prediction market growth, and the October 2025 Rio upgrade, which overhauled the chain's architecture for payment-specific performance. Polygon hasn't explicitly tied this pivot to POL token value accrual yet, but the strategic direction is clear. > Sonic: Vertical Integration. Alt-EVM L1 @SonicLabs takes a different approach. In their [early February post, Sonic announced it's abandoning the "gas fee only" model entirely. With blockspace commoditized, gas fees no longer sustain chains. Sonic's solution: build and acquire core DeFi products—trading infrastructure, lending, liquidity provision, stablecoins, staking—to operate in-house. Revenue flows directly back to the S token rather than external apps. Base serves as a cautionary tale, highlighting the dangers of relying on external parties to generate chain value. Unlike Polygon, Sonic explicitly addresses token value accrual. Buybacks will only come when real protocol revenue develops from these integrated solutions. The sequencing matters. Optimism announced last month they'd allocate 50% of Superchain revenue to token buybacks, then their primary revenue vehicle left. What Comes Next @VitalikButerin by reiterating that existing L2s and EVM chains can bring new features to the table: Privacy (@aztecnetwork). App-specific efficiency. Ultra-low latency. Expect chains to respond in three ways: > Alt-L1s → Rollups. Some may follow Celo's path from last cycle, converting into rollups and trading sovereignty for tighter Ethereum alignment. > Acquisitions. Well-capitalized chains will pursue acquisitions to accelerate pivots, as Polygon has done and Sonic suggests it will. > Verticalization. More chains will pick a specific category and build infrastructure to own it. We'll likely see buyback talk, but hopefully as a secondary priority. Chains announcing buybacks before making adjustments put the cart before the horse. The market will punish them if they lack revenue to support it. The era of "we do everything" L2s is ending. What replaces it looks like Polygon's payments focus or Sonic's vertical integration: chains that identify their category, build revenue around it, and earn the right to reward holders. It's a step in the right direction, but will cause pain.