Every part of onchain inference is starting to accrue value to a different set of projects Supply → Chutes Routing → Surplus, UsePod Tokenized access → Venice Agent demand → OpenServ I’m using the leaders as anchors, then building a watchlist of smaller names around each market. 1/ Inference Supply @chutes_ai sets the benchmark. 6.12B requests and 35.8T+ input tokens in its released dataset. Banger names: @engyai verified inference @TargonCompute confidential compute @dphnAI - $POD consumer GPU inference @CestusNetwork early distributed inference The shift here is from renting GPUs → selling served intelligence. 2/ Routing & Inference Markets OpenRouter proved multi-provider routing can scale. Now crypto is opening the supply side. @AskSurplus - base:0xc52aedec3374422d7510e294cfaa90799595cba3 seller-priced inference; 2.5M requests / 107B input tokens by July @UsePodAI GPU + API-key supply, USDC settlement @miniroutersh 1.2B tokens / 60.4K requests / 239 models in 30d @openservai - $SERV agent routing + execution layer More suppliers → better routing → tighter pricing. 3/ Tokenized Inference @AskVenice - $VVV is the reference point. $DIEM turns inference access into a transferable asset, while Venice recently reported 250B tokens/day. Smaller experiments: @orbiodotso - robinhood:0xaa07a0e9209e16ac99708c3ec70159c6ef3128a3 → live OpenRouter credit market; ~$35-37K asks, 22.5% best discount @usedotai - $DOT IRM + stake-for-inference @AileLabs - solana:8d3bQS9vNy3BfqK9dX95JRoaTbWoiuL2hzmSFLDopump idle Claude / Codex / Gemini access @Cerebro_RH - robinhood:0xeb24a2663af4ee979dcbfe39cb95b5f12d369c4b agents + inference credits @manyways_rh - robinhood:0xa26992c4268a8a78a4d872fe4bdad2ed03ac287d early model-access market This is where inference starts moving from usage → credits → transferable inventory. 4/ Agent-native Demand Agents may become the biggest native buyers of inference. @openservai - $SERV agent infra @AskSurplus x402 inference @AileLabs - solana:8d3bQS9vNy3BfqK9dX95JRoaTbWoiuL2hzmSFLDopump x402 settlement @Cerebro_RH - robinhood:0xeb24a2663af4ee979dcbfe39cb95b5f12d369c4b agents consuming credits @BlockRunAI USDC per LLM call @dgrid_ai pay-per-inference Humans pick brands. Agents can route around price, quality, latency, privacy and availability. ➟ Onchain Inference Map Supply Chutes · Engy · Targon · Dolphin · Morpheus · Cestus Routing / Markets Surplus · UsePod · MiniRouter · OpenServ Tokenized Inference Venice · Orbio · DOT · Aile · Cerebro · Manyways Agent Demand OpenServ · Surplus · Aile · Cerebro · BlockRun · DGrid My approach: keep the leaders as core exposure, then selectively build positions in smaller names as usage and liquidity show up. The leaders already own the attention. Now I’m looking for which smaller names capture the next leg.

After my previous options post I went through the replies and realized I missed quite a few smaller venues. So here’s an update. Based on Onchain Greeks, 30D premium volume is currently concentrated around: @DeriveXYZ ~$77M @paradex ~$2.7M @aevoxyz ~$1.9M @ryskfinance ~$1.7M @SynapseProtocol ~$353K @hypersurfaceX ~$94K There are also a few other names worth tracking: @CallPutApp @HegicOptions @Panoptic_xyz Panoptic is especially interesting because the model is quite different from traditional orderbook/RFQ venues, using Uniswap liquidity to create a more continuous options market. I’m also watching @convallax for its options on prediction markets angle, but I’m leaving it out of the data comparison for now since there still isn’t enough public trading data. The bigger takeaway is that onchain options are still small and fragmented, but the design space is getting much broader: - CLOB/RFQ, - structured products - LP-based options - perpetual options - prediction-market options. If I’m still missing any live venue with real volume, drop it below. I’ll keep updating the list.

Onchain options projects spotlight So instead of only watching the leader I’m mapping the long tail: 1/ Liquidity leaders @deriveXYZ - $DRV current onchain options benchmark @aevoxyz - ethereum:0xb528edbef013aff855ac3c50b381f253af13b997 established derivatives venue with live options market 2/ Pure options plays @HegicOptions - $HEGIC peer-to-pool options @stryke_xyz - $SYK concentrated-liquidity options @KyanExchange - $PREMIA permissionless options + RFQ @divergencedefi - $DIVER composable volatility/options AMM @DeriProtocol - $DERI everlasting options + derivatives 3/ Structured / strategy layer @ThetanutsFi - $NUTS altcoin options + structured strategies @TypusFinance - $TYPUS Sui options + structured products @IthacaProtocol - $ITHACA auction-based options execution 4/ Smaller / legacy names @AuctusOptions - $AUC tokenized American-style options @BaseVolApp - $ATM 0DTE / automated options strategies @epicentral_ - Solana options venue @exotic_markets_ - $EXO Solana structured options The market is basically splitting around: liquidity → better UX → structured strategies → new assets Derive has already proved there is flow. Now I’m watching which smaller names can actually turn that flow into OI, premium volume and revenue.

Perps already had their breakout. Options are still early, but the numbers are starting to move @deriveXYZ - $DRV printed its biggest week at ~$1.2B notional and ~10x YTD growth. That gap is what I’m adding to my watchlist > @deriveXYZ - $DRV onchain options leader, strongest liquidity + revenue proof > @HegicOptions - $HEGIC peer-to-pool options + one-click strategies > @ThetanutsFi - $NUTS modular altcoin options infra > @PremiaFinance - $PREMIA permissionless options liquidity + RFQ > @TypusFinance - $TYPUS Sui options, hourly-to-monthly expiries > @IthacaProtocol - $ITHACA auction-based options + capital-efficient collateral > @divergencedefi - $DIVER composable volatility/options AMM > @Rain__Protocol - $RAIN prediction-market infra adjacent to the options trade The real bottleneck isn’t whether options work. It’s liquidity + UX + distribution. Derive already has the liquidity moat. The opportunity now is finding which smaller protocols can abstract the complexity and bring the next wave of traders onchain.

Solana’s AI stack is getting its consumer layer. @ZetaChain - $ZETA is moving toward the application layer for AI on Solana, starting with Anuma. @AnumaAI is already live with: - 300K+ users - 35 AI models - 1M+ requests Anuma adds the consumer side: private AI + user-owned memory + one identity across models Its Private Memory Layer has been running in production since February. Users control access to their memory, Anuma passes only the context each request needs, and private mode uses open models on zero-retention infrastructure. The plan is to make $ZETA the common access token for more AI apps and AI agents built on the same layer.

➟ @solana already has a $1B+ lending market. But the next competition may be about the shape of the loan: its rate, duration and risk. A few projects are already moving in this direction: - @JupiterExchange Offerbook lets borrowers and lenders agree on a fixed rate + fixed duration from 1-30 days. Once the loan starts, the terms stay the same until maturity. - @xitadel_fi takes the same idea into project financing. Its first LTT let Flash Trade raise 334K USDC for 90 days against treasury assets instead of selling them. - @ExponentFinance goes one layer further: instead of only borrowing at a rate, users can trade the rate itself through PT/YT and rate markets. - @kamino is also moving toward fixed-rate lending, bringing the model closer to an existing credit base that already has ~$1.03B in active loans. - Then @perena, @solsticefi and @hylo_so are experimenting with another part of the stack: splitting yield and risk into different products, rather than giving every depositor the same exposure. These projects are not doing the same thing. But they point to the same shift: Rate -> Term -> Risk are becoming things users can choose. That matters because a real credit market needs more than a big lending pool. It needs borrowers who want different maturities, lenders willing to price those terms, and enough liquidity to move between them. Solana has already proven demand for borrowing. The next layer is turning that demand into a real market for term, rates and structured risk. DeFi credit then stops being only an APY game. It becomes a market for how capital is priced and structured.

You don’t need to sell your coin to make money. Its price swings can still generate yield. That’s basically volatility yield a new market starting to form onchain. You give someone the right to buy or sell your asset at a fixed price, and in return you receive a premium upfront. Premium = the fee they pay you for taking that obligation. I’ll break down this market in this post. Right now, it can be split into a few main groups: ❱ Yield products @ryskfinance @ThetanutsFi Focused on holders who want to earn extra premium from assets they already own. Rysk alone currently has around ~$46.7M TVL, ~$150–165M in 30d notional, and ~$1.4–1.6M in premiums generated. ❱ Options markets @derivexyz @aevoxyz @KyanExchange @strykefi Where traders directly buy and sell options and trade exposure to volatility. ❱ New primitives @Panoptic_xyz @GammaSwapLabs @ithacaprotocol @PolynomialFi New models for trading, packaging, or building strategies around volatility. This market is still early. But beyond hold/stake/lend, ct now have another layer to earn yield from the volatility of their assets. More ways to print money onchain.

Robinhood Chain already has the infrastructure to turn stocks into collateral. The 8 largest Stock Tokens I checked have around ~$75.7M in circulating supply. Meanwhile, @Morpho on @RobinhoodCrypto Chain already has ~$1.01B in deposits and ~$447.7M in loans. But NVDA, AAPL, SPY, GOOGL, SPCX, and TSLA combined only account for around $14.8K of collateral inside Morpho Blue. Credit rails are already here before Stock Token credit has really started to scale. ❱ The stack is gradually taking shape: @Uniswap + @0xProject + @1inch handle liquidity/routing @chainlink provides the pricing layer @Morpho provides the base credit infrastructure @longbowlend brings Stock Tokens into isolated lending and leverage @Lighter_xyz and @arcus_xyz open up margin/perps ❱ These projects are also doing same work: @0xHedgehood weekend stock borrow/short @lightlending lender-set credit markets with dedicated LTVs for SPY, NVDA, TSLA... @PareStocks splits price exposure and dividends into PT/YT, then brings PT back into collateral @TermMaxFi fixed-term borrowing + options on SPY/QQQ/NVDA @BitGo + Robinhood issuer/AP layer → custody + primary inventory underneath the whole stack ❱ There are still a few clear gaps waiting to be filled. The market still needs better inventory support when primary mint/burn closes, better risk standards for Stock Token collateral, and reliable liquidation infrastructure when the underlying market is offline. Stock-backed stablecoin credit is still tiny, while options, hard borrow, and structured products still have no clear leader. The next phase needs to focus on quality before scaling quantity. If Stock Token supply keeps growing, growth could move in this direction: liquidity → collateral standards → stock-backed credit → leverage → structured products. Stock Token credit is still very small. But the infrastructure is being built before the capital arrives. Leader will take all, then copycats follow later.

Security is moving up the stack. Ethereum is subsidizing security for mainnet builders through the Ethereum Security Subsidy Program. But some of the selected teams are working beyond smart contract audits. They are pushing security closer to the user. ❱ Privacy @theinterfold private computation with FHE, ZK and MPC @opaquecash private payments @xochi_fi private trading and intent execution @axol_io ZK and privacy infrastructure ❱ Safer transactions @Kai_Sign clear signing, helping wallets show users what they are actually signing @glossifi security intelligence and monitoring ❱ Wallet UX @ambire smart accounts and account abstraction That moves Ethereum security from the contract layer closer to the actual user. Privacy + signing + smart wallets + monitoring are becoming part of the same security stack. I’m mapping the other sectors being backed through the program too.

The significance of @pendle_fi only continues to compound as newer narratives + asset classes emerge onchain. Pendle has always been more than simply a yield layer. It has increasingly serves as an avenue for value discovery across entire sectors, allowing the market to price + trade the future yield generated by whatever assets happen to matter at the time. Narratives rotate, but each new yield-bearing primitive simply expands Pendle's addressable surface area. That's probably one of the strongest characteristics of its PMF. Pendle doesn't need to predict which sector wins next. As long as increasingly more assets move onchain + generate yield, there will be demand to price, trade + hedge those future cash flows. “As long as DeFi wins, Pendle wins." $PENDLE

Aave V4 reached $1B in deposits for the first time. Here are the key onchain drivers behind that growth. 🧵,

. $DEPEG is now ~10x from when I first started tracking it. It was the first token to graduate on @liftdotfun, and the Arc Network pad has kept adding pieces around it since then. Lift now has: > GMGN + Based Bot integration > instant liquidity with no bonding/migration > buybacks from day one > a mystery drop for the first 500 Arc wallets > stock/gold pairing planned once those assets land on Arc If Lift can move from meme launches into stock, gold and other asset-paired markets, the product becomes broader than a normal meme pad. $DEPEG gave Lift its first runner. Now the real test is whether @liftdotfun can turn that early attention into repeat launches and trading flow once Arc opens up. Early will win

➟ @arc might make meme launchpads compete on trust structure That’s why @liftdotfun caught my attention. Arc settles around USDC, so liquidity flows are easier to trace. Lift adds another layer by reducing what the launchpad itself can control after launch. - LP stays outside team custody - supply can’t be expanded later - launch settings stay fixed - 70% of fees go back to the creator There are already plenty of Arc launchpads fighting to be first. I think the more important question is which one still has users once the novelty fades. Lift’s bet is pretty clear: keep the structure simple, reduce trust assumptions, and let the market decide who survives.

RWA perp volume is more concentrated than I expected. Just 4 venues have handled roughly 86% of total RWA perpetual volume in 2026: - Binance $1.59T - Hyperliquid $542.8B - OKX $345.1B - Bitget $238.2B @HyperliquidX alone accounts for around 17.2% market share. There’s also a strong second layer of venues building meaningful flow: @edgeX_exchange @Lighter_xyz @Ostium @extendedapp @Aster_DEX @grvt_io @OndoPerps @OfficialApeXdex @verantaxyz These venues are expanding RWA perp activity across equities, commodities, FX, and indices. The market is getting deeper fast, and the real competition is starting to move toward liquidity, execution quality, and repeat trading flow.

I think the way @paretocredit is scaling the FalconX credit book shows a pretty clear playbook for onchain credit assets. Instead of creating a new product for every market, they are opening more ways to access and use the same exposure. > AA_FalconXUSDC LP is now expanding in 3 directions: @sygnumofficial opens access to institutional and private wealth capital through regulated rails. @NestCredit bring the same exposure to crypto-native users while adding more liquidity for RWA index products on @plumenetwork. @monad + @megaeth open up cross-chain access, paving the way for the asset to be used as collateral in local money markets. One credit book, more capital sources, more distribution, more utility.

➟ @arc’s launchpad war has already started. Different teams are competing on launch mechanics, liquidity and distribution: @liftdotfun - no bonding phase, direct liquidity and day-one buybacks @arcpad_meme - direct Uniswap V3 launches @minarafun - USDC-native launches @Archemistdotfun - X-native launch bot @actfunxyz - bonding curve + marketplace @circlewarp - cross-chain buys via CCTP @ubidotfun - Uniswap V4 + fee sharing @Fliptfun - Stay bonded, earn your share $DEPEG is Lift’s first graduate on Arc. Arc already has several different launch models before the ecosystem really starts competing for volume. Now it comes down to which model can attract creators, traders and liquidity consistently

. $DEPEG is the first token to graduate on @liftdotfun on Arc Network. Lift is a token launchpad on Arc, and $DEPEG is the first live token to go through its graduation process. So for me, this is a simple first case to follow: - how $DEPEG performs after graduation - how Lift develops with the next launches - whether more activity starts forming around the pad Still very early. I’ll keep tracking the numbers from here.