
APAC BD @AlloraNetwork @RoboNetHQ | ex @chainlink 👀 Views are my own
Joined November 2021
- Tweets597
- Following277
- Followers461
- Likes303
Difeng 2.0 retweeted
Stargazer joins the @AlloraNetwork ecosystem
Every trade is two questions. What happens next, and who is making it happen
Allora answers the first with predictions from a network of competing models
Stargazer answers the second by reading clusters behind every token on 7 chains
Soon you'll get both answers in one place
我们宣布了一篇名为《Flawed in Nature, Perfect through Evolution》(天生有缺陷,通过进化变得完美)的新论文,发表在 Allora Decentralized Intelligence(ADI)上。
论文提出了一种新的 AI 设计原则:通过给单个 AI 模型故意引入“基因突变”(随机改变模型参数,使其个体表现变差),反而能让整个模型群体在环境变化时表现更好。
主要观点:现有 AI 系统最大的局限是无法适应训练时没见过的新环境变化。
论文从数学上证明,借鉴自然界进化原理可以解决这个问题。
就像生物进化一样:大多数突变对个体有害,但种群因为保持了多样性,当环境突然改变时,总有一些变异体更适应新环境。
真正的智能(适应性、创造力、原创性)需要在从未遇到过的情况下也能表现良好,而当前的优化型 AI 做不到这一点,进化中的模型群体可以。
Volatility Topics are good.
If you have creative ideas about using Volatility Topics, let us talk.
Allora mainnet performance, for last week, now including Volatility Topics!
Price and log-return direction: NEAR/USD 8h led at 55.3%. The 8h topics averaged 50.1% across the week, up from 49.3%.
Direction of volatility change: SOL, XRP and ETH at 15m all cleared 67%.
1
209
Allora’s price + volatility topics unlock a powerful new use case for options protocols.
Options need more than direction. They need an estimate of how much the market will move.
Price forecasts → directional signal for calls/puts and strike selection
Volatility forecasts → forward-looking vol input for option pricing models like Black-Scholes / SVI
Together, protocols can move beyond relying purely on historical volatility or indirectly derived IV, and start incorporating AI-predicted realized volatility into pricing and vault strategies.
Price predicts where. Volatility predicts how far.
Introducing Volatility Topics.
Until now, Forge topics have focused on price and log-return prediction. Four new topics are now forecasting 15-minute realized volatility for BTC/USD, ETH/USD, XRP/USD, and SOL/USD on mainnet.
Deploy your worker today forge.allora.network/
The "stock-paired meme" narrative isn't a vibe anymore
Two live pools right now:
🟡 MarsCoin on BNB — paired against SPCXB, a tokenized SpaceX stock
• $51.8M mcap
• $2.04M 24h volume
• 6,000+ trades/day
• liquidity turning over ~2.3x every 24h
🟣 FROGE on Solana — paired against tOpenAI, a tokenized OpenAI stock
• ~$730K mcap
• $553K 24h volume — ~76% of its entire mcap traded in a single day
• 3,800+ trades/day
• liquidity turning over ~4x every 24h
Here's the part most people miss:
The stock is the story. The meme is the engine.
Tokenized RWA on its own barely trades. Nobody flips SpaceX 4,000 times a day. Wrap a meme around it and the real-world asset gets dragged through a high-velocity casino. Every degen trade is another lap of volume on rails built for RWA.
That's why flow is concentrating on Solana, Robinhood Chain and BNB. It isn't RWA pulling volume — it's memes borrowing RWA's rails and bringing their own liquidity.
The chain that nails "memes as the volume layer for RWA" wins the next cycle.
98
The "RWA-meme" phase I keep pointing to isn't theory anymore. It's shipping this week — on two chains at once.
On Robinhood Chain, @ponsdotfamily went live launching memecoins paired against tokenized stocks. First token $INITIAL was $AAPL-paired and graduated into a Uniswap v4 pool in ~6 minutes. They've already approved 7 Robinhood stock tokens as quote assets — $NVDA, $SPCX, $GOOGL, $TSLA, $GME, $AAPL, $SPY — plus $USDG.
On BNB Chain, Four.meme + bStocks + @BinanceWallet just opened "Stock Memes" — launch a token against a tokenized US stock like $NVDAb, liquidity seeded from day one, with a dedicated Stock Meme zone in the wallet.
Same idea from two directions: use the memecoin casino as the on-ramp, but make the quote asset a real-world stock. Buyers spend the stock, sellers receive it, the curve collects it, the graduated pool is seeded with it. Every trade walks users one step closer to holding tokenized RWAs — without it ever feeling like "RWA."
This is the meme → RWA-meme → RWA path, built in production. The bridge just went live.
2
352
Robinhood Chain is quietly one of the most active new chains right now, and the sequence it's running is smarter than people give it credit for.
Most new chains beg for liquidity. Robinhood doesn't have to. It already owns the hardest thing in crypto: millions of funded TradFi users inside a trusted app. The moat is distribution, not tech.
The playbook, as I read it:
Step 1 — meme. The cheapest, fastest way to bootstrap on-chain activity and get normies to their first transaction. Low stakes, high virality. Not the product, the on-ramp.
Step 2 — RWA-meme. The clever bridge: real-world assets wrapped in the speed and culture of meme trading. It gets their users comfortable holding tokenized real assets without it ever feeling like "crypto."
Step 3 — RWA for real. Once the habit is there, graduate them into serious tokenized assets: equities, treasuries, credit. Now the TradFi edge compounds — they already have the assets, the license surface, and the users.
The key move: they're NOT going TradFi to RWA directly. Too boring, no liquidity. Meme is the trojan horse that builds on-chain habits first, then walks users up the risk curve into RWA.
If it works, Robinhood becomes the rare TradFi giant that builds a native on-chain ecosystem instead of just listing tokens. A very different, and much bigger, bet.
1
60
Difeng 2.0 retweeted
Allora is extending its decentralized AI Strategy Vaults to @R25official's infrastructure, bringing verifiable, crowdsourced forecasting to institutional-grade derivatives markets.
More on the integration soon.
Difeng 2.0 retweeted
AI agents are showing up everywhere. Precision markets might be next.
We will be joined by @guyukyukgu from @trepa_io on Tomorrow, Friday, July 10, 8am EST to talk about what happens when autonomous agents start playing 60-second precision markets alongside humans.
🔥Ethlabs 正式成立,团队成员都是以太坊生态中长期的核心开发者与生态中最优秀的工程师与研究员,@snzholding 作为贡献者参与其中,我个人也参与到了与 Ethlabs 的沟通,希望站在亚洲社区一员的角度,我们能够更多的链接优秀的亚洲开发者,社区,欢迎对此感兴趣的贡献者一起讨论。
Replying to @ethlabs_org
Ethlabs is supported by a broad coalition across the Ethereum ecosystem:
DeFi builders, core devs, L2 founders, cypherpunks, investors, institutions, and researchers.
People who care deeply about Ethereum’s future, protecting its core properties, and bringing them to the world.
We are coming together to accelerate the next wave of Ethereum adoption by contributing to critical core R&D and building new infrastructure and products where the ecosystem needs them most.
We are incredibly grateful to be supported by @BitMNR, @Sharplink, @ethereumJoseph, @snzholding, and more than 50 others across Ethereum.
Difeng 2.0 retweeted
Mantle Mentor Clinic Session 04 nitter.cf/i/broadcasts/1mxPaaRkv…
Difeng 2.0 retweeted
You had three Mentor Clinics to bring your questions, blockers, and ideas you've been overthinking.
But the fourth is where it ends, and it's your last chance to ask anything you need for The Turing Test Hackathon.
We'll stay on to answer as many questions as we can ↓
Something important happened in May 2026 that most people framed as a price story.
It's actually an infrastructure story. Here's the real Hyperliquid narrative.
Three structural moves in rapid succession:
1. Stablecoin deal — redirects roughly $80M/year of revenue that previously went to Circle and Coinbase, back to HYPE holders. This isn't a tokenomics change. It's a revenue model change that makes Hyperliquid more comparable to a financial exchange than a DeFi protocol.
2. First US spot ETFs — Bitwise and 21Shares launched HYPE ETFs. Grayscale filed (now on third amendment). The institutional access layer is being built.
3. HIP-3 synthetic pre-IPO markets — SpaceX, Anthropic, OpenAI markets now live on Hyperliquid. You can get exposure to private companies that haven't IPO'd yet, onchain, 24/7.
Grayscale's framing: "Hyperliquid is not directly comparable to another project in either crypto or traditional finance. If it continues to execute well, we think Hyperliquid could become a financial services juggernaut."
The $200B daily perp volume market has historically been dominated by Binance and Bybit. Hyperliquid is now 3rd or 4th largest by open interest as a DEX, with self-custody, onchain transparency.
This is what DeFi becoming infrastructure looks like from the inside. Not a whitepaper. Not a narrative. $2.9T in volume processed, $800M in revenue generated, regulation as the key remaining unlock.
The headline is the price. The story is that Hyperliquid stopped being an exchange and started being financial infrastructure.
2
94
happy pizza day!
16 years ago, someone traded 10,000 BTC for two pizzas.
Today, that transaction is remembered as the moment Bitcoin proved it could become real-world money.
From one simple purchase came an entire financial revolution.
Happy Bitcoin Pizza Day 🍕
The future is built one transaction at a time.
Powered by @ATokenxyz
Amplify Any Token, Powered By AI.
2
116
Cerebras just had one of the most dramatic IPO debuts in recent tech history.
+108% at open. $100B valuation at peak. $5.5B raised.
But the more interesting story isn't the stock price. It's what Cerebras represents for the AI infrastructure layer and why this matters for anyone building in AI + crypto.
What Cerebras actually does:Cerebras builds chips specifically designed for AI inference — running frontier models after they've been trained. Their Wafer Scale Engine is the largest chip ever built, designed to handle the kinds of workloads that GPU clusters struggle with: massive context windows, real-time inference at scale.
OpenAI launched its first AI model running on Cerebras chips earlier this year. Anthropic is a customer. Amazon is a customer.
CEO Andrew Feldman put it clearly in an interview today: "If you ask Anthropic, if you ask OpenAI, they have vastly more demand for their offering than they have compute to make it. That is a profoundly different scenario."
The compute shortage isn't hype. It's the constraint that's shaping every AI roadmap right now.
Why this matters for AI + DeFi:
Every serious AI agent running onchain, executing trades, managing vaults, processing inference signals, is ultimately constrained by compute availability and cost.
Decentralized AI inference networks are competing for the same resource that Cerebras is trying to unlock: cheap, fast, scalable inference compute.
When inference gets cheaper and more available through Cerebras, through Render Network, through decentralized GPU markets, the economics of running AI agents onchain change fundamentally.
Lower inference cost = more agents running continuously = more onchain AI activity.
The IPO pipeline signal:Cerebras is being watched as a bellwether for OpenAI and SpaceX listings later this year. If this debut holds, it signals that public markets can price AI infrastructure companies — not just AI software — at growth-stage multiples.
The infrastructure layer is going public. The cycle is maturing.
1
82
Prediction markets are the most underrated AI + DeFi intersection right now.
Here's why I think about them differently than most.
What prediction markets actually are:Not betting sites. Information aggregation mechanisms. When thousands of participants put real capital behind a belief, the resulting price is often more accurate than any expert forecast, poll, or model output.
The 2024 US election was the moment mainstream finance started paying attention. Polymarket called the result hours before traditional media. Not because it's smarter, because it aggregates skin-in-the-game conviction at scale.
Where AI makes prediction markets dramatically more powerful:
AI agents can now participate as autonomous traders in prediction markets. They monitor news, onchain data, and social signals continuously, then take positions when their models identify mispriced probabilities.
Yoni Assia (eToro CEO) described at Consensus Miami this week experimenting with AI agents that "independently open wallets, bridge assets, research trades, and execute transactions across prediction markets." That's not a concept. That's running now.
Where the infrastructure is getting serious:
Their CEO's framing: "Polymarket was built onchain because transparency matters. Trades are open, traceable, and accountable by design."
@xomarket (just raised $6M from Coinbase Ventures, 20VC) is building the "YouTube of prediction markets" — user-generated markets with pooled liquidity vaults. Anyone can become a market maker.
The verifiable layer is being built on top of prediction markets the same way it's being built on top of DeFi: onchain settlement, traceable positions, auditable resolution.
The AI + prediction market feedback loop:
AI generates better signals → more accurate prediction market pricing → better calibration data for AI models → more accurate signals.
This loop is just starting. The infrastructure connecting AI inference to prediction market execution is where the most interesting BD conversations I'm having right now are happening.
2
2
67
Three things happened in the last 48 hours that tell the same story:
— Elliptic raised $120M from Nasdaq Ventures + Deutsche Bank for AI-driven crypto compliance
— BlackRock + Apollo backed Circle's $222M round to bring blockchain to banks
— CLARITY Act just cleared the Senate Banking Committee
The institutional layer isn't coming. It's being built right now.
179
I've been thinking about why information advantage feels different in 2026 than it did in 2022.
In 2022, alpha meant knowing about a project before anyone else. Getting in early. The edge was access.
That edge is mostly gone. Information moves too fast. Crypto Twitter surfaces everything within hours. On-chain data is publicly readable by anyone with the right tools.
The new information advantage is layered differently:
Layer 1: Verification speed. Everyone sees the announcement. Very few people can verify whether it's real within the same hour. Onchain data is public but reading it accurately and quickly is still a skill gap. The teams and individuals who've built that muscle have a structural edge.
Layer 2: Pattern recognition across sources. The signal is rarely in one place. It's in the gap between what a project announces, what their onchain data shows, and what their APAC partners are quietly telling you on calls. No single source gives you the full picture. Connecting them faster than everyone else is the actual edge.
Layer 3: Knowing what to ignore. In 2026, the noise-to-signal ratio is worse than it's ever been. The information advantage isn't finding more information. It's having a clear enough framework to dismiss 90% of what comes across your feed without hesitation.
There's a useful test for any AI + DeFi project that I've started using: remove the AI component. Does the product still work? If yes — it's not an AI project. It's a crypto project with an AI wrapper. Most things fail this test. The few that don't are where the real signal is.
The people winning in Web3 right now aren't necessarily the best connected or the earliest informed.
They're the ones who've built better filters — and the discipline to use them consistently.
1
3
51

















