168X Talks to Lao Bai: What to Buy in This Bull Run? Robinhood, RWA, Perp DEXs, AI, and the Next Alpha

Summary

• Lao Bai is running more ETH than BTC this cycle: BTC is the higher-quality asset, but ETH/BTC just broke above a long-term downtrend around 0.03 on the weekly chart. His scenarios are $200,000 for BTC in the bull case, $150,000 in the base case and $120,000 in the bear case; if BTC reaches $150,000, ETH should deliver another 50%-60% on top of BTC doubling. But he stresses that Ethereum cannot lead the market—how high it goes depends on BTC.
• He sees RWA, Perps and privacy as this cycle’s core themes. The privacy names are Zcash, Zama and NEAR; in Perps, it is Hyperliquid and LIT, although neither is cheap. RWA lacks an ideal vehicle: UNDO has no token value capture and continues to sell, while he instead sees UNI potentially benefiting from the combination of Robinhood trading and Uniswap. In AI, VVV and Bittensor are worth watching, with a potential entry only after a pullback if the data has not materially deteriorated.
• Robinhood has been his main battleground over the past month, and he bought HOOD in the low $100s. He likes Vlad and the team because they understand both stocks and Crypto, while Robinhood keeps expanding from Cash App-style products and Meme-stock pairings to Uniswap v4 Hook-based products. He once tweeted, when HOOD was around $76, that “the final form of every exchange in the world should be Robinhood,” but did not buy the stock.
• Among L1s, HYPE and BNB are both solid: BNB is Coca-Cola and Nvidia, HYPE has more upside optionality, and Solana has the most beta. Solana’s onchain exits rely mainly on native DEXs such as Raydium, Orca, Meteora and Jupiter. He owns SOL and hopes it reaches $150 within one or two months. Whether it breaks its prior high still depends on BTC.
• Lao Bai sees only two ways for Perp DEXs to challenge Hyperliquid: distribution or institutional-grade trading capabilities. Lighter could differentiate itself if Robinhood uses it as backend infrastructure; Variational is pursuing a better trading experience through RFQ and hedging. In his recollection, roughly 2% of Hyperliquid addresses generated about 80% of early trading volume, led by several hundred to roughly 500-1,000 large North American traders—though that does not mean Asian retail users have no value.
• AI×Crypto is still in its third year of “looking for a nail with a hammer,” and trading agents have not yet earned his trust with capital; he sees VVV’s privacy inference as the clearest real need. OBO’s design—allocating OpenRouter inference credits to token holders and building a secondary market for idle compute—is clever, but its flywheel depends on Meme-token trading fees. He sold at $0.04 and missed the subsequent rally, and remains unconvinced by the model.
• FOMO’s real edge is not buying Memes with Apple Pay—Moonshot already did that—but linking KOL trading activity to X accounts and compressing the lag from seeing a post to verifying it and placing a trade. FOMO has attracted a large number of KOLs with subsidies. If X continues integrating X Money, token price charts and trading, it could become a more complex version of FOMO. If SmartX receives support from Binance and fills the Social Trading gap, Lao Bai thinks it could potentially receive an airdrop, but that remains conditional.
• He thinks this could be Crypto’s last serious, or last large-scale, bull market. A shrinking primary market, the October 11, 2025 market shock, Hyperliquid’s shift toward revenue and buybacks and burns rather than “dream-multiple” valuations, and AI’s pull on capital and attention have combined to produce the coldest bear market. The more likely future is a long period of sideways trading punctuated by smaller bulls and bears: Crypto is increasingly resembling US equities, while US equities are increasingly resembling Crypto, rather than following a clear four-year cycle.

Deep dive

1. Fish Where the Fish Are: The Main Battleground Shifts to Robinhood, with More ETH than BTC

• Lao Bai opened with a clear framing: “Fish where the fish are.” Over the past month, there have been more fish on the Robinhood side, so he has been spending more of his time there.
• Among the major coins, he considers BTC the higher-quality asset, but his personal positioning this cycle will favor ETH over BTC. On the weekly chart, ETH/BTC has just broken above a long-term downtrend around 0.03; he thinks “whether or not you understand candlesticks, it should be obvious at a glance.” If the bull market continues for several months, the pair should not quickly fall back below the breakout.
• The host added Tom Lee’s BitMine, 易总 and the line about “always finding it impossible not to be bullish on the industry,” linking them to the path dependence of diamond hands since 2017 and 2018. This was the host’s industry commentary, not a complete statement of Lao Bai’s views on those people.

2. BTC Scenarios: $200,000 Bull, $150,000 Base, $120,000 Bear

• Asked whether ETH could reach $8,000-$10,000, Lao Bai said he does not know and can only take it one step at a time. ETH cannot lead the market; how high it goes depends entirely on BTC.
• If BTC reaches $150,000 and ETH has broken its long-term downtrend relative to BTC, he thinks ETH can rise another 50%-60% on top of BTC doubling.
• The host noted that $150,000 would be only about 25% above the prior high. Lao Bai agreed that $200,000 is his bullish scenario and laid out three cases: $200,000 bull, $150,000 base and $120,000 bear.

3. The Low-Hanging Fruit Is Gone: This Cycle’s PMF Is Mainly RWA, Tokenized Equities and Privacy

• Lao Bai said most past innovation was pushed outward from technology: whenever there was a technical breakthrough, the market was willing to assign a project a “dream valuation” without first requiring proof of data, users or product-market fit.
• Most technical routes and sectors have now been tried, including gaming, privacy, ZK, Layer 2, finance, sports and cards. He estimates that 80%-90% of them have already been disproven.
• He thinks the biggest areas to have found real PMF this cycle are RWA and privacy. Perps look more like a continuation from dYdX to GMX, with Hyperliquid extending the category through its trading experience and HIP-3. The genuinely newer direction is Robinhood’s tokenized stocks—the combination of Memes and tokenized equities.
• Privacy, in his view, may only qualify as “half a sector”: institutions will need some privacy once they move onchain, but the sector’s early gains this cycle were so strong that many participants appear attracted more by price appreciation than by conviction in the thesis.
• The host cited a post by 张胖胖: privacy is perhaps 20% fundamentals and 80% sentiment; Uniswap is 50/50; Morpho is 90% fundamentals and only 10% sentiment. They used the comparison to discuss how rallies often reflect sentiment running ahead of fundamentals.

4. L1 Ranking: BNB Is Coca-Cola, HYPE Has Optionality, and Solana Has the Most Beta

• HYPE and BNB are both highly stable. Lao Bai compares BNB to Coca-Cola in beverages and Nvidia in AI: high certainty, but perhaps less room for another leg higher. HYPE is already expensive at a $100B valuation, but if it enables permissionless trading of everything, continues expanding through HIP-3 and delivers good HIP-4 data, its upside could exceed BNB’s.
• Solana may have the most beta of the three. Lao Bai recalled its price rising from $2 to $250, falling back to $8, then recovering to more than $200, before dropping to the $40-$50 range and returning to above $100.
• Solana’s biggest advantage and disadvantage are the same thing: after FTX collapsed, there were few Solana assets that could get listed on exchanges such as Binance, so exits depended mainly on native DEXs including Raydium, Orca, Meteora and Jupiter. That makes the ecosystem more Crypto Native, cyberpunk and grassroots.
• Lao Bai holds some SOL and hopes it reaches $150 within one or two months. Whether it breaks the prior high still depends on BTC: “the big brother has to lead the little brothers.”
• He also thinks Solana has not yet become what it originally set out to be. Perps based on either CLOBs or onchain AMMs have generally been weak; Drift never truly recovered after its incident. The better Perp product on Solana today may instead be GMX-linked GTrade, which is AMM-based rather than CLOB-based.
• In tokenized equities, Solana previously had a clear advantage through xStocks. Today, Solana has the Stonks and Pump.fun launchpads, while Robinhood has Launch and Pounce. Lao Bai sees the two ecosystems as strong in the launch mechanics for tokenized stocks and Memes.

5. The Robinhood Thesis: Buy Vlad and the Team; HOOD Was Bought in the Low $100s

• Lao Bai likes Robinhood primarily because he likes the company and Vlad. He sees the team as one of the few that understands both stocks and Crypto-native culture. The GameStop short squeeze cited by the host was another example of that combination.
• Robinhood started with Cash App-style products, first using Memes to capture attention, then pairing Memes with stocks, before expanding into stock brokerage, AI and Nvidia paired assets, and Uniswap v4 Hook-based products such as liquidity management and launchpad plays through Hook and Delta.
• Lao Bai bought HOOD two or three weeks ago, when it was already trading in the low $100s. He had previously tweeted, when HOOD was around $76, that “the final form of every exchange in the world should be Robinhood,” but did not buy it at the time and now regrets it.
• His “Everything Exchange” includes traditional stocks, traditional Crypto spot markets, Perps and prediction markets. Perps were directly connected to Hyperliquid, while the prediction-market connection was, he recalled, Kalshi. Add options and the product is essentially an Everything Exchange.

6. Tokenized-Stock Exchanges Have No Clear Moat: Everyone Could Plug Into Alpaca on the Backend

• Asked what the ultimate competitive advantage would be once exchanges such as Bitget, Gate and Binance all offer tokenized stocks, Lao Bai said it would mainly come down to user experience and regulatory credentials.
• Early movers can capture users who want to try the product first, but the backend for tokenized stocks could in principle all connect to Alpaca, with a broker buying the underlying shares on Nasdaq or the New York Stock Exchange. That may not create a true difference in liquidity or depth.
• Once everyone has launched, the experience should be broadly similar, with the main differences in frontend UI/UX and fees. Lao Bai might still choose Binance or OKX because he wants to trade Crypto and stocks in one place, while Binance currently has an advantage in token variety and trading depth.
• If a user only trades tokenized stocks, Crypto depth can be ignored in favor of a platform with lower fees or one the user considers safer.

7. How to Research the Robinhood Ecosystem: Study 5-10 Tokens and Separate Data from Narrative

• Newcomers should first study the leading assets on Pump.fun and Launch, then understand the trading patterns, narratives and actual businesses of 3-5 leaders. Lao Bai thinks researching 5-10 tokens is enough to understand the Robinhood ecosystem.
• Data-driven investors can examine burn volumes, TVL and trading-volume dashboards for projects such as AI and Pawns to estimate their value.
• Shroom, the new Delta, Hook, Orbil and CDZ are much harder to value and depend more on attention and narrative. CDZ doubled after the GMX founder confirmed that he had indeed launched the token, but Lao Bai does not think it had necessarily built much actual product at that point.
• Lao Bai likes Stonks Broker and finds its options product interesting, but the product has been delayed and attention has faded, leaving him stuck in the position.
• On price action driven by celebrity attention, Lao Bai says investors must act immediately. By the time a KOL they follow posts about it on Twitter and they rush in, they are likely already providing exit liquidity for someone else.

8. AI Agents: Trading Still Cannot Be Trusted with Capital; VVV Targets Private Inference; OBO Is Clever but Not Yet Convincing

• Lao Bai has personally tested AI trading platforms. After a user deposits several hundred dollars and sets a strategy, the AI may misunderstand the intent, encounter a bug or fail to execute the framework perfectly. He might accept 500-2,000 U, but would not feel comfortable entrusting it with more than $10,000.
• Mojo AI, Travis and SmartX are still early; some have not reached mainnet. Lao Bai thinks the narrative is logically sound, but whether the idea can be engineered into a working product is a separate question.
• In his view, the better AI×Crypto combinations today are not trading products. VVV, or Venice AI, focuses on decentralized private inference: a privacy version of OpenRouter with open-source models, GLM and uncensored models.
• Demand includes creating restricted content, generating adult images, and asking private or dark questions that users do not want OpenAI or Anthropic to know. Lao Bai says he currently cannot identify a second use case with equally clear must-have demand.
• On value capture, he estimates that 70%-80% may accrue to the equity side and 20%-30% to the Token side, while explicitly noting that he has not done particularly detailed research.
• The host told an anecdote about a Goldman banker in the US who asked OpenAI how to murder his girlfriend, triggered an alert and was referred to law enforcement, then joked that he might have used Venice AI if he had known about it. This was the host’s example, not an independent verification of the story’s authenticity.
• OBO changed the content allocated to Vault token holders from tokenized stocks to OpenRouter inference credits, while also creating a secondary market for idle OpenRouter compute. Lao Bai thinks the combination is clever, but he sold at $0.04 and missed the rally. OBO’s flywheel depends on token trading fees, which are used to buy credits and distribute them to holders; it therefore still relies on a Meme-token buy wall and leaves open the question of “why not just go directly to OpenRouter?”

9. Token Middlemen: Legitimate Discounts and Gray/Black Markets

• The legitimate model relies on channel access to obtain official credits at a discount. Lao Bai said OpenAI may be available at roughly 66% of list price and Claude at roughly 70%, then resold slightly below OpenRouter prices. The core advantages are the channel discount and the amount of supply.
• Gray and black-market models may involve stolen credit cards or hiring university students to claim free educational credits.
• The host said Sun Yuchen’s B2B API appeared to have roughly 100 trillion tokens of request volume, while explicitly stating that he was not sure who was using it.
• Based on his understanding of Sun Yuchen’s style, Lao Bai thinks the credits “should” contain a substantial amount of gray- and black-market supply, but he would not say that all of it does, nor that 100% comes from unofficial channels.

10. Three Years of AI×Crypto Looking for a Nail with a Hammer: Circle and L1s May Capture Payment Value

• Lao Bai said he has reviewed more than 100 AI startups and AI×Crypto teams since late 2023, including distributed inference, Game Theory and ZK verification. Many were cases of “looking for a nail with a hammer.” After nearly three years, there is still no perfect point of integration.
• A meme he posted two years ago still holds: Crypto is enthusiastic about AI because it needs AI to help with its narratives, while AI looks at Crypto and says, “Who are you? I don’t know you.”
• The host argued that Crypto already has payment rails for AI through USDC, Solana and Ethereum. Lao Bai agreed that if an Agent Economy emerges at scale, value may be captured by Circle, Solana, Ethereum and other parts of the supply chain and infrastructure, rather than by one standalone project.
• x402 is only one option. There could eventually be 5-10 parallel protocols; Google’s ACP, Visa and Stripe may each develop their own approaches, with onchain, offchain and hybrid models all coexisting.

11. The Semi-Standardized RWA Niche: USD.AI and Dawn Have Room, but Scale Is Limited

• Lao Bai understands Dawn as an expanded version of USD.AI. USD.AI mainly provides GPU-collateralized lending, while Dawn may extend into memory or Edge Computing revenue, but the overall business model is similar. USD.AI is currently the leader in the category.
• Assets can be divided into standardized, non-standardized and semi-standardized categories. Treasuries, tokenized stocks, gold, crude oil and commodities are easier to put onchain. The non-standardized assets that Centrifuge explored early, such as music-rights revenue, are the hardest.
• GPUs are semi-standardized assets. Common models include H100, A100 and A200, but pricing, recovery periods and accounting depreciation are not fully uniform. They can still support a relatively standardized collateralized-lending framework, with scale and simplicity between standardized and non-standardized assets.
• The business is essentially fixed income built around earning a spread. If banks eventually take on these projects, the protocols’ addressable space could be affected. But the host cited Goldman’s forecast that AI power consumption could be 24 times current levels by 2030, leading Lao Bai to argue that electricity, chips and memory will still be in strong demand. Capturing even 1%-2% of the semi-standardized-asset market could create meaningful scale.

12. FOMO Wins on Latency: From Seeing a Tweet to Copying a Trade in One Push

• The host discussed FOMO’s fundraising and Benchmark, saying total funding may exceed $90M and that the company may still have $30M on its balance sheet, while using qualifiers such as “it seems” and “if I remember correctly.”
• After trying the product, he found FOMO’s smoothest feature to be buying Memes with Apple Pay, without learning about wallets, bridges or Gas. The platform also combines KOL trading activity with social distribution.
• Lao Bai had heard that FOMO offers qualifying KOLs roughly $20,000 and requires them to move from Pump.fun to FOMO and display FOMO trades on their Twitter pages. This was an unconfirmed rumor, which Lao Bai compared with the early subsidy war between Didi and Kuaidi, except that the subsidies now target KOLs.
• The host pointed out that Apple Pay purchases were not invented by FOMO—Moonshot had already done that. FOMO’s real advantage is latency: after a user follows 大宇 and sets an alert, 大宇’s purchase of roughly $200,000 of a token on FOMO can trigger an immediate push notification. Copy trading could be added later.
• X already has X Money and may continue integrating token price charts and trading. Lao Bai thinks X could become a complex version of FOMO, while FOMO is a simplified version of X; both could eventually put calls, charting and order placement on the same platform.

13. Asian Teams Trade at a Discount; Late-Mover Advantage: SmartX Depends on Binance Support

• Lao Bai acknowledged that North American teams often receive valuation and fundraising premiums, while qualifying his point by saying that “a lot of innovation” comes from Europe and the US, not that all innovation does. He thinks Chinese teams are better at taking a product from 1 to 100, as with GMGN, WeChat and Douyin—optimizing and scaling products—whereas 0-to-1 innovation is more common in Europe and the US.
• He sees SmartX as an Asian version of FOMO. Just as the West has Hyperliquid and Asia has Aster, the advantage of Chinese teams is taking an existing product with a poor experience and making it smoother, first acquiring users in Asia or Southeast Asia and then attracting Western KOLs and traders.
• The host said SmartX would likely play a role within Binance and 赵老板’s broader ecosystem. Lao Bai said that if 赵老板 is indeed supporting SmartX in Social Trading, the project could command a high market cap after launching a token, and encouraged listeners to register first. His statement that there “will definitely be an airdrop” was a conditional expectation, not a confirmed fact.

14. The Case for OKX’s $25B Valuation: One-Quarter of the Leader

• Lao Bai said he left OKX at the end of last year. At the time, the company’s two main priorities were its own IPO and X Layer. X Layer was about six months behind schedule and only began formally developing and attracting TVL in the second half of this year.
• He currently sees X Layer potentially working on Agent Trading and tokenized stocks, but does not yet have a clear view of its true focus.
• The host revisited the news that ICE invested in OKX at a $25B valuation. Lao Bai said that if BNB is treated as Binance equity and Binance is valued at roughly $100B, then OKX at one-quarter of that would make $25B fair.
• $25B was roughly Hyperliquid’s market capitalization at the time, but Hyperliquid is an emerging, undisputed leader in its niche, while OKX is the number-two or number-three centralized exchange. Their valuation frameworks are different.

15. Perp DEX Differentiation Has Only Two Paths: The Whale Structure Behind “Asian Users Aren’t Worth Much”

• Lao Bai sees two ways to challenge Hyperliquid: distribution, such as Lighter being used directly as Robinhood’s backend and preventing users from flowing to Hyperliquid; or institutional-grade capabilities, such as Variational using RFQ plus hedging to deliver a better experience and lower slippage than a CLOB.
• Lighter initially focused on ZK and zero fees, without a qualitative difference from Hyperliquid; some users were mainly there to farm volume and airdrops. If Robinhood eventually integrates Lighter, Lao Bai thinks it could resemble Aster attaching itself to Binance and lift its valuation to a new tier.
• The head of a Perp DEX once told the host that Asian users were not very valuable. Lao Bai thinks the comment may have reflected two considerations: perpetual futures were first developed by Western teams such as BitMEX, and early Hyperliquid volume came mainly from large North American traders.
• Lao Bai recalled their onchain analysis: in the six to twelve months before Hyperliquid took off, roughly 2% of addresses may have generated about 80% of trading volume. The core users may have been 500-1,000 large North American traders. This does not mean Asian users have no value; it means retail represented a limited share of Hyperliquid’s volume structure.

16. The HIP-3 Frontend War: Trade Can Win on Depth; Challengers Must Fight on Distribution

• HIP-3 frontend deployers can list Asian stocks, assets from other countries and even pre-IPO tickers. These are essentially synthetic assets that track prices; they do not involve buying the actual spot asset or placing it in an SPV or Treasury.
• But CLOB-based trading feels like a CEX. If Trade already offers a ticker, its market-making depth and liquidity may give it an advantage. In theory, Trade could even hire a trader familiar with Taiwanese, Japanese and Korean stocks and continue expanding its global ticker coverage.
• The host said Entropy has a strong capital base behind it, but he currently sees no decisive advantage beyond subsidies. Asian teams’ familiarity with Asian users and assets could offer a cultural and regional edge, but he is not yet fully convinced.
• Para, or Para…, on HIP-3 is taking a different approach: no frontend, just monitoring market demand and launching tickers quickly. The host thinks its data is ordinary and that tickers are easy to copy—competitors can list the same ticker the next day and overtake it with greater volume and market-making depth.
• The decisive factor will therefore be backend market-making capability. If challengers cannot beat Trade on liquidity, they will have to compete on distribution, acquisition channels and frontend experience, either with substantial capital or by finding an entirely new interaction model resembling FOMO.

17. Onchain Trading Cards Actually Broke Out: The Host Credits IP and Childhood Imprinting

• The host does not play Pokémon cards, but his son does. He noted that the Card project’s revenue briefly exceeded Pump.fun. It lasted only a day, but still showed that onchain cards are not merely a narrative: real collectors and trading flow exist.
• Lao Bai first asked whether the growth came from genuinely moving cards onchain or from top-tier Japanese IP such as Pokémon. The host said the core driver was still Pokémon’s broad consensus.
• The host also argued that onchain trading solves the ownership-transfer and transaction-friction problems of physical cards. In the past, buyers might have had to use platforms such as Xianyu; now they can trade onchain.
• Lao Bai summarized generational preferences as: “young people buy Pokémon, middle-aged people buy Nvidia, old people buy Moutai.” Each generation’s worldview is imprinted by what it encountered in childhood.

18. Anatomy of the Coldest Bear Market: Market Makers Broke, Not Just Retail

• Lao Bai said the dominant model over the past seven or eight years was “VC funds a token, the token lists on an exchange, and market makers control liquidity and distribution.” Outside a few leaders such as Uniswap and Aave, many projects had little connection to fundamentals.
• Around 2023, he counted the VC-backed tokens awaiting listing or unlocks and estimated their potential value at hundreds of billions of dollars. There are now too many assets: “too many wolves and too little meat,” unlike the scarcity of assets during the 2017 ICO cycle.
• Regarding October 11, 2025, he repeated the market claim that market makers may have lost roughly $50B. Token prices could not hold not simply because retail stopped buying, but because market makers could no longer keep providing liquidity.
• Hyperliquid offered an important model by putting revenue, buybacks and burns onchain, allowing the market to value it on P/E or P/S rather than the traditional “dream valuation.”
• The October 11 market shock, Hyperliquid’s demonstration effect and the wealth effect created by AI combined to produce what Lao Bai considers an exceptionally brutal bear market.
• The host proposed three causes: Hyperliquid’s anti-VC narrative and points system, Asian VCs being hurt by DAT, and the possibility that Ethena was the only project from the prior VC-backed cycle that truly made money. Lao Bai considered the summary broadly accurate, but added that Primitive is more like half a Western VC: David has spent years in the US and studied at Stanford, Dovey is in New Zealand, and the team also has partners in New York.

19. The Death of the Primary Market and the VC Endgame: “A $1B Fund Might as Well Roll the Dice”

• Lao Bai contrasted the current environment with 2018-2019, when Western VCs still funded MakerDAO, Uniswap and Compound despite the bear market—projects that later became infrastructure for DeFi Summer.
• After joining ABCDE in 2023, he spoke with 4-5 projects per day on average, and as many as 9 on the busiest days. Over more than two years, he reviewed more than 1,000 projects; only 40 or so ultimately emerged, and perhaps 60%-70% of those could still go to zero.
• After this cycle’s mini bull market, Asian VCs including ABCDE began shutting down. Many other Asian VCs subsequently closed, died or moved into secondary markets. Most second- and third-tier Western VCs also stopped investing; small projects could not raise, and small funds could not raise from LPs.
• Lao Bai expressed sympathy for a16z rather than simply mocking it. Market rumors say it has invested in more than 200 projects, with several dozen already shut down. If a fund has roughly $1B or more and faces pressure from LPs to deploy, it is difficult in Crypto to select only a small number of high-quality projects; the process can even make people feel they might as well roll the dice.
• They also discussed Dragonfly, a16z, Paradigm, Multicoin, Founders Fund and Sequoia. The host speculated that Multicoin may have bought Hyperliquid or invested in Maple’s Treasury, but neither was confirmed. Paradigm was mentioned as recently focusing on prediction markets, while Founders Fund and Sequoia were described by the host as appearing to stage a capital contest in prediction markets.

20. “Maybe Crypto’s Last Serious Bull Market”

• Lao Bai’s view is conditional: if the current cycle is still in its early bull-market phase, this could be Crypto’s last serious or last large-scale bull market. He is not saying there can never be another bull market.
• His reasoning is that most technical routes and PMF categories have already been explored. Privacy and tokenized stocks this cycle look more like incremental innovations on existing directions, rather than events on the scale of the 2017 ICO cycle, 2020 DeFi Summer or 2025 AI Memes.
• If tokenized stocks, privacy and Perp DEXs have all been fully traded, the remaining areas within his imagination may be circular Memes, lending, DEXs and similar categories.
• The market could enter a long period of sideways trading punctuated by smaller bulls and bears. Projects will be judged more on revenue, cash flow, buybacks and burns; Crypto will increasingly resemble US equities, while US equities increasingly resemble Crypto. The original line was that “US equities are increasingly becoming Crypto,” not that they are becoming increasingly Meme-like.
• The host cited the performance of stocks such as SanDisk and SK hynix, arguing that their rallies already look a lot like Crypto rallies.

21. Media’s Next Blue Ocean: Shape Agents’ Minds, from SEO to GEO

• Lao Bai thinks X and Podcasts currently shape human attention. The next direction may be GEO—Generative Engine Optimization—where the goal is to have content cited by large models, not merely indexed by search engines.
• Media outlets and KOLs could become AI KOLs and AI media, producing content primarily for AI or Agents to read and influencing the Agents participating in the Agent Economy. He sees AIXBT as an early version of this model. Projects attempting to build a “lobster Reddit” or “lobster X” failed as the related hype faded.
• He does not expect a mature product necessarily to appear over the next one or two years; the market may need to be watched over a 5-10 year horizon.
• The host cited Cloudflare statistics suggesting that machine traffic had already exceeded human traffic on websites in May or June this year. Lao Bai inferred that more than 90% of future internet traffic could come from Bots and Agents, though this remains a forward-looking view.
• Over the next decade, more businesses may be built for machines and Agents rather than having two Agents listen to a two-hour Podcast. For an Agent, that is an extremely inefficient way to transfer information.

22. Closing Ticker List: Specific Names and Levels Across the Three Core Themes

• Lao Bai thinks this cycle’s core themes are already clear: RWA, Perps and privacy. Launchpads such as Flaunch and Pump will ultimately launch assets around those themes. He did not elaborate further on BTC and ETH.
• The three privacy names are Zcash, Zama and NEAR, described respectively as “privacy Bitcoin,” “privacy Ethereum” and “what Ethereum should have been.” Lao Bai noted that these are popular formulations on X today.
• In Perps, the names are Hyperliquid and LIT. Neither is cheap: Hyperliquid is approaching $100B. Lao Bai thinks upside could be difficult if HYPE rises from $100, but would consider buying around $70-$80.
• RWA lacks a particularly ideal vehicle. Lao Bai mentioned UNDO as a strong tokenized-stock asset, but it has no token value capture and continues to sell, making it a difficult hold. He instead thinks UNI could benefit because much of Robinhood’s trading runs through Uniswap, allowing UNI to capture value from the combination of tokenized stocks and AMMs.
• In AI, the names to watch are VVV and Bittensor. Bittensor is the leader, but Lao Bai thinks its narrative is stronger than its data. VVV’s data currently looks good, but the token has already risen several times over; it belongs on a watchlist, with a potential entry only after a pullback if the data has not materially deteriorated.