Crypto is at a low point, and Lao Bai says the old playbook is breaking down
Crypto sits near an industry bottom in August 2026, according to the 168X interview summary. The piece describes a deep bear market, with BitMEX and BitMart announcing the end of their trading businesses, former cycle standouts such as Zapper and Fantasy Top shutting down, and repeated hacks hitting DEX and DeFi projects. At the same time, talent and capital are moving into AI, while primary-market crypto VCs are starting to liquidate.
Against that backdrop, 168X invited Lao Bai, also known as @Wuhuoqiu, to discuss a question that has become harder to ignore: if crypto has in some sense won, why do so many people inside the industry feel like they lost?
Lao Bai has 10 years of network engineering experience. He entered crypto in 2017 because he wanted to buy Bytom, and later worked in research and investment roles at Amber, ABCDE and OKX Ventures. His research now spans AI, semiconductors and the tokenization of traditional finance. Within crypto itself, he said his attention has narrowed to a few sectors he still believes have product-market fit: stablecoins, perpetuals, RWA and prediction markets.
Across the conversation, he made a string of forceful calls. Token issuance, in his view, is debt rather than financing. Crypto VC as a distinct category will disappear. Prediction markets have a much lower ceiling than perpetuals. The era of treating blockchain as a hammer and every problem as a nail is over. Crypto’s two strongest inventions are not a new public chain, but stablecoins and perpetual contracts. And Robinhood, he said, looks like the end-state model for exchanges.
The interview was part of 168X, an interview platform focused on AI, blockchain, robotics, space technology and bioengineering. The hosts were Mr. Z (@168MrZ) and Victor (@vcmktasa).
His investment focus has narrowed sharply
When asked to introduce himself, Lao Bai said he had worked as a researcher at Amber, then as an investment and research partner at ABCDE, before spending several months investing at OKX Ventures.
His current focus inside crypto is much tighter than before. He said he is mainly watching perpetuals, prediction markets and RWA, which he sees as some of the few remaining sectors with real PMF. Outside crypto, a growing share of his time now goes to U.S. equities, especially AI-related names, as well as options, which he said have expanded the ways he can express a view on price and value.
Blockchain was treated like a hammer for too long
Looking back across prior cycles, Lao Bai said many VCs and founders made the same mistake: treating blockchain like a hammer and every opportunity like a nail.
His first-principles view is that once the usual language around decentralization, privacy and censorship resistance is stripped away, blockchain’s core value is simple. It is the first network that made peer-to-peer transfer of value possible. The internet let people move information. Bitcoin, and blockchain more broadly, made it possible to move value rather than just messages. The key technical breakthrough, as he framed it, was the ability to prevent double spending and stop digital value from being copied without limit. He compared blockchain to an upgraded BitTorrent: a peer-to-peer download network turned into a peer-to-peer value network.
He also pointed to Satoshi’s forum posts from roughly 2009 to 2012, saying they suggest an early vision in which Bitcoin could carry more assets on top of it, including invoices and bonded contracts. In his telling, Bitcoin originally gestured toward some of what Ethereum later executed after Satoshi disappeared and Vitalik Buterin took the idea further.
That helps explain why he was especially bullish on Bytom in 2017. Changjia’s idea of bringing real-world assets on-chain fit what he believed blockchain should do. Bytom became the first altcoin he bought. In hindsight, though, he said the project was too early and ended up as a casualty rather than a pioneer.
By the time DeFi Summer arrived in 2020 and late 2021, he said he had also been carried away by fast-rising prices and had lost sight of the original thesis that assets should be the thing moved on-chain. By 2024, he felt something had already gone wrong. In an internal ABCDE meeting that year, he reviewed the tokens he still considered valuable and found that nearly all of them were infrastructure plays that improved programmability, network speed or capacity. He cited the evolution from Uniswap V1 to V2 and V3, from Aave V1 to V2 and V3, and later projects such as Morpho and Pendle.
But then came the harder question: where were the assets?
He said the industry had tried NFT collections, GameFi items, inscriptions, ORDI and now memes, yet none had proved to have durable asset qualities or long-term value. Even gaming assets, one of the early use cases Vitalik talked about in 2017 alongside finance, failed to hold up in practice.
Lao Bai mentioned an article he wrote six months earlier on why World of Warcraft succeeded while GameFi failed. His argument was that anything scalable can be industrialized, and once that happens a structural tradeoff appears. In his view, even if World of Warcraft assets were made freely transferable in the way blockchain advocates imagined, the system would still be overwhelmed by studios and industrial exploitation. That problem is even worse for blockchain games with weak gameplay.
He broke former cycle stars into two rough groups. One includes ideas that were simply too early, such as Bytom and Augur. The other includes products that were overtaken in execution, such as early DEXs like Bancor and Kyber before Uniswap pulled ahead. When Victor noted that Bancor appears to have shut down recently, Lao Bai replied that the market had already treated it as a failed project, so the shutdown itself no longer mattered much.
His current way of judging a project is straightforward: remove the token and see whether the product still stands. Uniswap does. Aave does. But GameFi names such as Axie Infinity and STEPN do not. If a product collapses once the token is removed, he now treats it as evidence that the direction was wrong from the start. The real question, he said, is what problem blockchain solves that traditional finance could not, and whether it genuinely cuts costs or improves efficiency in the real world.
Why token issuance worked before and struggles now
Asked why token launches used to work but no longer do, Lao Bai gave two reasons.
First, crypto was still early enough that markets were willing to pay what he described as a dream multiple. Projects with no users and no revenue could still receive high valuations as long as KOLs and VCs believed the model might work five to 10 years down the line. Retail buyers were not approaching these tokens like stocks. They were often following the judgment of KOLs and major VC firms. If a project raised from names like a16z or Paradigm, retail investors often treated that backing as validation and bought something like an option on a future dream multiple.
Second, there were simply fewer projects then. Hundreds of billions, or even more, could flow into a market where only dozens or a few hundred projects were available. That was a classic case of too much capital chasing too little supply. Over time, more teams entered the market, more VCs were launched, and the cycle shifted into a model built on pitch decks, institutional endorsements, high fully diluted valuations, low float and heavy lockups, with retail left as the final source of liquidity. Once supply passed a threshold and the market flipped from “more meat than monks” to “more monks than meat,” the system stopped working.
On whether Polymarket should issue a token, Lao Bai said the answer depends on perspective. From a retail angle, one can argue that Blur briefly succeeded and OpenSea missed its moment by not launching a token at the right time, perhaps giving up what could have been a $10 billion valuation window. But from the point of view of Polymarket or OpenSea, a business with strong cash generation may have no real need to issue a token because the token is, in essence, a liability.
He said issuing a token does not necessarily maximize a project’s financial or brand value. He also noted that Blur itself does not appear to be thriving now, even if it briefly overtook OpenSea at the time.
As for whether Polymarket could use a token to outgrow Kalshi, he acknowledged that token incentives can boost traffic. Still, he pointed to opinion, which used an airdrop wave from late last year into early this year to push trading volume to nearly Polymarket’s level. After the token launch, he said, both trading volume and token performance almost fell back to zero. His conclusion was that Polymarket has no urgent need to issue a token. It can go either way.
Crypto won in one sense, but old users still feel they lost
Lao Bai argued that two statements can be true at the same time. Crypto has matured, and long-time participants still feel they did not win.
His reason is that crypto’s original ambition has narrowed. The industry once aimed to build a new native financial world. Over time, it became a reformer and auxiliary layer for the existing system instead. He compared the shift to Liangshan being absorbed by the establishment in Water Margin.
He said that if someone in 2018 or 2019 had been told that stablecoins would become a major global dollar infrastructure with a scale in the high hundreds of billions, that Bitcoin would get ETFs and attract pension and institutional capital, that crypto companies would participate in IPO markets, and that RWA would start entering traditional finance through names like Stripe, Visa and BlackRock, they would probably have concluded that crypto had already won.
But from the perspective of asset returns, he said, the industry clearly did not win. It is much harder to imagine another DeFi Summer, or altcoins rising 50x to 100x within a year. Over the last decade, he said, crypto also failed in its original effort to create large, native on-chain asset classes, whether through NFTs, inscriptions or memes. What the industry actually built was an on-chain financial rail. Now it is increasingly being used to bring real-world assets onto those rails.
Even so, he thinks crypto has produced two genuine breakthroughs: stablecoins and perpetual contracts. Those, in his view, are the two most successful crypto-native products with proven PMF.
He agrees that “crypto VC” will likely disappear
When the discussion turned to Dragonfly managing partner Haseeb’s view that “crypto VC” may no longer exist as a category over the next decade, Lao Bai said he broadly agrees.
He does not think VCs or traders should keep treating crypto as a standalone industry. The analogy he used was the internet: nobody now brands themselves as an “internet VC” because the internet became part of the basic commercial stack. He expects crypto to follow the same path and turn into infrastructure embedded inside broader business models.
In that world, a company might use stablecoins or move part of its workflow on-chain, but that may only account for 10% to 20% of the business. It becomes much harder, he said, to define a business as a “crypto project” or a fund as a “crypto VC.” As crypto dissolves into the real economy the way the internet did, those labels should fade.
Perp DEX competition: Hyperliquid leads, but not every challenger looks the same
Lao Bai said he is very bullish on Variational. In his own ranking, Hyperliquid sits alone in the T0 tier. Behind it, he said the names he watches most closely are Aster, Lighter and Variational.
What he likes about Variational is its RFQ structure plus off-exchange hedging, which he said makes a lot of sense. He also praised the team, the founder and the backing from capital and institutional networks.
His broader view is that the market only needs four or five perpetual venues, just as the exchange market itself only needs a small number of dominant players. Hyperliquid sits at T0. Lighter, edgeX and Variational are among the names he would place in the T1 group.
He also raised a more speculative observation from a discussion with a friend. As crypto perpetual trading and centralized exchanges start to resemble leverage-based stock brokerage products in traditional finance, illiquid small and mid-cap stocks could eventually use a playbook similar to Binance Alpha. That would mean controlling 60% to 70% of the spot float, buying continuously to push prices higher, spinning a narrative such as a dying traditional company suddenly pivoting to AI or DAT, then building a large amount of open interest on Hyperliquid or Binance and using the perpetual market as the exit venue for existing shareholders or related parties.
He said the cash market might not show an obvious regulatory violation if the company only buys back stock and never sells in spot. The selling could happen in perpetuals instead. In that sense, he described it as a form of long-term regulatory arbitrage that could give globally illiquid smaller names a new route through Perp DEX infrastructure.
When Mr. Z asked whether later entrants should avoid head-on competition and list Asian equities instead, mentioning companies such as UMC, MediaTek, TSMC, Alibaba, Tencent, Meituan and JD.com, Lao Bai said that direction makes sense for CLOB and market-maker based perpetual venues such as Lighter and edgeX. He recalled that Trasia, a product built by Mable, is moving in that direction. He also mentioned another friend preparing to leave VC to build an on-chain FX perpetual venue, a market many have discussed for years but never fully cracked. If stablecoins on-chain grow into the trillion-dollar range, he said, the TAM for on-chain foreign exchange could be very large.
But he separated that group from products such as Variational and Ondo Perp. In his reading, their structure allows tighter spreads and better slippage on certain assets than Hyperliquid or even Binance. Hyperliquid, he said, effectively relies on a single market maker for liquidity. Ondo behaves more like an RFQ system that can hedge through traditional finance channels and lock risk inside the venue. On some RWA names, especially large-cap stocks like NVIDIA and Micron, he said Ondo Perp shows better spreads and slippage than Hyperliquid and Binance. He added that Variational’s design could theoretically achieve something similar, and that its swap product launched last month already introduced traditional finance hedging into commodities such as oil and gold.
That difference in market structure, he said, gives Variational and Ondo Perp a legitimate path to compete more directly with Hyperliquid.
HIP-3 may become even more winner-take-most
The discussion then moved inside Hyperliquid’s own ecosystem. Mr. Z pointed to HIP-3, where TradeXYZ remains the leading deployer while Paragon has emerged as a fast-moving challenger focused on speed, efficiency and ticker auctions rather than building its own front end.
Lao Bai said he has followed the space, though not in extreme depth. So far, he mainly sees TradeXYZ and Paragon on the Hyperliquid front end. Some of Paragon’s listed assets differ from TradeXYZ’s, but the volume gap remains large. Paragon, by his account, is still operating in the hundreds of thousands range on many assets, while TradeXYZ can regularly run into the hundreds of millions.
He explained HIP-3 as a system where the ticker itself originally represented a deployment opportunity. The first three slots were free, and additional slots were auctioned. What gets launched under that slot is ultimately up to the deployer.
His view is that this setup is likely to reinforce leader effects, and not only because of user awareness. Market making is done by the deployers themselves, not by Hyperliquid at the protocol level. If TradeXYZ can consistently provide deeper liquidity and tighter spreads on the same asset, users naturally move there, creating a self-reinforcing flywheel. That is the challenge new entrants need to solve.
He added that HIP-3 is unlike an AMM ecosystem such as Uniswap, where protocol design changes can still create meaningful differentiation, as Curve once did in stablecoins. HIP-3, he said, is much more of a direct contest over capital, market-making depth and liquidity quality. There is less room to get clever through mechanism design alone.
Security is mostly a time test, and lending is often the weak point
Victor brought up a series of recent incidents in perpetual markets and adjacent sectors. TradeXYZ had a Hynix wick event and later compensated all user losses. Ostium, an RWA project, was hacked for about $18 million. Paradex was also hacked earlier.
Lao Bai said trust around DEX security comes down to two things.
The first is time. Ninety-nine percent of users will never inspect a codebase, so what matters is reputation and survival. He pointed to AAVE as an example of a lending protocol with unusually strong market trust because it has not suffered what he considers a core protocol-level security failure. Even the prior incident people discussed, he said, was related more to issues from the underlying asset layer than to AAVE itself.
He said that in lending, most attacks happen through the lending function itself. Weak assets get manipulated through oracle problems, price spikes or governance control, then used as collateral to extract stablecoins. He cited Mango on Solana and a number of poor-quality lending platforms that suffered repeated attacks. He also recalled that Cream, associated with Machi, had been attacked three or four times.
That leads to his key advice for perpetual platforms. First, do not expand into a broad structure where perpetuals sit alongside collateralized borrowing and lending. He referred to Paradex, or what he thought may have been Drift on Solana, as examples of platforms that tried to do everything at once: spot, collateral, lending and perpetuals. In his view, that is where the opening appears. If you run a perpetual platform, he said, then just run a perpetual platform. Do not add a mechanism where users can post assets as collateral and pull out stablecoins.
Second, there is no shortcut beyond time. If Hyperliquid reaches the four- or five-year mark without a serious security incident while competitors do not, users will naturally trust it more and keep larger balances there, much as Binance built dominant mindshare over time.
Will Hyperliquid launch spot markets?
Lao Bai said he does not expect Hyperliquid to launch native crypto spot in a serious way.
His argument is that spot has become far less important relative to perpetuals. He said 80% to 90% of Binance revenue now comes from perpetuals, and that within perpetual trading, Binance and HTX have seen more than 60% of that business come from U.S. equities, or RWA, since last month rather than from native crypto assets.
If Hyperliquid does add a spot leg, he thinks the more likely route would be spot equities, similar to Binance’s bStock concept, not BTC, ETH or SOL spot pairs. He cannot rule out some kind of “hyperstock” product, though he does not think the odds are high.
Even if it happens, he said the main reason would still be to support perpetual trading through basis trades and internal hedging. He used Ondo as an example, saying part of Ondo Perp’s strong liquidity comes from having Ondo Stocks within the same venue. A market maker can quote the user a $1 million Micron long and take the other side immediately while holding $1 million of Micron spot, all with minimal delay. If Hyperliquid eventually makes a similar move, he said, it would not surprise him. What would surprise him is if it tried to build a full native crypto spot exchange around BTC, ETH and SOL.
Tokenized stocks are the ideal, but perpetuals fit how people actually trade
Lao Bai drew a clear line between the ideal version of tokenized equities and the form users actually want today.
In the ideal case, he said, tokenized stocks would mean real shares, clear legal ownership and free circulation on-chain. He recalled that the U.S. Securities and Exchange Commission recently approved Nasdaq to pursue on-chain stocks, which means the ultimate price competitor for products such as StableStock, and for BIT, with which he said he works, may end up being Nasdaq itself. What Nasdaq may not permit in the short term is fully open stock trading settled directly in stablecoins.
In practice, though, he said most users do not need legal title or actual ownership. They just want a price exposure that tracks the underlying stock. That is why perpetuals remain the dominant format. The numbers he cited earlier fit this view: more than 60% of revenue comes from perpetuals, and within perpetuals, 60% already comes from RWA.
In other words, the ideal is long-term investors holding real tokenized equities. The reality is traders want exposure and then want to add 5x or 10x leverage on top.
Stablecoins are crypto’s most important contribution to the world
On stablecoins, Lao Bai’s view was unequivocal. He said he is “super bullish” and sees stablecoins as the most important invention in crypto outside Bitcoin, or even more important than Bitcoin in terms of changing the real world.
His reasoning is that Bitcoin is now mainly a market for a relatively small group of large holders, institutions and whales. Stablecoins, by contrast, are the technology that actually pushed blockchain into the daily operation of global finance.
He noted that the market was talking last year about using stablecoins to help absorb U.S. debt. Last year, he said, U.S. Treasury debt stood at $36 trillion. It has now reached $39 trillion. Against that backdrop, the hope is that stablecoins can grow to $1 trillion to $2 trillion by 2030 and absorb some meaningful portion of demand for U.S. debt, easing pressure on the government. Over another 10 or 20 years, he said, that market could reach $5 trillion or even $10 trillion.
He sees both business and consumer adoption pushing in the same direction. Stripe and Visa are already entering on the business side, from payments to yield. On the consumer side, stablecoins are becoming a way to buy and sell a wide range of global assets. In his framework, stablecoins and tokenized U.S. equities will work together as a global extension of both dollar hegemony and dollar-asset hegemony.
As for new startups entering the space, he was skeptical. He said only very large players that already control payment rails, monetary gateways or critical financial distribution channels—names like Stripe, BlackRock or Visa—are likely to capture meaningful economics in stablecoins. For ordinary founders, his advice was simple: stay away. This is a business that depends heavily on institutions.
Agent payments and machine commerce look real to him
Lao Bai said the narrative around stablecoins and AI agent payments is real, though not necessarily inseparable from crypto as an industry label.
He broke the argument into two parts. The first is whether an agent economy will happen at all. His answer was yes. He referred to an interview in which Cloudflare’s CEO said that on one day in July, global bot traffic exceeded human traffic for the first time. For a company that sits at one of the world’s largest web gateway and CDN layers, that was, in his view, a major milestone.
The reason, he said, is straightforward. A person researching a camera may open five web pages. A large language model answering the same query may issue 5,000 HTTP requests to compare merchant pages and reviews across the web. Cloudflare’s CEO, he said, expects agent traffic to increase another 1,000x over the next five years.
The second issue is the business model. The old internet economy was built around Google Ads and a loop of search, recommendation and ad clicks. Once agents become the interface, that loop breaks. The most credible answer he sees is micropayments for data access. If an agent uses an e-commerce or review site’s content in thousands of retrievals, the site gets paid a small amount each time.
How that gets paid is where stablecoins matter. Lao Bai said the likely path runs through systems like x402 or stablecoin-based payment rails developed gradually by Cloudflare, Visa and Stripe. He does not think it has to look like traditional card payments, and he does not insist it must be framed as “crypto.” But he does think stablecoins on some chain will be part of the solution. Whether that chain is Solana, Ethereum, Tempo or Arc is still an open question.
Public chains are splitting by use case, and the universal-chain era may be ending
On how stablecoins and agent payments could reshape chain competition, Lao Bai expects a world of clearer specialization.
His shorthand was that Caesar gets what belongs to Caesar, and God gets what belongs to God. TRON and USDT may remain the choice for gray-market activity. Ethereum and Solana, likely with USDC, continue to serve retail users who want DeFi and memes. Institutions, business payments and foreign exchange, meanwhile, may end up on dedicated enterprise chains such as Tempo, Arbitrum and the privacy-focused Canton, which he said has also been developing well.
He does not think public-chain tokens will remain the central beta trade for VCs or retail traders in the next phase.
Hyperliquid, he said, has already shown the new model: a chain should exist to serve a powerful application. Hyperliquid is a chain, but the chain matters because the app matters. It is not a universal-chain project trying to be everything for everyone.
In his view, the universal-chain era effectively ended after Ethereum and Solana, even if the market was reluctant to admit it and kept producing names such as Aptos, Sui, Monad and MegaETH. He pointed to a recent post from MegaETH’s Shuyao saying the MegaMafia incubation plan would stop supporting new ecosystem projects and that the team would instead build applications itself. To him, that was another sign that the age of generic chains is over and application-specific chains are taking over.
He compared that shift to what happened in AI. First, infrastructure mattered most and received the richest valuations. Now the market is paying more attention to applications, PMF, users and cash flow. In his words, crypto is becoming more like U.S. equities, while some parts of U.S. equities—semiconductors and memory, for example—have started to trade with a more crypto-like mix of narrative and volatility.
Large models are starting to resemble public-chain competition
When asked whether large language models are moving in a similar direction, Lao Bai said that process is already underway.
He said China’s so-called six little dragons, including Kimi, DeepSeek and MiniMax, already resemble an earlier public-chain competition where each platform differentiates on technical features, price-performance and even censorship standards. MiniMax’s recent uncensored model was one example he gave.
In his view, the winners in AI will be decided on two fronts. One is business adoption. OpenAI and Anthropic have already built frontline deployment engineering teams and are copying something closer to the Palantir model, because most enterprises still have not deeply integrated AI into internal workflows. Many companies, he said, have simply bought access for employees to OpenAI, Anthropic or Doubao without changing their operating process. AI may improve individual productivity by more than 500%, but enterprise-wide gains may still be under 20% to 30%. That is why those model companies are now helping firms deploy AI directly.
The second front is the consumer side: who creates the next killer app beyond ChatGPT, Codex and Claude.
He also noted that Palantir’s CEO recently said many of the things Anthropic now talks about publicly are things Palantir has long been doing. To Lao Bai, that helps explain why Palantir’s stock has performed well: the market has realized that high-quality SaaS platforms are not so easy to replace and that large models are not easy to plug into enterprise workflows at scale.
Prediction markets have PMF, but their ceiling is much lower than perpetuals
Lao Bai’s stance on prediction markets is more restrained than it was a year ago.
He said he recently wrote a post describing prediction markets in four phases: underrated before the election, validated after the election when Kalshi and Polymarket showed the model works, overrated before the World Cup, and now moving back toward a more rational valuation. He said Polymarket’s volume has been falling since July 18, after the World Cup.
That is why he said he would short Polymarket without hesitation if it were really valued around $20 billion, or if a future token launch pushed it into the $20 billion to $30 billion range. His reason is not that prediction markets have no PMF. It is that he thinks they are nowhere near the scale of perpetuals.
At ABCDE, he said, he had looked at more than 10 prediction-market projects. Many pitches involved scenarios where friends would settle private bets in custom prediction rooms. In reality, he said, that use case shows up far too rarely to be a true PMF. It is a product dream that people attached to the category rather than one validated by behavior.
Most people, he argued, do not have frequent reasons to express a view and stake money on a large number of topics. For most users, the only consistent categories are politics and sports. And in sports, existing betting markets already satisfy much of the demand. In Europe, North America and Australia, people already go to stadiums, sports bars and racetracks, drink beer, watch games and place small bets or buy lottery slips. Kalshi may be doing reasonably well in sports, he said, but that means it is tapping an existing demand pool rather than unlocking a massive new market.
He also recounted a conversation with a founder building prediction-market tools, effectively a “GMGN for prediction markets” with smart money tracking, money flow, hot markets and AI recommendations. Lao Bai thought the product was solid. Yet the team eventually pivoted into perpetuals. The reasons they gave, he said, were convincing.
- First, prediction markets do not capture the benefit of liquidity spillover. DeFi Summer exploded partly because zero rates and heavy liquidity drove capital from U.S. equities into crypto and from Bitcoin into altcoins. Prediction markets are a format for expressing views, not an asset class that can soak up excess market liquidity in the same way.
- Second, the frequency of attention is too low. Politics and sports generate large events every few days or every few weeks. Memes generate new hooks daily or even hourly. He used pump.fun as an example of a product aligned with much higher-frequency attention.
- Third, prediction markets do not have the same “lead car” effect and upside ceiling. In memes or conventional trading markets, copy trading and leader effects can generate extraordinary upside. He cited DOGE reaching a $70 billion market cap after Elon Musk’s promotion, with some participants seeing 100x or even 1,000x gains. Prediction markets do not offer that kind of payoff profile. In the best case, one might move from a 10% implied probability to 100%, a theoretical 10x, and in most real cases the mispricing available when a user enters may only be 20% to 30%.
- Fourth, the format demands too much rational filtering. Users usually only trade prediction contracts where they believe they have an information edge, analytical edge or better judgment. That limits each person to a small number of topics. Meme markets work differently. Once a meme spreads on social media, it can quickly accumulate tens of thousands or even hundreds of thousands of holders.
For all those reasons, he said he is no longer as bullish on prediction markets as he was last year. He still believes the PMF is real. He just thinks the ceiling is much lower than the ceiling for perpetuals.
Why he invested in 42space despite that caution
Asked about products that combine prediction markets with meme elements, including 42space, which he invested in and which uses a bonding curve structure, Lao Bai said he used to dislike memes and had long been against the category. Prediction markets changed part of that view.
His current position is that memes are unlikely to disappear and may remain structurally resilient. One reason is that crypto differs from traditional finance in how many high-risk users it attracts and how many high-volatility tools it offers them. In his telling, crypto is still one of the highest-volatility markets in the world, and that is an advantage.
Prediction markets, by comparison, have a problem. Many of their functions in traditional finance can be replaced by options or credit default swaps. In crypto, they still do not offer enough intensity for users who want a more speculative experience. That is why he sees some logic in making prediction markets more meme-like and more speculative.
His reasons for investing in 42space were straightforward. He knows the founder personally, and while he does not claim 42space will definitely win, he believes the path of combining memes and prediction markets is worth exploring.
Robinhood’s model is clear: RWA for investors, memes for speculators
When the interview turned to meme activity on Robinhood’s chain and the shutdown of its native launchpad, followed by Uniswap launching pools.trade on Robinhood Chain, Lao Bai said Robinhood’s strategy is unusually clear.
As he put it, the chain does two things: RWA and memes.
Long-term investors want assets like Micron, NVIDIA and Hynix, not low-quality VC-backed tokens with little intrinsic value. High-risk traders want volatility, and that means memes. He said Robinhood understands this split very well.
He also noted that Base used memes for early traffic when it launched, because memes naturally create the possibility of 100x or 1,000x early growth and are powerful attention magnets. Robinhood’s approach, in his view, is to pull users and capital in through memes, then let those users shift into more durable assets once they want to preserve profits, for example by buying NVIDIA on the same platform.
That, he said, shows a clear reading of what both crypto-native users and traditional-finance users actually want.
Exchange shutdowns and project closures are more severe this cycle
Lao Bai said every bear-market bottom produces exchange exits and project closures, but he thinks this one is much worse than prior cycles.
In previous bottoms, he said, older exchanges such as BitMEX would not normally shut down absent a major theft or security failure. Products like Zapper, which had real users and real utility during DeFi Summer, would usually survive as well. This time, they have not.
To him, that is a sign of maturity. In prior years, growth could paper over almost every problem. Even if a business was not profitable, new users and new capital kept the system alive. Now there is no fresh capital, no user growth, and talent itself is leaving. Traders and KOLs are talking about U.S. equities instead. Once growth disappears, all the unresolved weaknesses appear at once, and businesses that once looked resilient stop looking resilient.
Talent is moving into AI, and he does not see a reversal soon
On talent outflows, Lao Bai was even more pessimistic than he was on market conditions. He said he does not see a credible reversal in the next several years.
Part of why he used to be so bullish on blockchain, he said, was the quality of people entering the field. Back in 2023, he wrote that among the four or five interns ABCDE had hired, three came from Tsinghua and one from Peking University. Among colleagues, Siyuan held a database PhD from HKUST and Joy was from the University of Pennsylvania. More broadly, he said he had spoken with 1,300 to 1,500 projects during his two-plus years at ABCDE, and around half of the founders he recalls were from Ivy League schools or elite universities such as Harvard, Stanford, Berkeley and MIT.
But that was in 2023, when AI and ChatGPT had only just exploded and the shift was less obvious. Today, he said, many people he knows, including former interns and colleagues, have moved into AI.
He does not think the explanation is only about making money. Another reason is that many of the truly interesting problems in blockchain were already addressed over the previous five or six years. After Ethereum was created, the space had major open questions around distributed consensus, open finance, on-chain trading, MEV, scaling, censorship resistance and privacy. By 2024, he said, most of those questions had already been answered, or at least pushed into mature solution space.
What remains often looks repetitive: a new chain or new Layer 2 launches, attracts traffic with memes, adds a wallet, token incentives, vaults, lending and versions of Uniswap or AAVE. In his words, everyone is playing the same game now, and it is no longer exciting. For graduates from schools like Harvard, Stanford or MIT, building copycats is not just less profitable. It is less interesting.
AI, by comparison, still has countless open problems to solve over the next five to 10 years. That is why he does not see a strong reason for AI talent to return to crypto at scale.
Ethereum did many things right, but on-chain financial innovation is nearing exhaustion
Lao Bai did not criticize Ethereum’s development path as harshly as some in the market have. He said Ethereum did a lot well. Much of the innovation that powered DeFi Summer came from Ethereum, including the Layer 2 concept and the smart-contract framework itself. Solana, in his words, found a different angle and executed well too.
He described Ethereum as the foundation layer. What remains to improve are issues such as privacy, staking ratios, decentralization and zero-knowledge proofs, but those feel incremental rather than revolutionary.
The larger issue is that over the last three to five years, crypto has already imported much of what Wall Street could imagine into on-chain form. That includes tranches, Pendle-style yield and risk separation, and even more complex structures related to CDS. Yet what survived as genuinely useful, he said, was mostly just DEXs and lending. More complicated financial engineering has not worked well on-chain.
Pendle, in his view, is one of the more innovative products, but even there the innovation mostly came from transplanting a Wall Street structure into crypto at about the maximum complexity users could tolerate. More complex products such as GammaSwap, or projects built around hedging, options, insurance or CDS-style structures, failed to gain traction. Even simple on-chain options have struggled. He said he has looked at more than 10 such projects, and none really broke out, while Deribit remains dominant.
That is why he thinks the number of deep open problems in on-chain finance—problems that demand the world’s smartest builders—is much lower than before.
How his trading framework changed after moving into U.S. equities and options
Lao Bai said his heavier focus this year on AI stocks, semiconductor names and options has changed his trading in three ways.
First, he is trying to build a more mature trading system. In crypto, his trading had often been driven by narrative. Once a trader makes money in a phase like DeFi Summer, it is easy to become path dependent and put outsized size into creative-looking altcoins. He said he gave back a meaningful amount during the Luna period in 2022 and again in the GOAT move at the end of 2024. Even without leverage or futures, he still lost money in alt spot because his framework was immature. U.S. equities forced him to think much more carefully about drawdowns, capital allocation and the cost of tying up capital.
Second, he learned to respect not knowing. In crypto, especially as a primary-market investor, it can be easy to read a white paper, inspect DefiLlama data, spend an hour or two with a founder, break down tokenomics and then think you understand the project well enough to buy. In U.S. equities, he said, a stock price reflects revenue, margin, inventory, capex, rates, options positioning, implied volatility, supply chains and competitors all at once. No one can fully know everything, and recognizing that is part of dealing with a mature market.
Third, and most important for him, there are more ways to express a view. In crypto, expression is blunt: if you are bullish, buy the token; if you are more bullish, add leverage through perpetuals. Traditional finance gives more tools. If you like NVIDIA but think it is expensive, you can sell a put, as Duan Yongping has done. If you like Hynix or Micron over the long run but worry about a near-term pullback, you can use a call spread. If the trade is really about volatility rather than direction, as in his attempt to trade around SpaceX’s second launch, you can build a straddle. If you hold a lot of semiconductor and AI exposure but fear the market is overheated, you can buy puts as protection.
Those instruments, he said, force a trader to clarify what the view really is: bullish on the asset, bullish or bearish on volatility, or bullish or bearish on time.
He avoids pre-IPO perpetuals if he cannot understand them clearly
Victor asked how Lao Bai would approach pre-IPO names such as Unitree, where there may be a pre-market perpetual contract on Hyperliquid before a listed stock or options market exists.
Lao Bai said plainly that if he does not understand something, he does not participate. He gave SpaceX as a similar example and said he did not trade it there either. Without a listed equity to anchor price discovery, he sees those contracts as pure short-term sentiment games in perpetual form. Even a correct directional view can be wiped out by a sudden wick. For that reason, he stays away.
His advice for founders, traders, VCs and exchanges
At the end of the interview, Lao Bai offered separate advice for different roles in the market.
Founders: do not sacrifice PMF just to remain “in crypto”
He said the startup environment is very different from the earlier years of primary-market enthusiasm. If a founder merely has a good idea or wants to import a traditional-finance mechanic into crypto, that is no longer enough. Teams now need stronger institutional resources, especially in North America, and they need to find real PMF and real users.
More than that, if removing crypto from the product makes the product better, then remove it. There is no need to sacrifice PMF just to stay inside the category. Echoing Haseeb, he said crypto should gradually become part of the underlying technical stack rather than the thing the company organizes its identity around.
Traders: do not build your identity around one market
His advice to traders was to avoid identifying with any single asset class. A person can trade crypto, U.S. equities, Taiwan equities, Korean equities or A-shares. That identity is not important. What matters is where the best odds and the strongest edge are, and that is where risk budget should go.
VCs: do not be a pure primary-market crypto VC
He joked that he may not be the right person to advise other VCs because “our own VC is closed.” Still, if there is one lesson, he said, it is not to operate as a pure primary-market crypto VC. The better model is to combine primary and secondary markets.
He said the firms that survived this cycle usually had both. In the primary market, they did not have to insist on token deals and could buy equity. In the secondary market, they could own Bitcoin and Ethereum, hedge through perpetuals, or even buy listed stocks such as Robinhood and Coinbase. The point is not to lock oneself into the identity of a pure crypto token VC.
Exchanges: the real competitors are no longer just Binance and OKX
Lao Bai said exchanges face the hardest challenge of all. They can no longer think of themselves merely as crypto exchanges. Their real competitors are not only Binance, OKX, MEXC or Gate. They are Robinhood, Interactive Brokers, Tiger Brokers and perhaps even traditional banks.
Users, he said, care about only a few things: where there are the most assets, the best liquidity, the lowest costs and the easiest experience. The exchange endgame is to stop being just a token-listing venue and become the global gateway to risk assets.
That is why he repeated one of his strongest lines from the interview: Robinhood is the final form of the exchange. In the model he described, users should be able to deposit USDT or USDC, buy U.S. stocks and crypto, use leverage, trade contracts and access prediction markets all in one place. He compared that to a WeChat-style super app and noted that Elon Musk has long admired that form, including through products such as X Pay and ideas about integrating stablecoins into X.
Victor and Mr. Z closed the conversation by thanking Lao Bai for the wide-ranging discussion, which moved from crypto’s past to its remaining opportunity set and then into AI and macro observations. The original piece also invited listeners to follow 168X on X, Substack and YouTube.