Authors: Victor, Mr. Z, 168X
From primary VC to cross-asset trader, after putting down the blockchain "hammer"
In August 2026, Crypto is at the bottom of the industry: the market is in a deep bear, exchanges like BitMEX and BitMart have announced the cessation of trading operations, star products from the last cycle like Zapper and Fantasy Top have shut down one after another, and DEX and DeFi have been hacked in succession. More critically, talent and capital are flowing massively to AI, and primary market VCs are starting to clear out. At this moment, 168X invited Lao Bai (@Wuhuoqiu) to discuss why Crypto has clearly won, yet everyone feels like they are losing? What stage are we at now, and what opportunities remain in the next cycle?
Lao Bai has ten years of network engineering experience. In 2017, he entered the crypto space because he wanted to buy Bytom, and subsequently worked in research and investment at Amber, ABCDE, and OKX Ventures. Now his research scope spans AI, semiconductors, and traditional finance on-chain. He has narrowed his focus within the crypto space to stablecoins, perps, RWA, and prediction markets—the tracks that still have PMF. In this nearly two-hour conversation, Lao Bai offers a series of sharp judgments: issuing a token is essentially a liability, not financing; the Crypto VC species will disappear; the ceiling for prediction markets is far lower than for perps. He believes the era of treating blockchain as a hammer and seeing everything as a nail is over, and the two strongest inventions in this track are not new public chains but stablecoins and perpetual contracts. And Robinhood is the ultimate form of all future exchanges.
I. Lao Bai's Investment Evolution: From Engineer, Primary VC to Cross-Asset Researcher
Victor: First, let's invite Lao Bai to introduce himself and share his experience. What tracks are you currently most interested in?
Lao Bai: Thank you, Victor, and thank you 168X. My previous work experience was first as a researcher at Amber, then as an investment research partner at ABCDE, and then I spent a few months in investment at OKX Ventures.
Now my personal focus is actually quite narrowed. Within Crypto, I have narrowed it down to a few tracks: perps, prediction markets, and RWA. These are also the tracks that I think are among the few that still have PMF (product-market fit) in Crypto. My other energy and time are spread to US stocks, especially AI-related US stocks, and some other investment opportunities. Recently, I have been paying more attention to options, because options greatly expand my way of expressing an asset, whether in terms of price or value, so I have been researching this area recently.
II. Treating Blockchain as a Hammer: Why Most Star Tracks Are "Blockchain for the Sake of Blockchain"
Victor: From 2017 to now, we have experienced several cycles. You previously wrote an article analyzing Web3 Game, mentioning that "we once treated blockchain as a hammer, seeing everything as a nail," and that many things in the world do not need to be financialized. Based on your past experience as a VC and investor, how do you view the development and evolution of these projects over the past cycles?
Lao Bai: I want to expand on this question because I think it's a particularly good one. When I was doing primary investments myself, I had this feeling: whether it's VCs or founders, many of us are making the same mistake, treating blockchain as a hammer and seeing everything as a nail.
If you start from the first principles of blockchain, setting aside terms like decentralization, privacy, and censorship resistance (which 99% of users don't actually care about), the core value of blockchain is actually what an article written by Li Xiaolai or Lao Mao in 2017 said: blockchain is the world's first network that achieves peer-to-peer value transfer. The internet allows us to transfer information, while Bitcoin, or blockchain, allows us to transfer value, not just information. How does it achieve this? Simply put, it prevents the infinite copying (copy paste) of information by preventing double-spending through technology. You can even think of blockchain as an evolved version of BitTorrent, but instead of a peer-to-peer download network, it becomes a peer-to-peer value transfer network. And an important feature is that assets on it can be programmed.
If you dig into Satoshi Nakamoto's earliest posts around 2009 to 2012, you'll find that his vision at the time was actually to carry more assets on Bitcoin after its success, including invoices, bonded contracts, etc. So Bitcoin had already thought about doing what Ethereum later did, but then Satoshi retired, and Vitalik emerged, using Ethereum to realize what Bitcoin originally wanted to do.
That's why in 2017 I was particularly bullish on Bytom. At the time, Chang Jia's concept of bringing real-world assets on-chain seemed to me exactly what blockchain should do, so the first altcoin I bought was Bytom. But looking back now, Bytom was too early and became a martyr rather than a pioneer. By the DeFi Summer wave in late 2020 and 2021, because asset prices were skyrocketing, I got carried away and forgot the original intention of "assets on-chain."
But by 2024, I already felt something was wrong. At an internal meeting at ABCDE, I discussed: after taking stock, I found that the dozen or twenty tokens I personally thought were valuable were all things that enhance the programmability of the entire blockchain network. From Uniswap V1 to V2, V3, Aave also V1, V2, V3, to later Morpho, Pendle, we have been continuously enhancing programmability, as well as the speed and capacity of the entire network—that whole Infra suite. But where are our assets? I found almost none. We tried NFTs, GameFi game items, inscriptions, Ordi, and even current Memes, and ultimately found that none have long-term asset attributes and value.
As for game items, this was first promoted by Vitalik, who mentioned in 2017 that finance and gaming were the two scenarios that could land first. So we thought of treating game assets as a type of RWA asset to put on-chain, and unsurprisingly, it has now failed. So half a year ago, I wrote an article titled "Why World of Warcraft Succeeded and GameFi Failed." My view is: anything that can be scaled can definitely be industrialized, and there is an impossible triangle in it. Even if you made assets in World of Warcraft freely tradable like on blockchain, World of Warcraft would be destroyed by gold farmers, let alone our chain games that have no playability at all.
As for the star tracks of each cycle, I think there are several types. One type is like Bytom, which was too early and became a martyr, including Augur, Ethereum's earliest prediction market, which also had the right direction but wrong timing. Another type is those where execution didn't keep up, like the earliest DEXs, Bancor, Kyber, which were later overtaken by Uniswap.
Victor: Bancor seems to have shut down recently too.
Lao Bai: Yes, that token has already failed, and everyone has defaulted to it being a failed project, so whether it shuts down or not doesn't matter much. So now when I look at projects, I basically look at how to distinguish: is it blockchain for the sake of blockchain, or is it blockchain to serve business, reduce costs, and increase efficiency? The simplest way is to remove the token and see if the product still stands. You'll find that Uniswap stands, Aave stands; but GameFi like Axie Infinity and STEPN completely don't stand. So projects where the product doesn't stand without the token, we can now all consider as not the right development direction. You ultimately have to figure out what problem blockchain solves that traditional finance cannot, and then see how it reduces costs and increases efficiency in the real world.
III. Why the Token Issuance Model Has Failed: From "More Meat, Fewer Monks" to "More Monks, Less Meat"
Victor: So now when you evaluate projects, you mainly look at whether it has PMF and can generate cash flow. Then, in the past, issuing tokens was a model that allowed VCs, retail, and various industry roles to participate. Why did token issuance work in the past but not now?
Lao Bai: Before issuing tokens, it worked because the industry was in a very early stage, and people could give it a high valuation based on "market dream rate." Even if you had no users or revenue, as long as KOLs and VCs believed your business model could succeed in the next five to ten years, they were willing to give you a high valuation. Retail investors' pursuit of these tokens follows a completely different logic from stock investing; it's entirely based on the judgments of KOLs and various VC institutions. For example, if you got money from a16z or Paradigm, retail investors would endorse the institution's view. After the project issued tokens, retail investors were essentially buying a "market dream rate option" for the next five to ten years.
The second reason was that there weren't enough projects in the market at that time. Funds might pour in with hundreds of billions or even trillions, but we only had dozens or a hundred or two hundred projects. That was a typical situation of more meat than monks. But this state wouldn't last long. Later, more and more projects entered the market, more and more VCs were founded, and everyone started making a PPT to get VC funding and endorsement, then high FDV, low float, and lock-up to control supply, making retail investors the last leg of liquidity. When this crossed a tipping point and became more monks than meat, the whole cycle broke, liquidity couldn't support it, and it became what it is now.
Victor: Many communities still discuss whether projects like Polymarket, which already have strong cash flow, need to issue tokens. People say OpenSea was overtaken by Blur and gradually declined because it didn't issue tokens at the most opportune time. Do you think this logic is correct? Can Polymarket use token issuance to capture a larger market and surpass Kalshi?
Lao Bai: From a retail investor's perspective, you could say Blur succeeded and OpenSea failed because it didn't issue tokens at the right time. If it had, its valuation would have been at least ten billion. But from Polymarket's or OpenSea's own perspective, with such strong cash revenue, there's no need to issue tokens. Because essentially, the token you issue is your debt. For the project itself, issuing tokens doesn't necessarily maximize their financial or brand commercial value. Moreover, Blur, as I understand, isn't doing very well now; it just briefly surpassed OpenSea during that period.
Polymarket could indeed surpass Kalshi in traffic and other aspects by issuing tokens. But looking at recent months, when opinion did its token airdrop from late last year to early this year, its trading volume almost approached Polymarket's. But after issuing tokens, both trading volume and the token have almost gone to zero. So I think from Polymarket's perspective, they don't have a strong need to issue tokens; it's optional.
4. Crypto Has Clearly Won, Why Do Old Players Feel They've Lost?
Victor: There's a situation now: many VCs and retail investors feel Crypto has already won, with Bitcoin entering the mainstream, getting ETFs, and gaining increasing recognition from Wall Street. But why do old players in the industry feel they've lost and aren't making as much money as in past cycles? Is it because Crypto is gradually being co-opted by traditional finance, leaving less opportunity for retail investors?
Lao Bai: Indeed, and I think both things are true simultaneously: First, Crypto has matured; second, Crypto's original ambition to change the world has indeed shrunk. In other words, we started with the intention of creating a new Crypto world natively, but gradually became reformers and vassals of the existing financial system, somewhat like Liangshan being eventually recruited by the imperial court in "Water Margin."
If you took a time machine back seven years and told VCs, founders, or retail investors from 2018-2019: stablecoins have now become a crucial global dollar infrastructure with hundreds of billions in scale; Bitcoin has ETFs with pension and institutional money buying; Crypto companies can participate in IPOs; RWA is truly entering the traditional financial system, with companies like Stripe, Visa, and BlackRock deploying Crypto infrastructure, building public chains, and doing PayFi. People from that era would definitely think: we've already won.
But from an asset perspective, we indeed haven't won. You won't see another DeFi Summer or altcoins going up 50x or 100x in a year. Moreover, over the past decade, the native issuance of assets on-chain, from NFTs to inscriptions to various Memes, I think we've also lost. We ultimately only built a track for on-chain finance, and now we're bringing real-world native assets onto the chain.
However, we have two great inventions: one is stablecoins, and the other is perpetual contracts. These are what I consider our great victories and truly Crypto-native PMF.
Mr. Z: I resonate with your remarks. A few weeks ago, I saw Dragonfly's managing partner Haseeb say that Crypto has matured, and he thinks there might not be Crypto VCs in the next ten years, just like in the mid-2000s when we invested in IoT and social platforms like Facebook, and now no VC claims to invest in those. So do you think Crypto VCs will cease to exist in five or ten years?
Lao Bai: I basically agree with Haseeb's view. In the future, as a VC or even a trader, you shouldn't treat Crypto as a separate industry or sector to worry about. Just as you wouldn't call a VC an "internet fund VC" now, the internet serves as a technology for the business system; Crypto will similarly become a technological foundation integrated into your entire business model. When you need to use stablecoins or put some things on-chain, you might allocate 10% to 20% of your involvement in Crypto-related business projects or companies. It's hard to say you're a "Crypto VC" or that this is a "Crypto project." I think Crypto should gradually integrate into the real world like the internet, and people won't need to worry about whether it's a Crypto asset, Crypto project, or Crypto VC; these labels will disappear.
5. Perp DEX Competitive Landscape: Four Little Dragons and HIP-3, Who Can Challenge Hyperliquid?
Mr. Z: This resonates with the operations of Perp DEXs like Hyperliquid and Variational, which use blockchain as the base and bring assets on-chain, such as using protocols like HIP-3 to allow deployers like TradeXYZ and Paragon to bring assets on-chain, enabling users to trade stocks from various countries 24/7. What do you think the competitive landscape for Perp DEXs will look like in the next two to three years?
Lao Bai: First, I'm personally very bullish on Variational. Beyond Hyperliquid as the T0, my preferred second tier includes Aster first, Lighter second, and Variational third. Variational's RFQ (request for quote) plus off-exchange hedging model makes a lot of sense logically, and the team, founder, and backing capital and institutional resources are all excellent.
I think the market only needs four to five Perps, just as exchanges only need four to five. Previously, it was Binance, OKX, plus some user favorites like Bitget, Bybit, MEXC, and Gate. In the end, only four to five Perps will form a leader effect: T0 is Hyperliquid, and T1 includes Lighter, edgeX, Variational, etc.
A couple of days ago, I chatted with a friend who raised an interesting point. As Perps and CEXs in the crypto space become more like leveraged contracts of traditional finance and stock brokers, many small and mid-cap stocks with poor liquidity might adopt the Binance Alpha playbook. The approach is: achieve 60% to 70% control of the float at the spot level, continuously buy to push up the price, and simultaneously create a narrative, like a traditional company on the verge of death suddenly announcing a pivot to AI or DAT (Digital Asset Treasury). At that point, you might stack a very high OI on Hyperliquid or Binance, and then the company's shareholders or stakeholders exit through the Perp. Regulators can't find any loopholes at the spot level because the company only buys back its own stock, only buys and never sells, so regulators can't say anything; but selling happens in the Perp. This is a long-term regulatory arbitrage, combined with our Perp DEX, creating a new playbook for other small and mid-cap stocks with poor liquidity worldwide.
Mr. Z: I've observed that latecomers other than Hyperliquid need different strategies. Hyperliquid already has many US stocks, and standard US stocks like interconnect and storage are common. Can latecomers like Variational explore Asian stock markets, such as listing Taiwan stocks like UMC, MediaTek, TSMC, or Korean, Japanese, or Hong Kong stocks like Alibaba, Tencent, Meituan, JD? Because directly competing with Hyperliquid is impossible. What's your take on this?
Lao Bai: Perps like Lighter and edgeX, which share the same architecture as Hyperliquid based on CLOB and market makers, should indeed pursue differentiated competition. The Asian assets you mentioned, TSMC, MediaTek, Korean stocks, all those directions are correct. I remember Mable's Trasia is focusing on Asian assets. Another friend of mine is also planning to leave VC to start a Perp targeting on-chain forex. We've talked about this for years but haven't succeeded. If stablecoins reach trillions on-chain, the TAM for on-chain forex would be enormous.
But projects like Variational, including Ondo Perp, are different. If you look at their slippage and liquidity, they can achieve tighter and more convergent spreads than Hyperliquid or even Binance. Because Hyperliquid essentially uses a single market maker to provide liquidity, while Ondo is similar to an RFQ quote, using its own TradFi for hedging, and it's extremely convenient to lock in hedges on-exchange. So Ondo Perp, on many RWA, especially big names like Nvidia and Micron, has better spreads and slippage than Hyperliquid and Binance. I checked at the time. Variational's mechanism can theoretically also achieve what Ondo does, and its swap launched last month has already introduced traditional TradFi hedging mechanisms on commodities like crude oil and gold. So I think these two have different mechanisms and can be considered to directly compete with Hyperliquid.
Mr. Z: Then looking inside Hyperliquid, regarding HIP-3, TradeXYZ is still the leader, but there are also rising stars like Paragon with quite shrewd tactics, emphasizing speed and efficiency. Wherever there is a hot spot in the market, they quickly bid for the ticker in auction and launch assets, and they don't build a frontend, believing Hyperliquid's frontend is good enough. Are you following this?
Lao Bai: I follow it, but maybe not very closely. On the Hyperliquid frontend, I can only see deployers like TradeXYZ and Paragon. I see some assets on Paragon that are somewhat different from TradeXYZ, but its volume is still very small, mostly in the hundreds of thousands, while TradeXYZ often has tickets worth hundreds of millions, a huge gap.
HIP-3's ticker originally only marked a deployment opportunity. The first three slots are free, and subsequent ones are bid through auction. What asset you want to do with the slot is entirely up to you. But I think this will likely be a winner-take-all effect, and it's not just about user mindshare. Because market making is not officially participated by Hyperliquid; it's TradeXYZ and Paragon themselves doing market making. For the same asset, if TradeXYZ's market maker provides very good liquidity and spreads, everyone will go to TradeXYZ, creating a self-reinforcing flywheel effect. How to break this is really something latecomers need to think about. And it's not like Uniswap, where you can use new AMM mechanisms, like Curve did for stablecoins with different curves to achieve different effects; HIP-3 is purely about competing on market maker liquidity and depth, it's about who has more real money, and it's hard to find clever angles.
6. Security is a Function of Time: Lessons from Ostium, Cream, and Perp
Victor: There have been some recent incidents in Perp, like TradeXYZ had a Hynix flash crash a while ago, Binance also had one but with less impact, and TradeXYZ chose to compensate all user losses, which is very generous. Additionally, some projects have been hacked or had bugs recently, such as RWA project Ostium being hacked for about $18 million, and Paradex also being hacked earlier. How will these security issues affect people's hesitation about DEX security, and whether Perp can be more accepted by traditional finance and investors in the future?
Lao Bai: I think this depends on two points. First, it takes time to prove, because 99% of users won't look at your underlying code; they rely purely on reputation and time. Why is AAVE so strong and has such high consensus? Because it's a lending protocol that has never had security issues; the previous incident was actually a problem from the underlying asset penetration, not AAVE's own problem.
You'll find that most security attacks on lending platforms occur in the lending itself: using a junk asset as collateral to extract a lot of stablecoins through oracle issues, price pumps, or vote control. Like Mango on Solana, and many junk lending platforms have been attacked, even multiple times. I remember Cream, the lending platform of Machi (the Taiwanese), was attacked three or four times, and I never understood why people still bravely deposited money after it was hacked.
So for Perp, I think there are two points. First, don't expand your Perp business to the point where you can collateralize certain assets and also do lending. I remember Paradex, or maybe Drift on Solana, wanted to be all-in-one, doing spot, collateral lending, and Perp, and eventually the lending side opened a hole and was breached. If you do Perp, just do Perp honestly, don't use collateral assets to extract stablecoins. Once you do that, something will happen one day. Second, there's no other way but to prove it over time. If after four or five years Hyperliquid has never had any security incident, while others have, then people will naturally trust it and be willing to deposit money there, just like Binance, completely occupying user mindshare.
7. Will Hyperliquid Do Spot? The Ideal and Reality of Tokenized Stocks
Victor: On the spot side, Hyperliquid hasn't launched spot trading yet. Do you think they might enter spot? If they want to keep up with the strategy you mentioned like Binance Alpha possibly helping stocks launch assets, wouldn't they need to do spot?
Lao Bai: I don't think they will do spot for native crypto. The influence of spot is too small now. You see, 80% to 90% of Binance's revenue comes entirely from Perp, and within Perp, Binance and HTX have had over 60% from US stocks, i.e., RWA, not native crypto, since last month.
So if you want to list spot, either you list a spot stock, similar to Binance's bStock. I'm not sure if Hyperliquid will have something like a "hyperstock" in the future; it's possible, though not very likely. But if they do, the sole purpose would be more to serve its Perp; you can do basis arbitrage and on-exchange hedging. Just like why Ondo's Perp liquidity is so good, because it has its own Ondo Stocks on-exchange. The market maker quotes you, you open a $1 million long on Micron, and it can simultaneously give you a short as the counterparty, then directly hold $1 million of Micron spot with zero delay. So if Hyperliquid does this, I wouldn't be too surprised, but it will never do spot for crypto assets like BTC, ETH, SOL.
Victor: Between tokenized stocks and Perp, if you had to choose one, which is the better model? Or are they complementary? Like StableStock, which positions itself as a stablecoin broker, letting people buy stocks directly with stablecoins, but doesn't do Perp; while Binance is a full suite, with Perp, US stocks, and tokenized stocks.
Lao Bai: The ideal state of tokenized stocks should be "real stocks, plus clear legal ownership, plus free on-chain transfer." This is something Nasdaq itself might do. I remember a couple of days ago, the US SEC approved Nasdaq to do on-chain stocks. So in terms of stock spot, StableStock, including BIT I cooperate with, the price competitor might be Nasdaq itself, but Nasdaq probably won't allow you to freely buy and sell with stablecoins in the short term.
But from a realistic perspective, the vast majority of users don't need clear legal ownership and real stocks; they just need a price exposure to bet on, as long as that exposure can anchor to the real stock price. So the most mainstream form now is Perp. That's why I say over 60% of revenue comes from Perp, and within Perp, 60% is RWA. Ideally, everyone would be holders and long-term investors, like traditional US stocks; but in reality, everyone is a gambler, needing exposure, and then adding five or ten times leverage on that exposure, frantically going long and short. Ideal is ideal, reality is reality.
8. Stablecoins: Crypto's Greatest Invention and the Global Extension of Dollar Hegemony
Victor: Next, let's talk about stablecoins. Recently, projects like Open USD have gathered traditional financial institutions like Visa, Stripe, and Mastercard, reminiscent of Facebook's Libra back then. How do you view the evolution of the stablecoin track in the next cycle? If new projects want to emerge, what differentiated advantages do they need?
Lao Bai: I am infinitely bullish on stablecoins, super bullish. I think this is the most important invention besides Bitcoin, no exceptions. Crypto's greatest contribution to the world is actually stablecoins, even more than Bitcoin. Bitcoin is now mainly a game for a few whales, institutions, and big players, but stablecoins are the technology that truly changes the world on the blockchain.
Last year we were talking about stablecoins paying off debt. Last year it was $36 trillion in national debt, now it's $39 trillion. Everyone is thinking, stablecoins should reach one or two trillion before 2030, at least helping to pay off two trillion of US debt and reducing the US government's burden; in another ten or twenty years, it might reach five trillion, ten trillion. You can see now on the B-side, Stripe and Visa are getting involved themselves, from payments to yield-bearing scenarios; on the C-side, needless to say, you can use stablecoins to buy and sell various assets in the world. So ultimately, I think stablecoins and tokenized US stocks have a synergistic effect: dollar hegemony plus dollar asset hegemony, penetrating all countries and corners of the world through stablecoins.
As for wanting to be a new player, I think the requirements are too high. New players can only be super players like Stripe, BlackRock, and Visa, who already control traditional payment, traditional currency and financial entry points, or intermediate circulation links, and are qualified to get a share of the stablecoin business. Ordinary entrepreneurs, I suggest not touching this area; it's a business that relies heavily on institutional connections.
9. Agent Payments and the Machine Economy: Cloudflare's Traffic Inflection Point and Micropayments
Victor: The narrative of stablecoins and AI agent payments was very hot last year, like x402, and recently Cloudflare launched agent wallet, and in the first half of the year Tether also launched its own wallet, all aiming to promote agent payments with stablecoins and the machine economy. Is this a viable narrative or more of a dream?
Laobai: This narrative is very viable, but whether it has anything to do with crypto is another matter. Let's look at it from two aspects.
First, will the agent economy actually happen? I think it will. Last month, Cloudflare's CEO said in an interview: On a certain day last month, bot traffic on the world's internet surpassed human traffic for the first time. This happened on a day in July and is a very big milestone. Previously, the internet had a lot of bots, but human traffic dominated. Now, as the world's largest gateway/CDN, Cloudflare's data shows bot traffic has finally exceeded human traffic. The reason is large language models: if a human wants to buy a single-lens reflex camera, they might check five web pages and compare prices, so HTTP requests are based on those five visits. But if you ask ChatGPT or Doubao to "recommend a single-lens reflex camera for 5,000 yuan with certain features," it might make 5,000 HTTP requests to browse all relevant e-commerce and review sites to give you the best answer. So the machine era represents an explosive demand for network traffic. Cloudflare's CEO believes agent traffic will expand a thousandfold in the next five years.
Second, he raised a question: Previously, the internet economy was driven by Google Ads. We completed the entire business loop based on search and recommendation through ad clicks. But when agents make recommendations, this model is broken. The only solution seen so far is to make a micropayment when an agent scrapes your website's data. For example, if you are an e-commerce or review site, and an agent uses your data in 5,000 scrapes, it pays you two cents. How is this two cents paid? It will likely be through something like x402, or stablecoin-based solutions gradually introduced by Cloudflare, Visa, or Stripe. It certainly won't be traditional Visa, and it may not necessarily be crypto, but it will definitely be a stablecoin on some chain. That chain might not be Solana, Ethereum, or Tempo or Arc—we don't know—but I think it will definitely be such a scenario.
10. The Endgame for Public Chains: The Era of General-Purpose Chains Ends, with Institutional Chains, C-End, and Gray Industries Dividing the Market
Victor: How will stablecoins and agent payments affect the competitive landscape of public chains? Currently, the largest is still Ethereum, followed by Solana, Circle's Arc, Stripe's Tempo, and last year there was Plasma, etc.
Laobai: In the next cycle, I think it will likely be "render unto Caesar the things that are Caesar's, and unto God the things that are God's," with everyone advancing side by side. If you are in the gray industry, you might still choose TRON and USDT; if you are a retail investor, you will likely choose USDC on Ethereum and Solana to play DeFi and Meme; if you are an institution, B2B, enterprise payments, or foreign exchange, you will likely use institutional chains like Tempo, Arbitrum, and also a privacy-focused one called Canton, which is also developing very well. So it will likely be a landscape of three or four institutional chains serving the B-end, Ethereum and Solana serving the C-end, and TRON serving the gray industry.
Victor: In past cycles, everyone treated public chain tokens as the most mainstream beta, like Ethereum and Solana. Will public chains still be the main axis for VC or retail investment going forward?
Laobai: Definitely not. Hyperliquid has basically set a benchmark for all public chains: public chains must serve applications. I am a chain, but I exist to serve a super powerful application, not to be a universal general-purpose chain. General-purpose chains may have reached their end after Ethereum and Solana, but we are still unwilling to give up, so we have Aptos, Sui, Monad, and MegaETH. You see, recently Shu Yao from MegaETH also tweeted that their MegaMafia incubation program will stop, they will no longer fund new projects on the chain, and they have decided to build applications themselves. This shows that the era of general-purpose chains can be declared over, and the future should be an era of application chains.
This is actually a bit like AI: first build infrastructure, we must stay one version ahead, get the infrastructure done first, and the valuation of infrastructure is also particularly high. Now people are finally starting to pay attention to applications, PMF, users, and cash flow. The crypto circle is starting to become more like the US stock market. On the contrary, semiconductors and memory in the US stock market are a bit like the previous version of our crypto circle, starting to "crypto-ize," so many people have died from 2x leverage long on SK Hynix.
11. Large Models Moving Toward "Public Chain-ization," Killer Apps, and the Inspiration from Palantir
Victor: Now large models are a bit like the public chains of yesteryear. The barrier to building large models is getting lower, and the main battlefield is shifting to models specialized in certain applications. Do you think large models will develop in a direction similar to public chains?
Laobai: I think it's already happening. Look at China's "Six Little Dragons," like Kimi, DeepSeek, and MiniMax. They are already similar to the feeling of public chains, each large model has its own characteristics. For example, MiniMax's new model recently came out as an uncensored version, and many people doing video are happy. So we have already differentiated in terms of technology, cost-effectiveness, and moral censorship, a bit like the public chain wars back then.
So who will ultimately break out still depends on two points. First, who can better integrate into the B-end. OpenAI and Anthropic have both set up their own forward deployment engineer teams, learning from Palantir, trying to deploy AI into enterprises. Because look, we are so hot, but more than 90% of enterprises have only equipped their employees with Anthropic, OpenAI, or Doubao, while the entire enterprise process and architecture remain traditional. AI's overall improvement to enterprises may be less than 20% to 30%, but its improvement to individuals may exceed 500%. So OpenAI and Anthropic are a bit anxious and have decided to go down themselves to set up companies to help enterprises integrate AI into their workflows. Second, on the C-end, whoever makes a killer app first wins, whether they can create other killer apps beyond ChatGPT, Codex, and Claude.
Victor: A few days ago, Palantir's CEO also mentioned in an interview that the narratives Anthropic is telling the public now are actually what Palantir has been doing all along, which is quite funny.
Laobai: Yes, so Palantir's stock has been rising very well recently. People realize that these powerful SaaS platforms are not easily replaced by large models, and large models are not easily integrated into an enterprise's workflow.
12. The Truth About Prediction Markets: Two Years to Complete the Gartner Hype Cycle, with a Ceiling Far Below Perp
Victor: Finally, let's talk about the prediction markets that everyone is very focused on. You have also invested in some prediction market projects, like the prediction market 42space invested by YZi Labs. How do you view the current landscape of Polymarket, Kalshi, and the overall prediction market?
Laobai: Yes, I just posted a tweet a couple of days ago about the four stages of prediction markets: undervalued before the election, proven viable after the election by Kalshi and Polymarket, overvalued before the World Cup, and now returning to rationality. You can see that Polymarket's volume has been declining basically since July 18th after the World Cup. So now I think if Polymarket really gets a valuation of around $20 billion, or really issues a coin to a level of $20-30 billion, I would short it without hesitation. Because from what I've observed, prediction markets are not on the same level as Perp at all.
At that time, we had many irrational imaginations about prediction markets, a bit like "if you have a hammer, everything looks like a nail." When I was at ABCDE, I talked to at least ten prediction market projects, and everyone would imagine scenarios like friends betting with each other, using a prediction market to open a private room and place bets separately. But in practice, the scenario and frequency of friends betting are actually very low. It's not a PMF at all; it's a PMF that people forced to find in order to build prediction markets.
To put it bluntly, a person does not have the need to express opinions on so many things and simultaneously bet on their opinions. For the vast majority of people, it might just be politics and sports. And for sports, traditional betting already satisfies the demand well. For example, in Europe, the US, and Australia, people like to go to stadiums to watch games, or go to sports bars, where a bunch of people drink beer, watch horse racing or ball games, and buy a traditional sports lottery ticket for a few dozen Australian dollars. If they win, they're happy; if they lose, it's just the cost of a meal. This is also why Kalshi is doing well in sports, but essentially, people's demand for these has already been satisfied by traditional markets and traditional betting to a large extent.
A while ago, I talked with a founder who makes prediction market tools. They make something like a "GMGN for prediction markets," with smart money, money flow, hot markets, AI recommendations, etc. I think it's very well done. But when I asked them why they finally chose to pivot to Perp, he gave me several points that made a lot of sense:
First, prediction markets cannot benefit from the spillover of liquidity. Why was DeFi Summer so popular back then? A big reason was that the US had zero interest rates and massive money printing, causing liquidity to spill over from US stocks to crypto, and from Bitcoin to altcoins, which perfectly matched the DeFi Summer explosion, lifting the entire industry. But prediction markets, in essence, find it hard to capture such dividends because they are a form of expressing opinions, not an asset form.
Second, the frequency of attention is insufficient. Sports and politics have major events only every few days or even weeks, but Meme is different. Look at why pump.fun is so powerful: you can catch new hotspots and new memes every day on Twitter, like the PNUT squirrel coin back then. Such things happen every day. The frequency of Meme coin issuance is on a daily or even hourly basis.
Third, there is no ceiling from the "leader effect." In the Meme market or traditional trading markets, many users follow others, creating a leader effect. The most typical example is Musk pumping Dogecoin, which reached a market cap of $70 billion. Some people made 100x or even 1000x gains on Dogecoin, with an extremely high or even unlimited ceiling. But prediction markets do not have such a ceiling. At most, you can go from a 10% probability to 100%, which is only a 10x at best. Moreover, the moment you place a bet, the market odds are likely already priced in, and the mispricing you can capture might only be 20-30%. Doubling your money is already a big win.
Fourth, rational decision-making accounts for too much. When you place a bet, you must believe you have an edge in intelligence, information, or understanding over others. I might bet on Apple, AI semiconductors, or Intel's yield, but I won't bet on a hockey or football game I've never heard of. You'll find that each person can only bet on five to ten markets, countable on two hands. But Meme is not like that: a meme that everyone on Twitter knows can quickly accumulate tens of thousands, hundreds of thousands, or even millions of holders.
So all these points make a lot of sense to me, and they also make me less bullish on prediction markets than last year: I acknowledge that its PMF is real, but its ceiling is far lower than Perp.
Victor: Then how do you view projects that combine prediction markets with Meme? For example, 42space, which you invested in, has introduced bonding curves and some Meme elements.
Lao Bai: I was actually very against Meme before and never liked it. But after seeing the prediction market form, I am not as bearish on Meme anymore. Rationally, I think Meme is a form that cannot die and even has lasting vitality. In summary, the biggest difference between the crypto market and traditional financial markets is that we have more gamblers here, and we provide tools with strong gambling nature for everyone to gamble. Crypto is also the most volatile market in the world, which is its industry advantage. Many functions of prediction markets in traditional finance can actually be replaced by options or CDS (credit default swaps). It can neither capture the share of options in traditional finance nor is it gambling enough to satisfy gamblers in the crypto space.
My reasons for investing in 42space are: first, I have a personal relationship with the founder; second, although I am not confident that 42space will definitely win, I think "combining Meme and prediction markets to make prediction markets more gambler-oriented" is a path worth exploring.
Thirteen, Robinhood's clear strategy: RWA for investors, Meme for gamblers
Victor: How do you view the on-chain Meme craze on Robinhood? Recently, Robinhood's original native launchpad was shut down, and Uniswap launched a new launchpadpools.trade on Robinhood Chain a few days ago. Although Robinhood focuses on RWA, a large portion of its trading volume is driven by Meme. How do you view the growth relationship between this chain and Meme?
Lao Bai: I think Robinhood's approach is very clear and very impressive. It openly states that this chain does two things: RWA and Meme. It understands what users really need. If you are a long-term investor, you need RWA assets like Micron, Nvidia, and SK Hynix, not those worthless VC air altcoins. For more gamblers, you want to play with high volatility, so I give you Meme.
When the Base chain first launched, it also attracted users through Meme first, because Meme easily creates early 100x or 1000x growth, naturally acting as a huge magnet for attention. So Robinhood is very adept at this: first attract users and accumulate funds. You might change your mind and say, "I made money on Meme and want to preserve it. What should I do? I'll buy some Nvidia on Robinhood." This strategy shows they understand what people in the crypto and traditional finance circles are thinking and wanting to play.
Fourteen, Exchange Exits and Talent Flow to AI: Why This Time Is Particularly Severe
Victor: Recently, quite a few exchanges have announced the cessation of trading operations, such as BitMEX, the former leader that invented perpetual contracts, and BitMart. Also, many projects have announced shutdowns, like Zapper and Fantasy Top, which were very popular last cycle. You have experienced several cycles. Does this happen at the bottom of every cycle? Is the exchange exit this cycle particularly noticeable?
Lao Bai: Actually, similar events happen at the bottom of every cycle, but this time it is especially severe. In previous cycle bottoms, old exchanges like BitMEX, unless they had security breaches, usually wouldn't shut down. Projects like Zapper, which I personally used and liked during DeFi Summer, had real users and real value, and would have survived previous cycle bottoms. But now you see they can't hold on, so we are truly entering a mature phase: the feeling that "growth can cover all problems" is gone.
Back then, if you could make money, or even if you couldn't but had incremental growth, new funds, and new users, everyone was happy. Now there are no new funds or new user growth; instead, our talent is flowing out, users are leaving, and all traders and KOLs are talking about US stocks every day. Without growth, all problems are exposed at once. Our circle has shrunk, causing some exchanges and projects that previously could hold on to now fail.
Victor: Talent is indeed a big problem. For example, Sui's co-founder recently announced a move to Anthropic. Compared to the 2017 and 2020 cycles, the world's smartest talent seemed to be flowing into Crypto, but now it's reversed. What phenomena have you observed around you? When might this trend reverse?
Lao Bai: I think we may not see a reversal in the coming years. I don't see any possibility of this talent drain reversing. I was particularly bullish on blockchain before, largely because of talent. In 2023, I even tweeted: at ABCDE, we hired four or five interns in total, three from Tsinghua and one from Peking University; among colleagues, Siyuan has a PhD in databases from HKUST, and Joy is from UPenn. In my two-plus years at ABCDE, I talked to 1,300 to 1,500 projects. Going through the founders in my mind, about half graduated from Ivy League schools. I talked to too many founders from Harvard, Stanford, Berkeley, and MIT. As a "second-tier college underachiever," I could talk to hundreds of founders from world-class universities in one industry, which is unimaginable in other industries. So at that time, I thought, how could I not be bullish on Crypto?
But that was 2023, when AI and ChatGPT had just emerged and weren't so obvious. Now, many people I know, including former interns and colleagues, have gone to AI. Besides the profit effect, I think the biggest reason is: In the past five or six years, after Ethereum was born, we had many interesting problems to solve, such as distributed consensus, open finance, on-chain trading, MEV, scalability, censorship resistance, and privacy. But by 2024, these problems have basically been solved to a large extent. You'll find that new projects from 2024 to now, whether a new chain or a new Layer 2, basically offer the same thing: use Meme to attract users, build a wallet, create a tokenomics incentive, set up a yield farm, set up lending, and port Uniswap and AAVE over. Everyone is playing the same game, which is not sexy or fun. If you ask Harvard, Stanford, or MIT people to do copycat work, besides not making money, they also find it boring. But on the AI side, we have countless problems to solve in the next five to ten years, just like blockchain in the past decade. So from a problem-solving perspective, I see no possibility of a large-scale return of AI talent to Crypto.
Victor: Could Ethereum's innovation path have an impact? How do you view the development of Ethereum's roadmap over the past few years? Many smart people came in with ideals back then, but the most idealistic things seem to have been disproven, and these people are starting to leave.
Lao Bai: In the past few years, I think Ethereum has done quite well. Many innovations from DeFi Summer came out of Ethereum, including the Layer 2 concept and the entire smart contract framework, which were all created by Ethereum itself. Solana is the only one that has done well by entering from a different angle. Ethereum is equivalent to building a foundation, and there are still some things on this foundation that can be patched up, such as privacy, staking rate, degree of decentralization, and using ZK for proof, but these are all minor details.
In the past three to five years, we have already tried all the gameplay that Wall Street can think of on the chain, such as tranches, splits based on risk and return like Pendle, and extremely complex risk default swaps like CDS. But in the end, you will find that what really remains useful is just a DEX and lending; things that are too complex actually cannot work on the chain. Pendle is already considered very innovative, but in fact, its innovation is just bringing over the Wall Street playbook, except that the complexity is just at the limit of what people can tolerate. For more complex things like GammaSwap, or things involving hedging, options, insurance, and CDS, I have talked about them all, but none of them have worked out. For the simplest options, I have seen more than ten on-chain projects, and none of them have succeeded; Deribit still dominates. So we have almost exhausted the innovations we can make; there are not that many problems that urgently need to be solved by the smartest brains in the world, and from an interest perspective, they prefer to explore AI.
Fifteen: From Crypto to US Stock Options: Three Cognitive Upgrades in Trading Framework
Victor: This year you have invested heavily in AI and semiconductor stocks, and you often trade options. After switching from Crypto to the stock market, what changes and insights have you gained in your trading framework? Options were an area with relatively few participants in Crypto in the past; how do you use them for trading expression?
Lao Bai: After switching to stocks, the three things that have had the greatest impact on me are as follows.
First, I will try to build a relatively mature trading system. I didn't really have a system for trading before; it was mainly narrative-driven. After making some money during DeFi Summer, it was easy to develop path dependence and take relatively large positions in some altcoins that I thought were creative. In fact, during the Luna crash in 2022 and the GOAT wave at the end of 2024, I gave back a lot of profits. Although I don't use leverage or trade futures, I still lost a lot on these altcoin spot positions. This is because my trading system was immature—too young, too naive. When you invest in US stocks that have real PMF, real markets, and real users, compared to investing in Crypto, which is driven by dream valuations, narratives, and fancy technology, the investment style is completely different. You need to establish a systematic trading framework that considers drawdown, capital allocation, cost of capital, etc.
Second, you learn to respect "not knowing." There is a dangerous habit in Crypto, especially for primary market investors like me: after reading a project's whitepaper, researching the data on DefiLlama, chatting with the founder for an hour or two, and breaking down the tokenomics, you feel like you already understand the project quite well, and then you buy. This is actually an illusion. In the US stock market, you will find it is much more complex: a company's stock price is affected by revenue, profit margins, inventory, capital expenditures, as well as macro interest rates, options positions, implied volatility (IV), supply chain, competitors, and various information that you simply cannot finish reading. So you choose to respect that "no one can be omniscient," which is a sign of a mature market.
Third, and most importantly: there are more ways to express your views. In Crypto, the way to express is very simple and crude: if you are bullish, you buy the coin; if you are more bullish, you use Perp to open leveraged long positions. If you win, you go to the club; if you lose, you go to work. But in traditional finance, there are many ways: if you are bullish on Nvidia but think it's expensive, you can sell a put like Duan Yongping; if you are bullish on Hynix and Micron in the long term but worried about short-term pullbacks, you can use a call spread; I previously bet on the volatility of SpaceX's second launch. Although the strategy failed, the idea was to bet only on volatility, not direction. If the rocket exploded, it would crash; if it succeeded, it would surge. In that case, you could use a straddle to construct an options portfolio. For another example, if I hold a lot of semiconductor and AI stocks and think the short-term market sentiment is overheated, I can buy some puts for protection. With so many ways to express, you will clarify your thoughts from vague to clear: are you bullish on the underlying asset itself, or bullish/bearish on its volatility, or bullish/bearish on its time? You need to think very clearly before choosing a way to express. This has had the greatest impact on me personally.
Victor: Speaking of targets like SpaceX, now Unitree Robotics is about to IPO, and there are no stocks or options yet, but there are pre-market contracts on Hyperliquid. How would you trade such Pre-IPO targets?
Lao Bai: To be honest, I don't quite understand this, so I don't participate. It's a bit like SpaceX's pre-market, which was only on Hyperliquid at the time, but I didn't participate either. Because this kind of thing is purely a short-term sentiment game. There is no underlying stock; everyone is expressing their views on Perp. Even if your view is correct, a sudden spike could liquidate both longs and shorts. So I don't participate in such pre-market Perps.
Sixteen: Advice for Each Role: How Founders, Traders, VCs, and Exchanges Stay in the Game
Victor: At the end of the interview, we would like to ask Teacher Lao Bai to give some advice to different roles on how to survive the bear market and stay in the game.
Lao Bai: I can talk about it by role.
For founders, the situation is different from when I was in primary market investing. If you just have a good idea, or want to bring a playbook from traditional finance, that era is over. If you start a business now, you may need more institutional resources, especially in North America, and you need to find real PMF and users. Even if you find that removing Crypto makes your product better, then remove it; you don't need to sacrifice your PMF just to stay in Crypto. As Haseeb said, Crypto should gradually become a technological foundation integrated into your product. What you should consider is which users your product serves and what it does, not "since it's a Crypto project, it must issue a token and have a tokenomics." These are not necessary.
For traders, don't identify with any market. You can be a Crypto trader, or a trader in US stocks, Taiwan stocks, Korean stocks, or A-shares; it doesn't matter. If there are no opportunities and volatility here, go to another market and look. Wherever there are opportunities and good odds, put yourself and your risk budget where the odds are best and where you have the most advantage or information asymmetry.
For my VC peers, I feel I am not qualified to give advice because our own VC has closed. But if I have to say something, it is to learn from our lesson: don't be a pure primary market Crypto VC; it's best to combine primary and secondary markets. I found that the VCs that survived this cycle all have both primary and secondary exposure. For primary investments, you don't have to issue a token; you can invest in equity, not necessarily tokens. For secondary, you can buy Bitcoin, Ethereum, or even short and long on Perp to hedge, or buy Robinhood, Coinbase, and some traditional US stocks. Don't limit yourself to being a pure primary Crypto VC; you are just a VC.
Exchanges face the most severe situation. In this cycle, exchanges can no longer understand themselves as Crypto exchanges. Your future competitors are no longer just Binance, OKX, MEXC, and Gate, but Robinhood, Interactive Brokers (IBKR), Tiger Brokers, and possibly even traditional banks. Users only care about where there are the most assets, the best liquidity, the lowest fees, and the most convenience. The ultimate goal of an exchange should shift from "listing coins" to becoming a gateway for global risk assets.
So I previously expressed a view: Robinhood is the ultimate form of all exchanges in the future. Whether in the crypto circle or traditional finance, one place should be able to do everything: deposit USDT, USDC, buy US stocks, buy Crypto, use leverage, trade futures, do prediction markets—I cover it all. In short, it's the WeChat model. Musk is also very envious of this Super App form; he has always thought China's WeChat is amazing. He is also working on X Pay, wanting to put stablecoins into X.
Victor & Mr. Z: Thank you very much, Teacher Lao Bai, for joining our Space today. We really talked a lot, from the history of Crypto to future tracks, and observations on AI, all with deep insights. Also, thank you to every listener who has stayed with us. If you like this episode, feel free to follow 168X on X, Substack, and YouTube, and share the show with more friends interested in Crypto, AI, and macro. See you in the next episode.
