Sun Yuchen’s Full Interview: Bearish on BTC and Nasdaq, Bullish on AI
Summary
Sun Yuchen is more bearish on BTC than in his last interview, and the core concern is not the traditional cycle but the risk that AI and Q day could crack cryptographic keys before markets have priced it in. He has shortened his risk window from a conservative estimate of 2 years to roughly 1 to 1.5 years, and worries that classical computing could make the breakthrough before quantum computing does. More than 1 million BTC linked to Satoshi Nakamoto, along with a total that could approach or exceed 2 million dead or lost coins, could return to circulation: “all kinds of ghosts and demons are coming back to life.”
His immediate response is not to sell everything, but to move assets into “bunker mode.” BTC should be moved to new addresses that have never signed a transaction or sent funds out; after each use, the remaining balance should be moved to another new address. That only protects people who actively move their coins and does nothing for old addresses with no one in control. Asked why he would not cut his position, he admitted he “doesn’t have a hedging strategy yet”; he will first isolate assets, watch the technical community closely, and then decide whether to sell.
ETH remains the main venue for large-scale DeFi yield farming, but it has moved from an offensive position to a residual holding while he waits for an opportunity. Sun says he sold at least 80% to 90% of his ETH around $3,600–$4,000, after a failed attempt to break $4,800 formed a fragile double top. He views BTC as a core holding he will not readily touch; he has not bought ETH back, and most of the freed-up capital went into U.S. equities.
He is still heavily invested in Nvidia and the Nasdaq, but is “slightly bearish” over the medium term and has not shorted either. His concern is that capital is crowded into AI hardware and software, while Treasuries—the “mother hen”—are being drained by equities and can pull money back with higher risk-free yields. Supply may eventually find a use, as the fiber-optic buildout did after the internet bubble, but a near-term chip glut is possible. “If something goes wrong, everyone goes down with it.”
Within the AI value chain, he would rather back real applications than memory, optical modules, and other hardware already being valued like software. He sees AI-powered health management as the best example: wearables and hospitals collect data, AI continuously analyzes it and proposes decisions, and experts provide a second opinion—reshaping healthcare and insurance in the process. Nvidia is already worth about $5 trillion; when the host suggested it might rise at most another 4x, Sun replied that there are many places in the application layer that could grow more than 4x.
He expects the renminbi to strengthen and thinks the market is too worried about a U.S. Treasury credit crisis. Treasuries can “issue new debt to pay off old debt,” supported by demand for the U.S.’s control over technology, the military, AI chips, and models. Even if rising yields reflect a shift in confidence and competition from AI assets, as long as that control remains, “no matter how you jump, you can’t escape the Buddha’s palm.”
The endgame for Hyperliquid, DEXs, and CEXs depends on whether the U.S. gives them sufficiently permissive regulatory treatment, while Tron is betting that RWA will be the second phase after stablecoins. If Hyperliquid receives exemptions close to Uniswap’s, many CEXs could shift to reduce compliance costs. Tron plans to bring most U.S. equities onchain in November, letting users in non-dollar countries move beyond holding dollars to buying QQQ and the S&P 500. More than $100B in capital parked on the network, along with its security and established user habits, supports its fee premium.
Deep dive
1. Hyperliquid opened up the possibilities for U.S. equity perpetuals, but regulation will decide the endgame
Sun Yuchen gave Solana a restrained “It’s alright.” Tron maintains normal cooperation with Solana, but Ethereum remains the venue for his team’s large-scale DeFi yield farming because liquidity on other chains is relatively small. The choice is driven mainly by yield and capacity, “not so much by the price of ETH itself.”
He sees Hyperliquid as the biggest breakout of this cycle, particularly because adding U.S. equity perpetuals gives it room to expand. Other platforms are also building U.S. equity perpetuals; the broader point is that Hyperliquid has opened up new possibilities for the industry.
The DEX-versus-CEX contest is not simply a matter of one technology replacing another. Hyperliquid sits between the two and still does not serve the U.S. market. The key question is whether the U.S. ultimately grants it a broad exemption similar to Uniswap’s or requires licenses and oversight comparable to those imposed on domestic centralized exchanges.
If the regulatory terms are highly favorable, traditional CEXs may proactively adopt a Hyperliquid-style structure to save on compliance costs and operate globally under exemptions. If exemptions are limited, Hyperliquid will face greater pressure. Sun does not think it will choose to operate onshore, given that its domestic regulatory readiness is unlikely to match Coinbase or Circle.
2. BTC’s biggest unpriced risk is Q day, not miner selling
Sun is “a bit more bearish” on BTC than when he last spoke. AI algorithms and Q day are the reasons: many investors holding BTC through ETFs such as BlackRock’s do not understand what a key-breaking breakthrough would mean, so he believes the market has not priced it in.
He sees OpenAI’s mathematical progress as a warning sign, while leaving room for uncertainty: classical computing might crack Satoshi’s wallet before Q day arrives. He explicitly said this was “at least one possibility.” The risk extends beyond Satoshi’s wallet; many exchange addresses also lack basic protection.
Before seeing OpenAI demonstrate relevant capabilities, he conservatively estimated that Q day was about 2 years away. He now thinks the possibility should be part of decision-making over a 1- to 1.5-year horizon. If nearly 1 million BTC came under selling pressure, he believes the market has priced Q day at a very low probability.
The host’s key challenge was that if the BTC community can upgrade to quantum-resistant addresses, why can’t it neutralize the threat in advance? Sun’s answer: the protocol “can only help people who are willing to help themselves.” An upgrade can give people who move their coins a new address, but it cannot act on behalf of Satoshi, people who have lost their keys, or the public that has not recognized the risk.
3. “Bunker mode” can protect individuals, but it cannot stop dead coins from coming back
For people who still want to hold BTC, Sun recommends moving coins to a new address that has never signed a transaction or been used before. The address should be used only for storage. After each spend, the remaining BTC should be moved again to another fresh address to reduce the exposure created when a public key is revealed.
This approach only protects an individual’s funds; it cannot deal with old coins whose owners do nothing. After years of Bitcoin’s operation, many private keys have been lost, and there are destroyed and dead coins. Sun cited a friend who lost 5,000 BTC and said similar assets could become exposed again after Q day: “all kinds of ghosts and demons are coming back to life.”
The community could discuss freezing early blocks, restricting the circulation of those BTC, or implementing a hard fork. But there is no KYC evidence to prove that these addresses belong to Satoshi, making the boundaries contentious. More fundamentally, Bitcoin has never handled assets this way before; miners, developers, and the community would struggle to reach consensus in such a decentralized system.
His extreme supply-shock scenario is that “up to nearly 2 million or more” BTC could become active again. The host concluded that holders should therefore cut their positions. Sun agreed that the logic held, but chose to move into “bunker mode” first and monitor the technology community rather than sell immediately.
4. The quantum threat could turn a crypto problem into an internet infrastructure problem
The host pointed out that cold-storing BTC barely affects returns because BTCFi is still immature. The same approach with ETH would prevent users from continuing to participate in DeFi and materially reduce yield. Sun accepted the trade-off, describing it as a challenge for the entire crypto industry and even cryptography as a whole.
His scenario extends well beyond crypto: website certificates, banking systems, AI services, and internet identity verification all rely on elliptic-curve or other classical cryptographic algorithms. He explained the basis of asymmetric cryptography with the example that it is easy to multiply two large prime numbers but hard to reverse the calculation and factor the result. If that asymmetry is broken, China Merchants Bank, Bank of America, and even Nasdaq could be affected.
This would be a “problem for everyone,” on a scale far larger than an incident on a single chain. He drew a comparison with Y2K, while stressing that the impact need not be as severe. Tron and the exchanges have an extreme-case contingency plan: back up pre-crisis data in advance and, if needed, reach “consensus to pause” until a new solution is ready, then restore the chain and trading systems.
Asked about hedging, he answered directly: “I don’t have one yet.” For now, the only steps are to move assets to safer locations and invest in post-quantum algorithms. He first said Tron’s quantum-resistant version was “running onchain,” then clarified that Tron and Huobi’s algorithms are currently on testnet and not yet deployed to mainnet, with launch planned for about a year before the Q day he anticipates.
5. Fears about U.S. Treasury credit are overblown; the real anchor is U.S. control of AI
Sun gave a clear, unexplained call on the renminbi: “It will strengthen.” On Treasuries, he takes the opposite view of the market’s concerns. Their scale has grown so large that there is no directly comparable substitute on Earth; the question is no longer simply whether the U.S. can afford to repay them in a static sense.
Treasuries can “keep refilling the cup”: issue new debt to repay old debt, as long as buyers continue to take the new issuance. In his view, buyers are ultimately willing to do so because of U.S. dominance in global technology and the military, as well as its control over AI process chips, the memory supply chain, models, and technology.
The host pushed back that this dominance is weakening and that rising Treasury yields increasingly reflect declining buyer confidence. Sun acknowledged that yields are rising, but sees this more as Treasuries competing with AI equities for capital than as proof that the credit anchor has broken.
He summed up his view with a vivid image: “Like Sun Wukong, no matter how high he jumps, he can’t escape the Buddha’s palm.” As long as the U.S. retains ultimate control over AI, Treasuries will not lack buyers. That is the condition behind his relatively sanguine view of a credit crisis, not a claim that Treasuries are unconditionally safe.
6. He still holds the Nasdaq, but crowded trades and the “mother hen constraint” make him slightly bearish over the medium term
His current allocation across major asset classes still includes ETH, BTC, and top-10 tokens with applications; on the U.S. equity side, he remains invested in Nvidia and the Nasdaq. He describes the Nasdaq position as large, but his stance has shifted to “slightly bearish” over the medium term. He still holds the position and has not shorted it.
The central constraint is the source of funding: “Treasuries are the mother hen.” Equities can use leverage, but cannot detach indefinitely from the bond market. When AI draws capital away, the Treasury market needs to offer higher yields to finance itself. Those higher risk-free returns then pull capital back out of stocks and raise the cost of the entire system.
The host framed the extreme case as a 7% risk-free yield and asked, “Why would anyone bother with equities?” Sun agreed and added that circular financing within the AI industry is also severe: capital, chip demand, and valuations reinforce one another, so “if something goes wrong, everyone goes down with it.”
He does not reject the long-term demand story. As with the internet boom in the 1990s and around 2000, many fiber-optic companies went bankrupt, but the fiber they left behind was eventually put to use. The large amount of chips and capacity being built today may also be absorbed over time, though a near-term glut cannot be ruled out given the “immense” supply.
He has no A-share positions, though he does some research. He noted that optical modules were once a popular A-share trade because they benefited from U.S. AI while also being made in China, where the sector already had some presence.
7. ETH was the offensive position, BTC the core holding; last year’s double top triggered an 80% to 90% cut
Sun says his team cut its ETH position around $3,600–$4,000, selling at least 80% to 90%, and has not bought back in. The trade was based on a clear chart setup: ETH’s first attempt to break through about $4,800 failed; it later rose again on ETF demand and narratives from people such as Tom Lee, forming a classic double top.
In his view, the second rally depended on Tom Lee continuing to fuel it, leaving the structure “very fragile” and looking like it was “running out of steam.” Because ETH was their largest position at the time, they proactively cut most of it while keeping a small core holding.
They handled BTC differently and did not reduce it then, because BTC was viewed as a core position. He compared it to Nvidia: as long as they had not given up on AI entirely, they would keep a core position. ETH, by contrast, served an offensive role and could be increased or reduced substantially based on the setup.
The capital freed up by the ETH sale was not left idle for long; it went into a large amount of U.S. equities because he was still confident in the crowding into U.S. stocks at the time. His view has since changed again: the crowded trade is showing signs of fragility, but his position has not yet caught up with that shift.
8. AI’s next phase is less about hoarding chips and more about applications splitting out by data use case
Sun’s clearest preference in the AI value chain is applications. He thinks that once Nvidia is worth about $5 trillion, the ceiling for a single company is probably no more than half the size of the Treasury market. When the host extrapolated that Nvidia might rise at most another 4x, Sun replied that there are many areas in the application layer that could grow more than 4x.
He thinks memory is overvalued, and also mentioned optical modules and other areas. That is not a denial that there will “always be a shortage of memory”; rather, he believes capital is crowding into higher-certainty areas because investors cannot see applications beyond large models. Software capital is flowing into Anthropic and OpenAI, while hardware capital is flowing into Nvidia and chips. The result: “a lot of hardware stocks are actually being valued like software companies.”
He expects the theme to start shifting as soon as next year or the year after, from a handful of software and hardware leaders to applications differentiated by data-processing use case. He cited Elon Musk’s issuance of about $40B in debt to buy chips as an example of the consensus that “you can’t go wrong buying chips.” But the strength of that consensus is precisely what makes valuations fragile today.
AI-powered health management is his favorite concrete model. Consumer devices such as Oura Ring handle day-to-day data collection; hospitals conduct specialized tests such as bloodwork about 4 times a year; AI mines the combined data and recommends decisions, with experts providing a second opinion. Healthcare could shift from “going to the doctor only when you’re sick” to continuous management, with insurers potentially adjusting as well. He thinks this could achieve more with less spending.
9. Tron wants to evolve from “Alipay onchain” into an asset gateway for non-dollar countries
Sun says the RWA opportunity is “enormous” and that Tron plans to bring most U.S. equities onchain in November. Stablecoins’ first value proposition is giving people in non-dollar countries exposure to the dollar. The next step is to offer access to the highest-quality, most widely accepted assets in the dollar system, giving people who previously could not access them a way to invest.
The host summed up the product roadmap directly: users will be able to hold more than dollars and buy QQQ and the S&P 500 in the Tron ecosystem. Sun agreed. This would add value beyond transfers and move Tron from a payments network toward a cross-border asset platform.
Asked about transaction fees rising from about $1 in the early days to about $5–$6, while new chains such as Plasma promote free transfers, he acknowledged that Tron is more expensive than before. He said fees were cut by about 60% last year, and users can also transfer for free by holding and staking TRX to obtain energy.
He calls transfers the first stage of “building Alipay” and says more than $100B is now parked on Tron. If other chains are growing slowly and cannot match Tron’s transaction volume or scale, he will not push fees down to rock-bottom levels. Users trust the brand, network, and security, and are willing to pay a premium. Fees could come down further if RWA becomes more successful.
10. “Old-guard assets” are being left behind; reputation, death, and life are seen as dynamic experiences
Sun draws a line between “old-guard assets” and “new-guard assets”: he is not bullish on old-guard assets such as baijiu, while actively embracing new-guard assets. It is his way of describing a shift in eras—capital should move closer to technology and new demand rather than rely on old narratives.
Asked about his biggest failure over the past year, he initially dodged the question, then offered only “failure in love” and “emotional failure.” He immediately added that while it might be a failure for the relationship, it “isn’t necessarily a failure for life”; it could also become a source of wealth.
On reputation, he insists that “it still matters,” but does not see public opinion as a permanent verdict. The source of one’s reputation changes across life stages. People should understand public sentiment and engage with it constructively, letting disagreements continue to play out through discussion, because “I have never believed that opinions are eternal.”
He admits that what he lacks most is an understanding of AI and how to apply it. His ultimate ambition is to integrate AI into his existing businesses and find a place for himself and his company in the new era. On death, he says he is “afraid of dying” and uses AI to monitor his health data. The goal is not to extend life indefinitely, but to make the most of it within a reasonable lifespan and “die with dignity and grace.”
11. “Sun studies” is framed as zero-cost bootstrap; the Sun Yuchen Prize must keep evolving for the AI era
If young people do not want to study “Sun studies,” his second recommendation is to take up combat sports first. For those who do, he says the system is “pretty perfect” and can be “distilled and installed at zero cost,” like an AI skill. It is aimed at ordinary young people, not those already born with a silver spoon who can develop their own playbook.
The host’s objection is worth preserving: “Sun studies” could be excessively self-interested and push society toward polished self-interest. Sun rejects the equivalence. He says the first step is to understand the conditions of the era and one’s own circumstances, then choose a strategic path that maximizes personal interests. It is a form of game theory and evolutionary theory, designed to help someone bootstrap from zero to one.
He points to his 2016 advice as an example of “going with the wheels of history”: young people should not get married or buy homes, but go all in on Nvidia and Tesla. He says those who followed that advice will see their wealth grow by orders of magnitude over 10 years. His recommended reading follows the same intellectual lineage: Ayn Rand’s The Fountainhead, Wang Xiaobo’s Golden Age, and his own Sun Studies.
The Sun Yuchen Prize currently receives only a small fraction of his wealth because he wants to “look at the cards first,” like in Texas Hold’em. He imagines that AI may produce 50 Nobel Prize–worthy results in a single day in the future, so the prize must keep evolving: anyone who solves a problem with AI and passes formal verification in Lean should be eligible to win. The prize’s first batch of winners has already been selected, but OpenAI has not yet come to collect its award.