Compilation & Editing: Shenchao TechFlow

Show:Kyle Chasse Crypto (hosted by Kyle Chasse)

Guest:Arthur Hayes (former BitMEX founder, Maelstrom founder)

Duration:57 minutes Released: September 2, 2026

Assets:Bitcoin, Ethereum, Ethena (ENA), Hyperliquid (HYPE), Flop (Hayes' new project)

Disclosure:Hayes is the founder of Maelstrom, and the Ethereum and Ethena mentioned in the show are held by his fund; he is also the initiator of the Flop project (Flop Labs), and large portions of the text discuss the coins he issued. The opinions expressed are his personal directional bets, not investment advice.

Summary of Key Points

• Believes 58,000 is the cycle bottom: Bessemer reveals his hand to push the yen, U.S. Treasury buybacks expand, while AI absorbs marginal credit to its peak, entering a "capital mismatch stage" that benefits crypto.
• Ethereum is Maelstrom's largest directional position, reasoning being the layer one narrative of trading real-world assets, ease of explosive growth after not breaking previous highs, and the Lindy effect; holds more Bitcoin but heavily invested in ETH.
• Cleared out Hyperliquid, reasoning that "everyone knows about HYPE now, the asymmetry is gone," does not favor the route of compliance concessions.
• Bets Ethena will rise five times in two to three months: basis arbitrage on synthetic USD, early investors have redeemed and circulation has rebounded, 58,000 is the bottom.
• Returns only for his own Flop coin: builds an AI computing power payment network, betting "either it goes to zero or is worth more than Bitcoin"; there are multiple references in the text to his holdings and project endorsement, requiring a discount on interests.

1. Changes stance after ten weeks: from "local top" to "load the truck"

Kyle Chasse starts with a contrast: back in June on the show, Hayes was sitting opposite him saying "local top", just ten weeks later, Hayes returns to reverse, saying "back up the truck," buying everything at the bottom. He asks Hayes what exactly changed in these ten weeks.

Hayes says it's not the market that's changed, but the statements from Bessemer and the authorities. "Basically, he revealed his hand; they'll intervene in the yen." The USD/JPY is still around 160; Japan has intervened with hundreds of billions of dollars but returns to square one, but both sides have agreed to weaken the dollar and support the yen. Hayes believes that to prevent Japan from selling off affecting the U.S. market, the U.S. will print a lot of money. Immediately following, the U.S. Treasury buyback scale was raised in early August, and the direction is clear: expand the balance sheet, give the market more easing, so that the U.S. government can repay its debts.

He says Bitcoin surged after the buyback announcement because "a large part was those shorting gamma and selling call options got squeezed," that surge was the result, and is now being digested. But he still believes 58,000 is "very likely the local bottom, after some fluctuations it will continue upwards."

2. 58,000 is the cycle bottom, AI absorbs marginal credit

Chasse follows up: is 58,000 a local bottom or a cycle bottom?

Hayes answers straightforwardly: "I believe it is the cycle bottom." He mentions the wave of about 125,000 last October, only slightly higher than previous highs, is the weakest bull market in crypto history. The reason he states bluntly: "AI has absorbed all marginal credit." This momentum will peak in 2025 because the capital expenditures poured into AI and the demand on the credit market will be the story from 2024 to early 2026. He predicts this second derivative will turn negative by the end of 2027, meaning the phase where AI takes away all marginal credit is over, entering what he calls "the overall capital mismatch phase in AI": people who once supported AI can't admit they were wrong, projects that aren't profitable still falsely report income growth, "we can't admit failure, so we can only continue to bet, covering up past foolish decisions."

He argues that this is a tremendous boon for crypto. Super-sized cloud vendors borrowing money without regard for cost, the U.S. government, fearing competition from Google, must first ensure they can secure funds. "Politicians have already said we are all in on AI, both the U.S. and China believe this, regardless of what it actually means." The expansion of AI credit, circular financing, and various accounting games "all benefit crypto, as politicians take AI as a shortcut, to cover up the money printing and waste we've seen over the decades in their terms." Money must have a destination, and he believes it will go to crypto. "Earlier this year, we crossed that point, at that 58,000 position, combined with all these accounting games and the chaos surrounding AI, and the panic over long-term interest rates all point to the same thing. So I believe Bitcoin has bottomed, and it will be much higher from here."

3. Macro or four-year cycle

Chasse continues: do you think Bitcoin and crypto are driven by macro factors or is it the four-year cycle turning by itself? Hayes says he has written research on this, the reason the four-year cycle aligns with the credit cycle is that Bitcoin was born at the start of U.S. unlimited quantitative easing in 2009, then following the ups and downs of Chinese and U.S. monetary policy, coincidentally around the halving time. "I still believe Bitcoin is the purest way to bet on the expansion of liquidity by central banks and commercial banks. But I won't dogmatically cling to the four-year cycle; I look at the macro, look at liquidity, trade according to what the market gives, and not be bound by dogma." It may be a four-year cycle, it may not be; he doesn't want to think that way. His argument is that "everything depends on liquidity," and he trades whatever the market provides.

4. Ethereum is the number one heavy position

When discussing Ethereum and Bitcoin, Chasse asks: do you think ETH will outperform BTC in the coming years? Hayes responds: "Yes, and by a large margin." He says Ethereum is now Maelstrom's "largest directional position." Chasse then asks about the ratio of Bitcoin to Ethereum in Maelstrom; Hayes laughs: "The numbers are very large, much more Bitcoin; we hold a lot of Bitcoin, I don't even know the specifics, it doesn't matter. It's just that Ethereum is our largest directional bet."

He explains why it is ETH and not something else. The narrative this round is "layer one for trading real-world assets," Robinhood's chain is built on Ethereum (he understands that it's an Arbitrum fork, with very little revenue actually flowing back to Ethereum's mainnet, but the narrative is "look, Robinhood is making stock and meme coin trades on it"). He argues that if you are the IT guy in a big company and your boss asks you to go on-chain, Robinhood is the S&P 500, the fastest-growing retail broker in America, they use Ethereum; would you use Sui or Solana? "You would use what Robinhood is using because they did it first, you can't explain otherwise. That's just trading, plain and simple." Moreover, Ethereum's price hasn't broken its 2021 all-time high, it is the only major coin that hasn't made a new high, "the Ethereum Foundation is just a bunch of puppets, there are tons of jokes online about how gay it is, but it’s precisely this setup that is easiest to explode. It has survived long enough, there's the Lindy effect, I’m not worried about it dropping 75% overnight, so I can comfortably hold this position, and I indeed have."

5. Cleared out Hyperliquid

As for Hyperliquid, he just cleared out. Early on he called for HYPE to rise thousands of times, held it up to 40 or 50 bucks, and later wrote that the market should start worrying about team selling pressure and low-fee competitors (like Aster, Lighter) compressing profits, which was pretty much the local top. When it fell to the 20s, he felt it was an opportunity, adding to his position to make it one of his largest, climbing to around 75 before he balanced his books to only Bitcoin before the Korean semiconductor crash brought down the market. Now that HYPE is back to 85, he is no longer on board. He says Hyperliquid is one of the best projects in crypto, the exchange is profitable, and 85 bucks makes sense, the problem is "everyone knows that everyone knows about HYPE, there is no asymmetry left in this trade." What's trickier is it has to cooperate with Kraken to enter the U.S., meaning both the clearing and margin models must change to cater to compliance, "that's not an exciting story about making more money, that's saying I'm going to make concessions to fit this market, and then compete with CME, Nasdaq, CBOE, and a bunch of similar products." He still holds a long-term optimistic view that HYPE can win, but "is it a ten-fold coin? Maybe, but I don't think so," so for his trading style, the asymmetry is gone.

6. Betting on Ethena to rise five times

He switched to talking about Ethena. Simply put, Ethena engages in basis arbitrage, going long on spot, short on perpetuals, packaging the returns into a synthetic USD stablecoin, staking which can yield basis profits, sometimes as high as 20% or 30%. Its stablecoin circulation surged to 15.5 billion during times of high basis, but later prices fell, volatility disappeared, and people asked, "Why should I bear the smart contract and counterparty risk, just to earn that half a point?" The circulation dropped to three to four billion, and the token fell 90%. Additionally, early investors wanting to exit continuously sold, crushing it for a year and a half.

But recently, things have changed: Ethena has repaid all those investors, the token no longer has a lock-up period, "finally considering profit sharing for token holders." Hayes says these are structural changes favorable for the token, and combined with 58,000 being the bottom, Bitcoin's volatility and price moving upwards, basis returns will rebound, everyone is more motivated to hold and stake, so he entered the market a few months ago. "Now this token is down 99% from its historical peak, yet it is one of the best projects in crypto, with product-market fit. That's why I believe it can rise five times in the next two to three months; the market cap is small enough that five times is not difficult." He calculated that returning to the historical peak is ten times from here. "I'm not looking for this kind of trade in HYPE, I'm looking for trades like Ethena."

Chasse asked a tough question: if you were broke and someone gave you $500,000 to invest in only one coin, would you choose Ethena, ETH, Bitcoin, or something else? Hayes replied: "If it's only $500,000 and I can only do one thing, it's most likely ETH." The reason is he has more confidence in ETH "not going to zero one morning." No matter how good the Ethena team is, the smart contract and counterparty risks are real and embedded in the structure, "I wouldn't put my entire fortune on that trade." He says at Maelstrom he likes trades like Ethena because it can produce nominal returns similar to HYPE without putting a ton of principal at risk, while HYPE must amplify its position five to ten times to produce the same returns, "but all of these have the same smart contract risks; you never know until one morning you wake up to find a bug, and it’s down 75%, by then it’s too late."

7. Returns only for Flop: a payment network for AI's computing power

Hayes says he is "coming out of retirement" for a coin called Flop. Chasse asks why a coin is worth his return. Hayes says the opportunity arose because he has always criticized these AI companies, one day he challenged all users of AI: "You tell me what a token is worth, what is it really?" "We pay X dollars for a million tokens; what's a token really? Don’t just tell me it’s information; what is it?" He discovered that each company has a different definition of tokens, different charges, and is extremely opaque, with no standard. Looking deeper, what runs the models is simply a computer doing floating point calculations, and how many floating point operations per second is what is actually traded.

He asked: Is there any market with any currency that can help me understand "the global unified price of computing power"?

None.

Thus, he followed the belief that "AI agents need a payment network, and this network will be enormous," thinking: people go to work for fiat currencybecause they need to eat and can immediately exchange money for calories; agents need computing power, which is floating point operations per second, "give me that kind of currency, I don't care which one, I want to go to that decentralized place, directly exchange the coins in my payment network for what I need to survive, which is floating point operations."

So he built a floating point computation commodity market, modeled after Bitcoin: the network issues its own native coin, people do "useful work proofs" to mine and process floating point operations, the coin is called flop, allowing this network to become the payment network for AI. Agents use flop because they can directly exchange it for the necessary computing power for their existence.

He says this is an extremely asymmetric bet: "Either the flop network completely fails, or we’re worth more than Bitcoin, there are only these two outcomes." He believes there will be a large number of AI agents working, needing a payment network, "I believe it will be something like flop, not USDC."

He explains a lot about the specific mechanisms. Miners plug idle computing power into the spot market, initially because of subsidies, the hourly income from GPU can be eight to ten times higher than rental income, just like back in 2012 when you could mine Bitcoin with a laptop and earn a lot. They also want to build "Technocore," an AI agent chat room, issuing identities to agents along with business functionality, "in our test net, you work, and we’ll give you some flop to spend on computing power, to get you used to using it." An airdrop will release 20% of the supply. Hayes emphasizes they "don’t do presales; once launched, the only way to get new flop is to buy it from miners who receive block rewards," miners can sell however they want, "we don’t care." The test net will launch by the end of October this year, in a 90-day competition, with an airdrop on the mainnet by the end of the year.

Chasse asks how ordinary people can participate; Hayes says in two ways: provide computing power or use reasoning. He repeatedly states that this is "not for people, it’s for agents"; ordinary users won’t even know flop exists, their orchestrating agents will say "go do this," at the base either through OpenRouter or the flop network, and flop will be cheap, almost free due to miner subsidies, so agents will naturally choose it. He gives a wild valuation ceiling: "2 to the power of n, where n is the number of agents," with tens of billions to trillions of agents running on a payment network, the explosive value lies in this upper limit, "we obviously can't reach it, but hope to land in a higher range than the metastable law (the one describing Bitcoin and social media)." He encourages speculation on flop, betting there will be billions of agents needing to hold a bit of flop for business, "just like having a bank balance, you save it, you won’t spend every cent of fiat currency on food, that will create upward pressure on prices."

Chasse finally asks if there’s anything else to add; Hayes says no. Chasse wraps up: "I can see why you came back, that idea makes a lot of sense, being worth more than Bitcoin or going to zero, I understand."