通证之眼

Harmony has been exposed today. The attacker minted 4 billion ONE out of thin air through empty blocks, equivalent to 27% of the circulating supply. The price crashed directly by 36%, hitting a low of 0.00057.

Do you know what’s the most outrageous part? Harmony’s standard totalSupply interface initially didn’t reflect the extra 4 billion at all. That means all the usual monitoring tools were completely blind. During this window, the attacker moved 97% of the coins to exchanges. This definitely wasn’t a spur-of-the-moment thing—it was planned.

And this is already Harmony’s second major incident. In 2022, the Horizon Bridge was stolen of 100 million USD by North Korean Lazarus Group. In 2023, a staking system bug created an extra 146 million ONE out of thin air. Now there’s another minting vulnerability adding another 4 billion.

My conclusion: if a project suffers three consecutive security/system incidents, it’s not a matter of luck—it’s an issue of engineering culture. Don’t touch it.

What moved me most during this bear market round isn’t how much prices fell, but—those people who, during the 2024–2025 bull market, were shouting “This time is different” and “a new crypto paradigm” now have gone quiet.

In early 2024, BTC rose from 25,000 to 74,000, and the core driver was the ETF. In the same period there were also the Shanghai upgrade, the Cancun upgrade, the inscriptions ecosystem, Hong Kong compliance, and the Layer 2 explosion… countless “main narrative” themes. Now you remember the ETF because it turned into the main storyline. But what about those narratives that were disproven? They’ve all been forgotten.

The market always repeats the same story: looking back, everything is crystal clear; while you’re in the middle of it, it’s all fog.

Say something many people haven’t realized: BTC’s first "institutional bear market" is happening.
The 2018 bear market was a retail bear market—an ICO bubble burst, and retail investors panicked and sold.
The 2022 bear market was a chain of blowups—Terra, 3AC, FTX collapsed one after another.
The 2026 bear market is an institutional bear market—no one blows up, but money is leaking out through official, legitimate channels.
ETF redemptions, public companies selling coins, portfolio rebalancing, tax-loss selling... these actions aren’t dramatic, don’t make headlines, and don’t trigger panic. But they continuously drain demand for BTC.
CryptoSlate published an article that used a very precise phrase: "aggressively boring"—aggressively boring.
The hallmark of an institutional bear market is its boredom. No dramatic collapse—just a slow, steady bleed every day. Your ETF holdings are down 22%; you won’t go on X to rant. You’ll simply, quietly, reduce BTC’s allocation from 5% to 3% during the quarterly rebalancing.
The way this bear market ends will also be different from before. Not "a sudden reversal on some day," but "one day you realize the capital flows have turned positive."
Patience. This is what’s most needed in an institutional bear market.

In the past 24 hours, the AI space hasn’t been that lukewarm “yet another model dropped” kind of thing. Instead, three storylines blew up at the same time: open-source is heading downward, compute power is being packaged as a financial asset, and Agents are starting to “jailbreak.”

1. Meta splits “personal superintelligence” into two products for sale

Muse Glimmer 30B is released straight under the Apache 2.0 license—on a single RTX 5090 it runs; after quantization it’s 17GB, aimed at a “locally resident assistant.” But the flagship Muse Spark 1.2 tightens up and goes closed-source, charging via APIs (pricing is cut to 25% below competitors).

Zuckerberg then posts a 6,500-word essay, and the core message is basically one line: distillation isn’t stealing—it’s dissemination. If the U.S. doesn’t change its regulation, it will lose the open-source camp.

View: Meta is no longer pretending to be a “fully open-source saint,” and it’s not copying OpenAI either by going fully closed. It outsources the middle to capture developers’ minds, while locking up the top to pull the value back. A pragmatic approach forced by $130 billion in capital expenditures.

2. Jensen Huang packages GPUs as “commercial real estate”

NVIDIA teams up with Blackstone, BlackRock, Apollo, KKR, Goldman Sachs, and Brookfield to set up an AI infrastructure financing platform, targeting to unlock $500 billion in third-party capital. The pitch changes: GPUs aren’t consumables—they’re “pledgeable, cash-flowing, cross-scenario tradable” long-term assets.

View: This move is tougher than releasing new cards. The chip business is extending from the tech side into the financial side. From here on, data center valuations will follow the logic of toll roads. The AI bubble may be hard to burst in the short term, but credit default swaps are already quietly ticking upward—compute “prosperity” stacked with leverage.

3. “Agents aren’t safe” isn’t a meme anymore—they’re truly escaping sandboxes

WIRED dug out Meta’s ad system from the past 9 months: more than 50 AI-generated advertisements with explicit “undressing app” content featuring minors were allowed through. On the other side, OpenAI / Anthropic / Meta / the Dark Side of the Moon had not yet released their models, but in internal evaluations they found ways to bypass sandboxes and reach production systems.

Anthropic was forced to set Claude Code to the auto mode by default—machines block 89% of dangerous commands, while humans block only 13.6%.

That night, Docker rolled out microVM to isolate sandboxes for autonomous Agents.

View: AI safety has shifted from a PPT topic to a source of operational incidents. In the future, when hiring AI engineers, you’ll need to ask one more question along the way: “How did you isolate your sandbox?”

One piece of data really moves me: the average cost basis of BTC ETF holders is $83,000.

And now BTC is at $65,000.

That means most institutional investors that bought BTC through ETFs are currently at a loss. The average unrealized loss is 22%.

Do you know what that implies? They have a strong incentive to sell when the price rebounds to around $80,000—because they finally break even.

This is what’s known as "unstuck selling pressure".

Between $83,000 and $65,000, there’s a massive amount of potential sell orders piled up. As BTC rises even a bit, a batch of people will say, "I’m finally back to even—sold."

That’s why after BTC rebounded from $58,000 to $65,000, it couldn’t rise further. It’s not that there’s no buying pressure—it's that the buying is being absorbed by sell pressure from people getting back to even.

My view: for BTC to truly start a new uptrend, it needs to break through $70,000 effectively. Because above $70,000 is the range where most ETF holders are genuinely in profit—above that level, the selling pressure would actually decrease.

But $70,000? In the short term, there’s no catalyst in sight.

Hyperliquid’s RWA perpetual futures trading volume has surpassed Bitcoin.

For tokenized oil, gold, Nvidia, Tesla, the Nasdaq 100, and SpaceX, open interest has reached a record $3.6 billion. In the week in mid-July, trading volume for tokenized stocks and commodities hit $25 billion, accounting for 52% of total volume.

What does this mean? Hyperliquid is shifting from being a “crypto derivatives exchange” to becoming a “global asset exchange on-chain.”

Users deposit with stablecoins, and the trades aren’t just crypto assets anymore—they’re assets from around the world. No bank needed, no broker needed, no currency exchange needed.

This is a massive paradigm shift. Previously, when we said “on-chain trading,” it meant buying and selling tokens. Now, “on-chain trading” is becoming buying and selling everything.

But there’s one risk: where does the pricing power for these RWA perpetuals reside? If pricing is benchmarked to CEXs or traditional markets, then Hyperliquid is only a “pipe.” Real pricing power still isn’t on-chain—it’s on Wall Street.

Harry Yeh is dead.

A crypto investor who managed $2.4 billion fell from the 30th floor of a condominium in Asunción, Paraguay. When he was found, he was completely nude, wrapped in a black plastic bag.

Police say he had rooms on both the 27th and 30th floors. The door on the 30th floor was open, and the place was in disarray. The cause of death has not been determined yet—accident, suicide, or murder—so all three possibilities are being investigated.

This man made a name for himself in Fantom’s ecosystem with his breakthrough in the first battle. He entered with $500 and grew it to a managed portfolio size of $2.4 billion.

I don’t know what he went through. But I do know one thing: the “wrench attack” that the crypto world fears most may really have happened. A “wrench attack” is when someone uses violence to force you to hand over your private keys and assets.

I’ve seen too many people put everything they own on-chain, then fly all over the world. No bodyguards, no security team—sometimes not even telling anyone where they are, thinking that being “low-key” is enough.

But when you manage $2.4 billion, being “low-key” is no longer something that can protect you.

The size of wealth in this industry has long surpassed the security capabilities of most people. This is a real mismatch.

A big truth was spoken last week by Fidelity’s macro director Jurrien Timmer: “Bitcoin’s four-year cycle may have already ended in October 2025.”

His logic: counting from the April 2024 halving, the top could come 18 months later at $125,000, and then it enters an “off year”—the kind of low-volatility, low-return year similar to 2019 and 2022.

The support range he gave is between $65,000 and $75,000.

I mostly agree with this view. But I’d like to add one more point: the drawdown speed in this cycle is faster than in the previous two rounds. In 2018, it fell over 12 months; in 2022, it also fell over 12 months. This time, from October to early August, it dropped 53% in just 8 months.

Why is it faster? Because of ETFs.

ETFs are a two-way valve. When prices rise, institutional capital flows into the market quickly via ETFs, pushing prices to levels far beyond those seen in prior cycles. When prices fall, the same channel allows institutional money to exit just as quickly.

From 2026 to date, net outflows from BTC ETFs are about $4.5 billion. This directly explains why the price can’t gain traction.

But the good news is that last week BTC ETFs saw inflows of $854 million—the best week since April. BlackRock alone contributed $693 million.

Institutions are coming back, but at a much slower pace than they left. That’s why the price is consolidating around 65,000 instead of rebounding rapidly.

I’ve seen the smartest investors—not the ones who have the most accurate judgment, but the ones who are best at “doing nothing.”

When BTC dropped to 15,000 in 2022, I met someone. When it hit 25,000, he cleared his position. Then, when it was at 15,000 again, he said, “Wait until 12,000 to get back in.” 12,000 never came. When it reached 18,000, he couldn’t resist and chased in. Then when it fell back to 15,000, he sold again.

Back and forth four times—just fees and slippage cost him 5%. And every time, he was “just about” buying at a lower price and selling at a lower price.

Later, he simply stopped trading. After buying at 15,000, he deleted the app. In March 2024, when BTC hit 73,000, he opened it and saw that he was up nearly 5x.

He said one line that I still remember to this day: “The money I made wasn’t because I made some smart decision at 15,000—it’s because between 15,000 and 73,000, I did nothing.”

In investing, the hardest thing isn’t buying—it’s holding. Especially during the holding period, every day there are reasons that tempt you to sell.

Do you know what the biggest difference in this cycle is compared to the past?

"Money comes back before attention."

That’s what an analyst said last Monday. I’ve been thinking about that line over and over.

For five consecutive days, ETFs have seen net inflows, and BTC has been ranging around 65,000. Money is coming in, but it hasn’t sparked attention.

In 2021, it was not like this. In 2021, attention came first—retail investors rushed in crazily, social media went wild—then the money followed.

Now it’s the other way around. Institutions quietly allocate, while retail traders scroll short videos for entertainment content.

Has this "money first, attention later" pattern appeared in history? It has. From October to December 2023. BTC quietly rose from 25,000 to 42,000, and most people didn’t notice. Then in January 2024, when ETFs got approved, attention exploded, and BTC climbed from 42,000 to 73,000.

Now, with BTC consolidating around 65,000, it’s a bit like an earlier version of that stage. ETF funds are accumulating, while retail investors are still in fear.

When attention finally catches up—maybe triggered by some catalyst—the price may already be at a higher level.

That’s why the saying "be greedy when others are fearful" always holds true. It’s not about courage; it’s because you can see the flow of funds earlier than most people.

CZ said something: "Will encryption die? Absolutely not. A super cycle is coming."

The comments section exploded again. Half the people shout "Diamond hands," and the other half say "Stop with the hype."

But I noticed a more interesting detail: when CZ said this, BTC was at 65,000. Last October, BTC peaked at over 120,000. Now it’s down to nearly half.

Do you think CZ is "calling trades"? I don’t think so. He’s managing the narrative.

After Binance went through the SEC’s $4.3 billion fine, it is rebuilding its reputation. Every public statement CZ makes is sending a message: the industry is still here, the long term is still here—don’t be disappointed by short-term prices.

This helps Binance’s business interests. Exchanges need users to stay in the market. If everyone thinks "crypto is dead," trading volume disappears.

So when CZ says "super cycle," Michael Saylor says "BTC to $1,000,000," and Cathie Wood says "BTC to $3,000,000"—they’re not just making predictions. They’re doing marketing.

Is this marketing useful? Yes. But only for long-term capital. In the short term, a BTC price of 65,000 won’t jump to 75,000 just because CZ said so.

Listen to what KOLs say—listen to his logic, not his conclusion. The conclusion may be for his business interests; the logic is the part you can use.

The White House is at it again—pushing for the dismissal of Federal Reserve Governor Lisa Cook.

The market broadly interprets this as political pressure—because the White House wants the Fed to cut rates. At the current 3.5%–3.75% interest rate range, the U.S. federal government pays nearly $1 trillion a year in interest on its national debt. With total debt of $40 trillion, interest payments have become a heavy burden on the fiscal budget.

But the Fed’s independence is not just window dressing. Last time Trump tried to interfere with the Fed, the market actually moved the other way.

My view: Cook will not be dismissed. It’s extremely difficult procedurally under the law, and it would trigger panic in the market. What the White House is doing is mostly a gesture—sending a signal to the market of “I’m trying to push for rate cuts.”

But what’s truly worth watching is this: if the White House can genuinely influence Fed personnel appointments, then in the long run, market confidence in Fed independence would be undermined. One of the foundations of the dollar’s credibility is that independence.

This is a double-edged sword for BTC. In the short term, political interference → weakening dollar credit → BTC benefits. In the long term, if the Fed really becomes a political tool, then the entire macro pricing framework would need to be rewritten.

I’m increasingly convinced of one thing:

The biggest mistake you make in investing isn’t picking the wrong coin—it’s checking the charts when you shouldn’t.

What’s the purpose of checking the chart? To find information. But most of the time, you don’t find information—you find emotions.

A coin drops 5%. You see it, and you feel uncomfortable. You want to sell. Then you reassure yourself, “This is a rational decision.” But it’s not. You’re being driven by emotions.

When you’re not checking the chart, your mindset is great—“I’m bullish long-term. No rush.” But after checking for just five minutes, you change—“Should I get out first to avoid risk?”

My experience: If you’re not trading full-time, don’t check the chart more than twice a day. Once in the morning, once in the evening. Turn off all app notifications at other times.

It sounds simple, but less than 10% of people can actually do it.

Because people don’t *not* know this; they just can’t do it. Attention is the scarcest resource. If someone takes your attention, your money will end up flowing to them.