$150 million, maybe more, walked out of Coldcard wallets and the theft rate is falling now, which is the part that keeps me up. Galaxy's read is that the lull means the vulnerable holders either moved their coins or already got emptied. Both of those are endings. Neither is a rescue. I've watched enough of these to know that when a bleed stops, it usually stops because there's nothing left in the vein, not because someone tied a tourniquet. Hardware wallets were the thing we told each other was the safe answer, the boring answer, the answer for the friend who kept coins on Mt. Gox and learned. 🩸

The same days, Calle and the Bitcoin Red Team are pointing Chinese models, Moonshot's Kimi K3 among them, at Bitcoin's open-source software and finding bugs. I sat with that longer than I expected to. Not because of the China angle, which will get overplayed by the usual suspects, but because of what it implies about timing. If a red team can rent that capability, so can the other side, and the other side doesn't publish. The Coldcard losses and the Red Team's bug harvest feel like two readings from the same instrument. Cheap, tireless code review just arrived, and it arrives for attackers first because they have a faster path to monetizing a finding. We spent a decade arguing about whether Bitcoin's security budget survives the halvings. Nobody I follow was arguing about whether the software around it survives machines that read faster than we do.

Then Tether sat for a real audit and apparently passed. I keep turning that one over. For years the standard reply to any Tether question was an attestation, a photograph of the pile on a chosen day, and the honest criticism was never "the money isn't there" so much as "you're asking me to trust a snapshot." An audit covers the whole period, samples the evidence, and ends with a firm putting its license behind an opinion. What actually changed isn't the reserves. It's who bleeds if the reserves are wrong. That's the whole thing, and it's the kind of shift that looks boring for two years and then decides which stablecoins are still standing. Every issuer still living on attestations alone now has a question they can't answer with a PDF.

World Liberty picked up a conditional bank charter and will take USD1 issuance over from BitGo. Set aside who's attached to it, which I know is a lot to set aside. Structurally it's the same movement: issuance migrating from a crypto custodian into a chartered trust. The rails are being absorbed into the thing they were built to route around. That's not a betrayal, that's just what happens when something works. What makes me uneasy is the pairing, because in the same stretch JPMorgan's exit from Polymarket surfaced, a relationship they killed back in late 2025 over regulatory concerns. One venue gets a charter, another loses its bank. Same regulatory weather, opposite outcomes, and the difference between them looks a lot less like risk models than like proximity.

The Multicoin exit is the one I'd underline. Eight months after helping launch a $1.65 billion Solana treasury company, they're out, and the numbers underneath are ugly in a way I recognize: cash down to $4.5 million, $120 million owed to Galaxy, over half the reported SOL equivalents pledged. Pledged. That word did all the damage in 2022 and it's back wearing a nicer suit. A treasury company that has pledged 52.7% of its holdings isn't a holder, it's a leveraged position with a ticker. Celsius told us it was a yield business. Three Arrows told us it was a trading firm. The pattern isn't fraud, it's the slow slide from "we hold this" to "we hold this and borrowed against it and the collateral is the same asset the whole thesis depends on." When the sponsor who built the vehicle walks first, that's information. Sponsors know the cap table's real shape.

Against that, Strategy is staring down MSCI's non-operating-company screen and an estimated $2.8 billion of forced passive selling, answering with "Bitcoin doesn't need you." Bravado aside, the mechanical point stands and it cuts the other way from Multicoin's mess: index membership was always a borrowed legitimacy, and borrowed legitimacy can be recalled. Saylor built the original of this design and the imitators built worse copies with more debt. If MSCI does write that rule, the copies break first.

UBS showing a 24-fold jump in IBIT call options, with direct holdings up 12% to 407,890 shares and puts down 53%, will get read as a Swiss mega-bank going long. It isn't. A dealer's call book is usually a hedge against something it sold that never appears on the form, and the snapshot is from June 30 anyway. Every 13F season since the ETFs launched has turned dealer hedges into conviction headlines. I've stopped correcting anyone about it. The form shows what a firm holds, not what it believes, and the gap between those two is where a lot of my worst trades were born. 📉

Ethereum walking away from Poseidon after eight years, moving toward SHA and BLAKE because binary-field proofs flipped the math, deserves more attention than it'll get. No breakage, no panic, just an admission that the ground moved and the old bet no longer pays. That's rarer than it sounds. Most of us defend a position long past the point where the reason for it expired.

Kazakhstan pulling the plug on rigs in its capital year-round to save grid capacity is the same lesson with the wattage turned up: hashrate is a guest, and guests get asked to leave when the host gets cold. ⚡

The thread, if there is one: everything this week was about who carries the consequence. The audit moved it onto an auditor. The charter moved it onto a regulator. The pledge moved it onto Galaxy's loan book. The Coldcard holders were carrying it alone and found out. The rails didn't change much. The liability did, and liability always finds someone.

I keep coming back to Multicoin walking out the door eight months in. In 2021 the sponsors stayed until the very end because they believed. Now they leave early and file the paperwork. 🫥