Bitcoin Reclaimed Its 50-Week Average After 20 Weeks Below It. Galaxy Says the Bottom's In
Galaxy Research opened its latest note with five blunt words: we believe the bottom is in. The 50-week reclaim is the headline. But the 13F rotation from hedge funds to sovereign wealth, and the AI capex debt story underneath it, are the parts that actually matter for where Bitcoin goes next.
Eighty-one thousand dollars is a strange place to call a bottom. Galaxy Research did it anyway. Their note opens with five words that leave no room for hedging: we believe the bottom is in.
The evidence isn't vibes. Bitcoin spent more than 20 weeks trading below its 50-week moving average, then climbed back above it. Alex Thorn, who runs research at Galaxy, went looking through history for a case where that reclaim was followed by a fresh cycle low. There isn't one.
Small sample size, sure. Bitcoin has only lived through a handful of cycles, and four of anything isn't a statistics department. But the signal earns attention for what happened around $80,000. The coin held. It held through a Fed decision, through a wave of AI-driven risk-off chatter, through the exact kind of macro noise that normally cracks a thin bid.
And the positioning is light. Thorn's read is simple: the crowd isn't here. Not in the euphoric, everyone's-a-genius way that marks a top.
Hedge Funds Out, Sovereigns In
The 13F filings are the part most people scrolled past. Every quarter, funds running more than $100 million disclose their long positions 45 days after the quarter closes. This round showed a clean handoff. Hedge funds trimmed. Sovereign wealth funds and wealth managers added.
Read that again. The fast money left. The slow money arrived.
That rotation changes who the marginal seller is. Hedge funds sell on momentum, on risk limits, on a Tuesday because the value-at-risk number got too big. Sovereigns and wealth managers don't operate that way. They buy in size, they buy slowly, and a 20% drawdown doesn't move them.
Here's my first take. The moving average is the story most people will write. The 13F rotation is the story that actually matters. Technicals describe where price has been. Holder composition tells you where it can go.
Worth adding one more layer. Corporate treasury vehicles are still out there hoovering up supply, and the daily dividend mechanics on those balance sheets are now part of the Bitcoin bid in a way they weren't two years ago. That's a structural buyer with a public reporting obligation, not a fund that can vanish in a quarter.
The AI Build-Out Is a Debasement Machine
This is where the two loudest stories in tech collide, and I don't think crypto has priced it properly yet.
The AI capex boom is enormous, and it isn't being funded out of cash flow. Data centers, power contracts, custom silicon, the whole stack. That's debt and dilution, financed into a rate environment already carrying a heavy sovereign load. Thorn's framing is blunt: the AI build-out functions as a debasement machine. Capital gets pulled toward compute, and the money to build the compute gets borrowed or printed.
Which is the entire Bitcoin thesis compressed into one sentence. Fiscal dominance doesn't need a crisis to show up. It just needs governments to keep issuing into a market that's simultaneously financing the largest industrial build-out in a generation.
So the AI trade and the Bitcoin trade aren't opposites. They're the same trade expressed on different ledgers. The compute layer needs a payment rail, and the balance sheet underneath the compute layer needs an asset that doesn't answer to a treasury department.
Second take, and this one is less comfortable. Tokenization is the most crowded low-conviction story in crypto right now. Everyone agrees it's coming. Almost nobody can name the timeline, the venue, or the chain that wins. Meanwhile the basis trade does the actual work while the narrative pendulum swings back and forth. Follow the flows, not the conference panels.
Who Holds the Keys
Now the part that should worry anyone running self-custody at scale.
AI is a cyber capability multiplier. Attack surfaces that used to require a patient human now get probed by systems that don't sleep, don't get bored, and can run a thousand variants of a phishing lure before lunch. Hardware wallets and multisig stopped being optional for serious holders a while ago. They're the baseline now.
But here's the tension nobody wants to sit with. We're being told agents are getting wallets. Machines will transact, settle, and hold value permissionlessly. Fine. If agents have wallets, who holds the keys? Because the honest answer today is either a human who can be socially engineered, or a key management system that's itself software and therefore attackable.
The AI-crypto Venn diagram is getting thicker. This isn't a partnership announcement. It's a convergence, and it forces a rebuild of custody assumptions from the ground up. The projects that win the next cycle will be the ones that solve key management for things that aren't people.
The Takeaway
Galaxy's call is contrarian right now, which is exactly why it deserves a hearing. Light positioning, a reclaimed trendline that's never failed, and a holder base that just swapped its most jumpy participants for its most patient ones. That's a reasonable floor argument.
But the reason to pay attention goes past $81,000. The AI build-out is going to force a global conversation about money, debt, settlement, and custody that Bitcoin has been having with itself for fifteen years. We're building the financial plumbing for machines, whether or not anyone planned it that way.
Somebody's going to own that layer. It probably won't look like a bank.