Galaxy's First Full Helios Quarter: $79M in Lease Revenue and a $435K Loss. Both Numbers Matter.

Galaxy Digital just posted its first full quarter with Helios fully leased. The headline number is $79 million in lease revenue. The other number is a $435,000 net loss in the Data Centers segment. Both are real. And half of crypto Twitter is about to pick whichever one fits their existing thesis.

Let's do the honest version.

The story, without the spin

Helios is Galaxy's big bet. It's a Texas site that used to be a bitcoin mining footprint and has been reborn as an AI and high-performance computing data center. Galaxy's pitch has been simple for a while now: we're not a miner, we're a landlord for compute. That quarter one of the leases going fully live should be a victory lap.

It mostly was. Seventy-nine million dollars in lease revenue in a single quarter is a real business. That's not a vibes trade. That's contracts, power, racks, and a counterparty paying on time.

But the segment books a $435,000 net loss. And here's where the commentary splits into two camps that are both kind of wrong.

Camp one says the AI data center pivot is a bust. A quarter of $79 million and you still can't clear breakeven? Pack it up. Camp two says the loss doesn't matter because growth. Both takes skip the actual mechanics.

The $435,000 isn't a failure of the model. It's the cost of getting a brand new facility up the ramp. Depreciation on new buildouts is brutal in the early quarters. Staffing, power contracts, and the boring operational stuff all hit before the lease revenue fully seasons. A company can bring in $79 million and still show a small red number at the bottom of one segment while the total picture works. That's accounting, not collapse.

And the total picture, per the disclosure, is that Treasury and Corporate did the heavy lifting. That's the part nobody expected to be the hero of an AI data center quarter.

Why the Treasury line is the real story

Look, there's something quietly funny here. Galaxy spent the last two years pitching itself as a compute landlord. The pitch deck is all power purchase agreements and hyperscaler demand. And the segment that saved the quarter was the trading desk.

That's not a knock. That's a flex disguised as a footnote. Very few crypto-native firms have a Treasury and Corporate arm that can actually absorb a buildout quarter and keep the lights on. Most miners pivoting to AI don't have that cushion. they've a hash rate, a power contract, and a prayer. Galaxy has a balance sheet with multiple engines.

But let's ask the question nobody in the bull camp wants to ask. When does the AI data center line put the trading desk back in the passenger seat? Because right now, for all the $79 million of lease revenue, the recovery was carried by the book. That's a transitional quarter. It's not the final form.

My first hot take: anyone comparing Galaxy's Data Centers segment to a bitcoin miner's margin is doing the wrong math entirely. Miners get paid for block rewards that move every block. Landlords get paid on a lease schedule with a counterpary that signed a multi-year deal. Those are different risk profiles. Stop benchmarking them like they're the same asset class.

My second hot take: the $435,000 loss is bullish, and I mean that specifically. A first full quarter with that small a hole in the ground is a sign the facility is basically operating at cost during the ramp. If it were losing tens of millions, you'd have a commissioning problem. A sub-half-million loss on $79 million of gross is a rounding error that tells you the lights came on correctly.

So what actually happens next? The boring answer is the right one. Lease revenue compounds as more of Helios gets signed. Depreciation evens out. The Treasury desk stops needing to be the hero. But there's a real risk too. If the AI trade cools and counterparties get picky about compute contracts, Galaxy's first-mover advantage becomes a capex millstone. That's the bet.

Is a company that just proved it can lease $79 million of compute in a quarter actually priced like a landlord, or is it still priced like a trading firm that happens to own some buildings? That gap is where the entire investment thesis lives.

The takeaway

The Data Center pivot at Galaxy isn't a meme anymore. It's a revenue line. $79 million is real money, and the small net loss is the tax you pay for building the thing that generated it.

But the recovery being carried by Treasury and Corporate is the detail that separates Galaxy from the miners trying to copy this playbook. The pivot works because there's a balance sheet behind it. The timeline is undefeated, and it's about to spend a week arguing over a $435,000 number that doesn't mean what either camp thinks it means.

Watch the next quarter's lease revenue and the Data Centers segment margin together. If revenue climbs and that loss shrinks at the same time, the model is working exactly as sold. If revenue climbs and the loss widens, that's when the receipts actually matter.

We regret to inform you that the AI data center trade just got its first real scorecard. And it's a lot more boring, and a lot more interesting, than the pump threads would have you believe.