CryptoBadNewsss 🐐 (Ø,G) ⌘ CLONE (Ø,G) .ink

TeraWulf fell 15% last week, right alongside every neocloud. But TeraWulf isn't a neocloud, so why the selloff? Because the market sold everything with an AI infrastructure label first and asked questions later. TeraWulf owns the power, the land, and the buildings. It rents finished data halls to AI companies on long leases. The tenant buys the chips, runs them, and carries the risk if they lose value. That's a landlord model. TeraWulf gets paid rent whether AI chip prices rise or fall. So the whole business comes down to two numbers: what it costs to secure the power and build the hall, and what a tenant will pay to rent it. And tenants keep paying up. Hyperscalers, AI labs, and neoclouds need more capacity now than ever, and most of them can't build it themselves, or fast enough. Money doesn't fix that. Even with unlimited cash, you can't connect a giant site to the grid and build it in two years. That's the bull case for data center landlords. They sell grid connections on their own land, under their roofs. TeraWulf's CEO said tenants must be investment grade or bring backing from someone who is. In New York, that's Google. If Fluidstack stops paying, Google either pays a termination fee or takes over the lease. That backstop is about $3.2B in total, but it switches on hall by hall, and only about $600M is live today. For Kentucky, he says the backstop is committed. No filing names of who's providing it yet, so we're waiting for that. The other part of his pitch is "no surprises": the next hall this quarter, the one after in Q1 2027. One rent start has already slipped once, so I'm watching delivery most. Ultimately, this comes down to execution. My take: the landlord pitch holds up on structure, and the stock is trading like it doesn't. I love companies with predictable and durable cash flows. I own WULF in my portfolio. You can see the rest of my book here: link.milkroad.com/Martin

Galaxy earns 45% a year on its own money. TeraWulf earns 35%. And yet I own both. Here's what's going on. Strip it down to 1 MW, the power that actually reaches the chips (see the chart below): - Build cost: Galaxy ~$14M, TeraWulf ~$9-11M - Rent per year: Galaxy ~$2.5M, TeraWulf ~$1.9-2.4M - Yearly return on what it cost to build: Galaxy 15.6%, TeraWulf 17-18% - What the debt costs: Galaxy ~9.8%, TeraWulf ~8.7% TeraWulf builds cheaper, earns more on each building, and borrows at a lower rate. So how does Galaxy end up with 45%? Take a $100M building. Galaxy puts in about $16M of its own cash and borrows $84M. The building earns about $15.6M a year. Interest eats about $8.2M. About $7.4M is left, on $16M put in. That's about 45%. TeraWulf puts in $32M and borrows $68M. Its building earns about $17.3M, interest takes about $5.9M, and about $11.4M is left. On $32M, that's about 35%. Galaxy's edge isn't a higher return on the building. It just puts in less of its own money. (This isn't what the stock pays you. It's what each project earns before taxes, wear and tear, and loan repayments.) Why does Galaxy cost more? It built its Texas power setup almost from scratch. TeraWulf's New York site already had 500 MW of power from its bitcoin-mining days. Why does Galaxy pay more for debt? It borrows more, so lenders have less cushion. And part of TeraWulf's New York rent is guaranteed by Google. Colocation is a simpler, lower-risk game than neoclouds that live and die by GPU pricing. At today’s levels, a lot of these operators look set up for great returns. If the model resonates with you, you can lock in those MW-style economics at a compelling entry point right now. Follow me @m0xt_ for more updates or join Milk Road PRO to get my moves in real-time: link.milkroad.com/Martin