I’ll start by saying IREN is a second-tier compute seller, showing subpar execution both on construction and sales.
However it’s also why it’s interesting.
The penalty markets are putting on IREN and other second-tier compute sellers is simply too high given the assets they sit on.
Let’s unpack this.
Nvidia has given us forecasts that show demand growing 80% next year, showing AI infrastructure builds outs are accelerating.
This is confirmed by hyperscaler CAPEX commitments and pretty much all ER across the supply chain.
Another key component of building out AI infrastructure is powering it, which is why you need land connected to the grid or at least with permission to connect.
Goldman Sachs estimates the current wait time to get on the grid at 4 years for a data center project (just for permits).
Compute sellers have reported deal size per MW has recently doubled (although chip prices also increased) leading to higher value for powered land.
Essentially, those with powered land are in the unique position to spend $40-50 million per mW and earn $20m per MW per year for 5 years, then spend another $25m per mw to get another $20m per MW for another 5 (these will both need to be adjusted for inflation in AI space, margin will likely remain similar).
Given IREN’s power access, it can invest $160-200 billion to generate $320-350 billion over 5 years and then another $100 billion to get another $320 billion in years 6-10, for a potential total of $380b over 10 years.
Now, of course, IREN has only 10% of that money locked in so it currently can guide only after 10% of that upside (\~38b over 10 years).
Then you need to take out the cost of debt and you need to apply a DCF valuation to the 10 year cash flow to get that $38b over 10y to be worth around $17-20 billion today.
This is in fact lower than IREN’s valuation today. However, even if you discount and stress test the DCF math, you can end up with it being what it’s worth today.
This means the 80% unpowered land IREN has is currently free in this valuation.
Of course, IREN investors won’t capture 100% of the upside here, as new investors will need to come in and fund the capex build out in exchange for a piece of the profits.
Expecting an automatic 5x is therefore not accurate (however that is what Nebius is doing almost currently).
It is highly plausible however that IREN will get to a trading range of 70-100 per share as this build out continues.