IREN Limited: Not Collateral Damage
IREN | Management Betrayal | Thesis Still Stands...
IREN Limited:
Verticality in The AI Compute Wars
Amid all the uncertainty already plaguing the Neoclouds like: GPU depreciation, competition and the risk of overexpansion. IREN's management apparently decided it was the perfect time to award themselves billions in dilution...
Everything I touch just decomposes somehow. But it gets worse. META 0.00%↑ who’s been going all-in on AI without a clear goal is now considering selling some of their cloud compute infrastructure to others.
This rumor sent literal shockwaves through CoreWeave (down 11%) and Nebius (down 12%), dragging IREN further down along with them. But despite the shitshow I’m still optimistic.
Trade is Still About Infrastructure
The ones innovating at the top aren’t the ones awarded high multiples, the ones supplying the infrastructure are. This can’t go on forever, but it most likely will for a little while longer.
My concern isn’t actually about GPU depreciation too much, the AI labs will need chips and a reliable power source connected to the grid. However I am worried about the chips potentially becoming exponentially better undermining the need for physical space.
If Blackwells, Rubins or whatever next-gen thing are exponentially more efficient per square foot, why do we need gigawatt pipelines of real estate? The current AI architecture saves us for now as both training and inference are compute heavy.
Inference especially is what currently consumes most of the computational workloads when solving hard problems with AI and is expected to scale through the 2030s but we don’t know for sure…
The Advantage of Owning Real Assets
Either way, if AI develops the way we expect, power is the single most valuable asset. IREN’s own hydroelectric sources provide an edge few can replicate. Datacenters require 97% uptime, good luck competing with that unless you unlock fusion.
The U.S. grid already operates with limited excess capacity, leaving little room for the explosive growth in AI compute. Expanding generation, transmission and grid infrastructure will take years, making existing power assets even more valuable.
IREN also has something most other Neoclouds don’t: Land. And land isn’t equal. Most is just mountains or dense forest, which isn’t suitable for much infrastructure. So usable, well-located land is rare and only becomes more valuable with time.
In this case it comes as a massive insurance policy. If this all goes haywire, the land can be repurposed or even sold. But hopefully repurposed, and with energy being such a bottleneck they can repurpose their hydro-assets to stabilize the grid for instance.
Bitcoin Mining No One Talks About…
Mining is for the most part a poor business, always soaring last when asset/metal spike and dipping first when it declines… But IREN is one of the largest miners in the world controlling 5% of the Bitcoin network. And few even know how it works.
It all boils down to energy and something called ASICs which are machines that mine Bitcoins. Bitcoin mining is random, but not equally random.
Think of it like buying lottery tickets. A Bitcoin is mined every 10 minutes. Suppose every 10 minutes there’s a lottery draw. Each lottery rewards 3.125 BTC…
Miner A owns 1 ASIC.
Miner B owns 100 ASICs.
Miner C owns 10,000 ASICs.
Everyone is playing the same game, but the more machines you have the more guesses you’re making per second.
Guess #1 → Nope
Guess #2 → Nope
Guess #3 → Nope
Guess #7,493,124,881 → Success!
For every new block, IREN is statistically winning ~1 out of every 20 lotteries. So if Bitcoin recovers and IREN continues to hold a majority of the network? They can withstand an AI winter better than most.
The Durability of Vertical Integration
Putting the pieces together it becomes clear that IREN has a headstart and is perfectly aligned with the new dawn of inference-time compute.
But it will come at a cost. Switching from Bitcoin mining to high performance computing is an expensive endeavor and the ASICs themselves need upgrading too. But there won’t be another time to expand than right now.
Does it justify the dilution? Maybe. But it doesn't justify the billions flowing to management. But the potential upside from cloud compute with both land and power secured is an unrecognized advantage especially when coupled with Bitcoin.
And it remains true even if Bitcoin declines further as it would benefit them, decreasing the hashrates temporarily leaving more for IREN to take. It’s not a flywheel but there is edge here.
Lower BTC price → Less incentive for other miners to mine → Decreased hashrate → Higher IREN BTC network share.
My Move?
Despite the dilution and the execution risks, I’m staying long.
The market is pricing IREN like another GPU rental company, but I think it’s much harder to replicate. Power, land and grid connectivity with BTC/AI infra all reinforce each other.
If one segment underperforms, the underlying assets retain strategic value. I don’t think AI has to perform just yet. Only that demand for electricity and digital infrastructure keeps growing, which it seems to be.
If AI accelerates → IREN benefits through HPC.
If AI slows → Its physical assets provide downside protection and optionality.
I don’t like dilution. I don’t like management enriching themselves. But the thesis hasn’t changed…
IREN still owns hard assets in industries where scarcity is increasing...
NOT GIVING IN.
🛰 Signal LOST: End of transmission — INVADER out… ⚡😭
Disclaimer: This post is for informational and entertainment purposes only. It reflects personal opinions and is not intended as financial advice. Always do your own research before making investment decisions.







oh shit, i thought it was monday today. but it's sunday...
This is the hard part with asset-heavy stories.
The assets can be real, scarce, and strategically valuable — and management can still become the largest risk in the thesis.
Power and land may create downside protection, but dilution changes who actually gets to own that protection.
Sometimes the business thesis survives while the shareholder thesis gets impaired.