So this won't be organized well — I'm pretty exhausted. But here we go: 1. To start, I definitely agree it hasn't gone how I thought it would, but in hindsight there's just too much complexity/ immense knowledge required to figure out how this whole AI ecosystem plays out. So many moving parts. I used to confidently say SW1 signs in the next 3 months — that was 8 months ago, and now I'm like, yeah, it could be as late as Q1 2027 lol. So on that front I've reset my framework to accept the complications and nuances of what they're trying to do. 2. Since you don't think I'm brainwashed — I was very frustrated with the comp package. It basically felt like your gf/wife cheated on you. But there were at least some takes that it brings their ownership in line with founders/CEOs of competitors once their Class B voting shares dissolve or time out. Also, what they've achieved in a year is incredible. Sure, the share price doesn't feel like it, but it's a seriously cool company pushing the envelope of data center design and AI factory prototypes. 3. I hear you on the flywheel and dilution, but honestly the ATM is a critical gap filler, and the flywheel is actually on track. They're trying to keep everything pushing across all lines of the business and scale as fast as possible. Say in a perfect vacuum they only had the legacy BTC mining business plus Horizons 1–4 with Microsoft. Once that had all its construction/GPU financing sources, it would no longer need an ATM to progress. But they'd have to wait until it was all stabilized, with project-level asset-backed debt in place, before doing anything next with the freed-up cash. All the anti-ATM people would be sitting on their thumbs far longer, and they'd only scale linearly and slow. What they're doing now is moving fast and scaling exponentially. Shooters shoot, and they're clearly working toward massive long-term plans. My other main point on the ATM: private companies dilute all the time in new series raises. Those get praised because it's an instant markup, so no one cares — but it doesn't mean the same thing isn't happening here. Running the ATM at $50/60/70 is incredibly strong compared to early 2025. To be clear, it hurts me deeply to imagine them running it in the $30s, but the logic technically still holds. It's also very easy to pause the ATM, so I certainly imagine it's not running at these levels. 4. Totally normal sentiment about the price now and what it looks like after an announcement — but to be clear, this stock goes up just as fast as it goes down. Just because a deal tomorrow might only be a 15% pop doesn't mean the stock won't be 100% higher a week later. AI sentiment and macro are crazy finicky but can flip bullish very quickly. IREN is mostly still in prove-the-model-works mode, and also caught in prove-AI-is-as-big-and-profitable-as-some-say mode. I'm strongly of the opinion they'll prove it, and soon -and that AI is as life-changing for humanity as any major breakthrough in our history as a species. IMO it's a matter of time, and not that much time, before both are proven fact and IREN violently rerates on superior economics, pipeline, and talent versus other neoclouds. 5. The coming months have a number of possibilities. I think H5/H6 sign with MSFT, dependent on H1 delivery (very soon) and potentially held up on VR pricing — but I suspect that resolves super soon. The economics will be much stronger all around and time-to-build will be very strong. There will also be a lot more individual smaller contracts with all the incoming GPUs already procured. They'll get better prepayments & terms than ever before. SW is the wildest of wild cards on timing, but whatever happens there will be huge. Some new model with NVDA's Nemotron, selling straight to enterprises, seems like the most likely outcome. Still trying to understand how they'd fund it — maybe deeply discounted GPUs for rev share, and maybe a partial MW allocation to Thinking Machines or a frontier lab
