Nobody buying SK Hynix tonight will remember how hated it was not too long ago. When I wrote the below in January 2024, the dominant narrative was about the inescapable memory glut. MU & SK had been flat to down for a year while NVDA had tripled. But…something that this cycle has taught me is that being early is a double-edged sword in more than just the classic “risk a drawdown before it works” way. I held on to these names for a little more than 2 years, but it’s pretty clear now that I sold too early. Now, despite a 500%+ return, I’m wondering if selling in February means I’ll have to withstand the mindfuck of watching them 10x again (by next Tuesday, at this rate). I’m already debating with myself whether I should bite the bullet and get back in or if doing so will align perfectly with the exact top. I love this game, but good lord does it mess with your mind!

For almost 3 years now, since CitriniResearch’s first AI Infrastructure article in May 2023, we’ve published our Dynamic AI portfolio. It’s our longest running actively managed portfolio and just crossed +400% since inception. We have attempted to stay ahead of rotations within AI - everything from GPUs to optics to memory to power and more. Through every drawdown and rip of the majority of this cycle, our AI allocation has been transparently open for our subscribers. It’s one thing to write research, it’s another thing entirely to publish a collection of positions based on that research. At this point, we’ve written about 20 articles related to AI and it’s sub-themes - each one has has trades, some of which I certainly wish I’d held longer, some I wish I hadn’t put on and some I’m quite proud of. Funnily enough, the biggest single winner by contribution rate remains when we went short with 50% of the portfolio on SMH leading into the “DeepSeek” moment and covered in April. Another amusing thing, I’ve been covering the AI infrastructure trade for more than a thousand days now and I don’t think a single one of them has gone by without someone calling it a bubble. 3 years is a long time to trade a single theme, but every time that I think I’ve seen all this trend has to offer, it surprises me with something new. Curious what the next thing will be, and hopeful we’ll be ahead of it.