You have no idea how proud this makes me. In just two months, the Discord I've built has turned into one of the best trading communities I've been part of, where people support each other, learn from one another and share high quality alpha. The best part is that I didn't build it alone. This is only the beginning ⚡

$NBIS $SNDK $MU $AMD $BE $LITE $MRVL $GLW $DELL $TSEM $AAOI $AEHR $MXL $PENG $SIVE $AXTI $CBRS $FLEXQ After giving it some thought, and for the benefit of the broader community, I'll be sharing ALL the research I create publicly moving forward, with a 2-week delay between when the subs get it and when everyone else does. Everything goes out as long articles, both on X and on Substack. You choose where, the research stays the same. That includes the 30+ long-form articles I've already written. I hope they help someone build their own thesis. The research belongs out in the open. What stays exclusive for subscribers is where the thesis becomes actionable: my portfolio updates, my trades (both long and short dated), and the X sub and Substack/Discord community. Funny thing, I've started re-sharing deep dives I wrote 3 months ago, and most names are getting back to the valuations they had in June, right before the unwind. My valuations, assumptions and thesis behind each name haven't changed. If anything, a really strong earnings season has only reconfirmed them. We're still very much at the beginning of the AI buildout. Maybe not the very beginning anymore, but there's so much more to come! So follow along and learn with me where the AI trade will take us. Bullish, Ren ⚡

$SNDK Citi expects enterprise SSD demand to rise 52.9% in 2027 and 41% in 2028, with total SSD demand up 45%. NAND demand grows 29% in 2027 and 33% in 2028 against supply growth of 21% and 25%, pushing the supply-demand ratio from −0.8% this year to −6.1% in 2027 and −5.5% in 2028. Expects industry NAND wafer capacity to grow only 3.2% in 2027, with Samsung's NAND capacity actually falling 4.7% as it redirects resources to DRAM and HBM, and SK hynix and Micron roughly flat. That's the mechanism behind the deficit: demand up 29%, wafers up 3%, with the gap covered only by process migration. Citi expects the tightness to persist until 2031, with DRAM deficits of −8.7% and −9.7% in 2027 and 2028 alongside the NAND numbers. And projects NAND average selling prices up 45.3% in 2027. Bullish for $SNDK

$SNDK Funny thing. I've been hyper bullish on agentic AI for over a year, and it's one of the main reasons I hold the stocks I do, memory above all, and NAND in particular. TrendForce just raised its NAND price forecast. Enterprise SSD Soars, Consumer Stalls. Enterprise SSD orders are still climbing into Q4 on North American cloud demand, and the two drivers they name are agentic AI and KV cache offload. Agents generate state. Every task an agent runs leaves behind retrieval results, working files and intermediate steps, and all of it has to be stored somewhere fast enough to be useful and cheap enough to keep. That's why cloud providers are scaling QLC SSDs, which pack the most bits per dollar of anything with an interface fast enough to serve inference. KV cache offload is the other end. Models need to hold context, and until now that lived in HBM and DRAM, the two scarcest, most expensive places on earth to keep anything. Push the parts you don't need every millisecond onto SSDs and you free up the expensive tiers for the work that actually needs them. TrendForce says the industry is building frameworks specifically to do this, and China's inference architectures are already leaning on it. The funny thing is the same report says consumer NAND is oversupplied, with YMTC adding capacity and a two-track pricing structure emerging in 2027. Enterprise NAND it’s becoming a different product from the stuff in your phone, with its own supply, its own buyers and its own price. The part of the market that agents need is the part that's short, and the part that's oversupplied is the part $SNDK doesn't depend on. Bullish $SNDK

This looks scary at first glance. Three companies worth more than 45 years of tech IPOs combined. Bubble, right? Follow the dollars instead. A frontier lab raises money. What does it spend it on? Compute, and not much else. Some of it they buy outright, but most of it they rent, which means multi-year contracts with hyperscalers and neoclouds. Those contracts are what a neocloud takes to its lenders to borrow against. So every dollar of equity a lab raises can end up financing several dollars of hardware further down the chain. Now, who takes the risk if the stock trades badly after listing? Whoever bought the IPO. The cash has already been raised, the contracts have already been signed, and the orders are already placed. And where do those orders go? Of course, to the AI buildout: GPUs. CPUs. HBM and DRAM. Flash. Optics and interconnect. Power. The suppliers get paid either way. So regardless of how these names perform after they have been listed, this is bullish for the AI buildout ⚡️

$META $AMD $MU $SNDK $INTC $ARM Meta's Muse just took what OpenClaw proved an autonomous agent can do and removed the barrier to entry. The result: people who don't care about setting up technical stuff, or about AI itself, using a product that makes their lives easier. Ben Thompson called it "the best and most approachable personal agent product I have tried," and regular people seem to agree. No. 1 on the US App Store in ten days, 1.8M US and Canada iOS downloads in its first 12 days against 1.3M for ChatGPT, and 642K US daily users against 231K. THE MODEL IS CHEAP. THE COMPUTER ISN'T. On Cursor's own benchmark, Muse Spark 1.3 at max roughly matches GPT-5.6 Sol at max, 41.6% against 41.7%, at about a third of the cost per task. Meta didn't build the smartest model, it built one good enough and cheap enough to give away, and that moves the bill from the GPU to the machine the agent lives on. Every user gets their own always-on cloud machine, 2 vCPUs, 8GB of RAM and 100GB of SSD, which stays on whether you're using it or not. AGENTS DON'T CHURN LIKE CHATBOTS Switching chatbots takes a click. Once an agent holds your email, your calendar and your habits, leaving means starting over. Sticky users mean machines that stay switched on for a long time. And Meta plays this game from strength: it already runs the best personalization engine in consumer tech, and Adam Mosseri says it's only now getting as sophisticated as people always assumed it was. WHAT IT MEANS FOR THR AI TRADE At 100M users: 200M vCPUs, roughly 0.5M to 1.6M server CPUs depending on core counts and sharing, 800 PB of RAM and 10 EB of SSD. Meta will share hardware and park idle machines, so the real number lands lower, but every lab now has to answer with its own cloud computer per user. Chatbots made GPUs scarce.Agents add CPUs, DRAM and flash to the list. Bullish AF for the AI buildout ⚡

Rosenblatt's Kevin Cassidy just initiated $SNDK at Buy with a $2,400 target, about 35% above where it traded today. Management's framework for FY28 to FY30 is mid-to-high-teens revenue growth, roughly 80% gross margins, and half of every dollar of revenue turning into free cash flow. Rosenblatt calls its $300 FY30 EPS estimate conservative. At $2,400 you'd be paying about 8 times that FY30 number. For a business with 80% gross margins and most of its output pre-sold, that isn't a hype multiple, it's the market still pricing SanDisk like the old cyclical it used to be. The target sits comfortably above the roughly $2,000 street average, right next to BofA's $2,500. $SNDK is one of my largest positions. I’m hyper bullish on this name.