Everyone wants to invest in AI, but most people stop at Nvidia. The way I think about it, AI is a stack. Understanding where money flows through that stack helps you understand what you’re actually betting on. Here’s the simple breakdown: 1. Chips - The Compute 🧠 AI requires MASSIVE computing power. $NVDA and $AMD design GPUs, $TSM manufactures many of them, $ASML makes critical equipment used to produce advanced chips, and $MU provides the high-speed memory AI processors need. And demand doesn’t stop once an AI model is built. Training builds the model, while inference requires compute every time someone actually uses it. The bet: More AI usage = more demand for compute. 2. Infrastructure - Where AI Lives ⚡ Millions of chips need data centres, servers, networking, cooling and a LOT of electricity to actually run. That’s where companies like $ANET (networking), $VRT (power/cooling), $CLS (hardware) and AI compute providers like $CRWV come in. It’s basically the picks-and-shovels layer of AI. The bet: More compute = more infrastructure needed to support it. 3. Platforms - How AI Gets Distributed ☁️ $MSFT, $GOOGL, $AMZN, $META and $ORCL are spending billions building AI infrastructure and then distributing AI through cloud services and products people already use. Instead of every business building its own AI infrastructure, they can rent compute and access AI through these platforms. And these companies already have massive businesses outside of AI. The bet: AI becomes embedded into more businesses → platforms capture more of that spending. 4. Applications - Where AI Gets Used 🤖 This is AI actually becoming a product. Cybersecurity, healthcare, finance, coding, automation, robotics - potentially almost every industry. But if everyone has access to similar AI models, the question becomes: What makes your product different? The winners may be companies with unique data, distribution, customers or workflows that competitors can’t easily replicate. The bet: AI creates new products and makes existing ones more valuable. So how do you invest in it? Broad-market ETFs already give you exposure to many major AI companies. Sector/AI ETFs let you target specific parts of the stack. Individual stocks let you make more concentrated bets on who you think wins. But always check your overlap. Owning an S&P 500 ETF + $NVDA + a semiconductor ETF containing $NVDA isn’t necessarily diversification - you may just be increasing the same bet. The more specific your AI bet becomes, the more position sizing matters. You don’t need to predict exactly who wins AI to benefit from it. You just need to understand what you own, where it sits in the stack, and why you own it.
9,968 views
6 Comments
Blossom
@blossom · 11d
Nate
@hoodnate · 10d
The overlap point really hit home. I hold an S&P 500 ETF and was eyeing a semiconductor ETF, so this saved me from doubling the same bet. Do you prefer broad ETFs or individual names here?
Tej
@investwithtej · 11d
Valid ✅
Daniel Newman
@patzkedog · 9d
So my problem is I'm not at a level of buying individual shares of these companies. I have a few etfs involved in all of this But what is the most stable productive fund to put your money going forward. Seems like it has been a roller coaster for awhile. Thank you and I appreciate any feedback. Funny I'm just a novice but I have been doing very well since the end of the pandemic.
See the full comment section 👀Sign up for the full Blossom experience!
