Latest / Stock Picking with Fexingo: Individual Equities, Research, and Building a Concentrated Portfolio / Why AI Data Centers Are Not Safe Bets
Transcript
- Lucas: The headline coming out of HSBC this morning is worth pausing for. Their CEO, Noel Quinn, just said the AI boom won't make every data center a good bet. It’s a direct challenge to the idea that pouring money into infrastructure guarantees a return. Luna: That feels like a necessary reality check. We’ve been told for two years that more compute equals more value, but the margins aren't always there. Lucas: Exactly. And look at the tape today. Meta is down six percent in the last five days. Amazon is up three percent. You have two giants in the same space, one struggling with efficiency questions and the other scaling aggressively. The divergence tells you everything about how the market is pricing risk right now. Luna: So we’re seeing a split in conviction? One side thinks they can build it cheaper, the other thinks they can monetize it faster? Lucas: Precisely. When I look at Meta’s slide, it’s not just about building the models anymore. It’s about whether those models actually drive engagement or just burn cash. The S&P 500 is hovering around seven thousand seven hundred, holding steady despite that drag from big tech. Investors are getting picky. Luna: I’m curious if you think the supply chain constraints are still the main bottleneck, or if it’s purely a demand problem now. Lucas: It’s a mix, but the demand question is louder. You’ve got hyperscalers spending billions on chips and power. But if the software layer doesn’t unlock new revenue streams, those facilities become white elephants. Remember when everyone bought server racks expecting immediate ROI? Many didn't get it. Luna: Right, so the capex cycle is running hot while the profit cycle is lagging. That’s a dangerous mismatch for a concentrated portfolio. Lucas: It is. If you own a few big positions, you can’t afford to assume every dollar spent on AI will pay off. We need to look at who has the moat. Is it the chip maker, the cloud provider, or the company using the AI to cut costs? The answer changes your thesis entirely. Luna: Let’s drill into that last point. Why do you think the end-user companies are the safest bet compared to the builders? Lucas: Because they control the customer relationship. A chip maker sells hardware once. A software company sells subscriptions forever. If the AI improves retention or upsells, the end-user captures that value. The builder just gets their margin. It’s a classic platform play. Luna: But isn’t the hardware scarcity creating a temporary monopoly power for the makers? That seems like a strong argument for buying the suppliers instead. Lucas: Short term, yes. NVIDIA and TSMC are printing money because no one else can scale fast enough. But look at Tesla. It’s up almost two percent today. They’re building their own chips. Competition drives prices down. The monopoly window is narrow. Luna: So you’re saying don’t chase the hype, buy the companies that are already profitable and just adding AI as a feature? Lucas: I’m saying focus on quality compounders. Companies with strong free cash flow that use AI to boost margins, not replace them. Look at Microsoft. Down slightly today, but they’ve integrated Copilot across Office. That’s recurring revenue, not a one-time sale. Luna: That makes sense. It’s less about the shiny new toy and more about the boring, efficient engine underneath. Lucas: Exactly. The HSBC comment is useful because it forces us to ask: who is actually making money here? If you can’t answer that, you shouldn’t own the stock. It’s that simple. Luna: What about the energy angle? Data centers are hungry. Does that create a secondary opportunity in utilities or grid infrastructure? Lucas: Absolutely. Power generation is the hidden bottleneck. Nuclear, natural gas, renewables—it’s all needed. But those sectors move slower. They’re regulated. They’re not going to jump six percent in a day. It’s a slow grind, which fits a concentrated strategy perfectly. Luna: I like that. Low volatility, high necessity. It feels like a defensive play within a tech-heavy narrative. Lucas: It is. And it’s safer than betting on the next unproven AI startup. When rates stay elevated, speculative burns die quickly. Utilities survive because they have tangible assets. Luna: So we’re pivoting from growth at any cost to sustainable growth with real assets? Lucas: Yes. And that means auditing your current holdings. Do you have exposure to companies just riding the AI wave without a clear path to profit? If so, it might be time to trim. Luna: Harsh, but fair. Most of us probably hold something just because it’s popular right now. Lucas: We all do. That’s why we talk about it. To keep each other honest. The market rewards discipline, not momentum. Luna: Speaking of discipline, if our conversations help you navigate these complex shifts without losing sleep... Lucas: Honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo — that's F. E. X. I. N. G. O. Luna: It keeps the lights on and the research sharp, so thanks for considering it. Lucas: Back to the portfolio. Let’s look at Berkshire Hathaway. Up two percent today. Buffett isn’t chasing AI. He’s sitting on a record pile of cash. Is that a signal to wait for a better entry point? Luna: Maybe. Or maybe he knows exactly what he owns and doesn’t need to add more risk. Lucas: That’s the beauty of concentration. You know what you own, why you own it, and when to sell. No guessing games. Luna: Do you think the current market breadth supports this 'pick and choose' approach, or are we too dependent on a few mega-caps? Lucas: It’s tricky. The Dow is up significantly, driven by industrials and financials. The Nasdaq is dragging. So breadth is mixed. It favors active management over blind indexation. Luna: So passive investing might underperform here because it forces you to hold the losers along with the winners? Lucas: Exactly. In a fragmented market, selection matters. You want to avoid the companies burning cash on AI experiments that go nowhere. Luna: How do you identify those red flags early? What metrics should we watch before the earnings call? Lucas: Look at operating expenses versus revenue growth. If OpEx is growing faster than sales, especially in R&D, be wary. Also, watch guidance. If they keep raising expectations without delivering, it’s a trap. Luna: Practical advice. Keep an eye on the burn rate and the guidance consistency. Lucas: Yes. And remember, AI is a tool, not a business model. The best companies will use it to work smarter, not just spend harder. Luna: It’s a reminder that technology serves the bottom line, not the other way around. Lucas: Always. So as we close out this week, ask yourself: does my biggest holding make money from AI, or just spend money on it? Luna: If it spends more than it earns, maybe it’s time to rethink the position. Lucas: Good question to sit with. See you tomorrow.