Latest / Stock Picking with Fexingo: Individual Equities, Research, and Building a Concentrated Portfolio / Why Jamie Dimon Warns About Hyperscaler AI Spending
Transcript
- Lucas: So I was reading Jamie Dimon’s comments this morning, and he dropped a number that stopped me cold: he thinks hyperscaler artificial intelligence capital spending is going to hit one trillion dollars next year. Luna: One trillion? That sounds like a round number meant to scare people, but is there actual data backing that scale of outflow? Lucas: Well, if you look at the combined capex guidance from the big five tech firms over the last two quarters, they are collectively committing roughly three hundred billion dollars annually to data centers and chip fabrication. So a jump to one trillion isn't impossible; it would mean they triple their current run rate. Luna: Triple the current rate in a single year? That feels aggressive even for the current hype cycle. Lucas: It is aggressive, which is exactly why it matters for our podcast today. Because when you are picking individual stocks, you have to decide if you are buying into the builders of that infrastructure or the companies trying to monetize it before the money runs out. Luna: Right, so we are looking at a tug of war between the hardware suppliers getting paid upfront and the software platforms hoping for a return on investment that might be years away. Lucas: Exactly. And here is the concrete thing I want us to drill into: the market has already priced in massive growth for the chip makers, but the hyperscalers themselves are trading at valuations that assume those one trillion dollar investments will yield immediate efficiency gains. Luna: But do they know how to use the compute power effectively yet, or are they just burning cash to keep up with competitors? Lucas: That is the million dollar question, literally. Let's look at Meta. Their stock is up ten point seven percent over the last five days alone, pushing past seven hundred forty four dollars. They are spending heavily on AI models to improve ad targeting, which directly drives their core business. Luna: Meta makes sense because the link between AI spend and ad revenue is direct. What about a company like Microsoft or Alphabet where the path to profit is more diffuse? Lucas: Alphabet is interesting because their search dominance is being challenged by generative AI features that don't necessarily generate more ad clicks per user. Google is spending billions to defend its moat, not necessarily to expand it immediately. Luna: So you are saying the defensive spend is riskier than the offensive spend that creates new revenue streams? Lucas: I am saying that when you see one trillion dollars in spending, you have to ask who is capturing the value. If all that money goes into building faster chips and cooler servers, the margins for the hyperscalers might actually compress rather than expand. Luna: That is a really counterintuitive angle. Usually we think bigger tech spend equals bigger tech profits. Lucas: Only if the products sell. And right now, consumer adoption of AI assistants is still pretty niche. Enterprise adoption is growing, but companies are cautious about paying premium prices for features that were cool demos a year ago. Luna: So we might be seeing an infrastructure bubble where the buildings are too big for the customers living inside them. Lucas: Or at least a period of significant volatility as the market realizes the timeline for returns is longer than the chart patterns suggest. This is why I think individual stock picking requires a much deeper dive into cash flow statements right now. Luna: Yeah, because free cash flow is going to get hammered by those capital expenditures before any earnings growth shows up. Lucas: Precisely. And that brings us to a practical step for listeners who want to navigate this. You need to look at the ratio of capital expenditures to operating cash flow for each company you hold. Luna: If that ratio is spiking above fifty percent, does that automatically signal a sell? Lucas: Not automatically, but it signals that the company is betting its entire future on a single horse race. For Apple, which is sitting at three hundred forty two dollars, they are spending less aggressively than their peers, which keeps their margins healthier. Luna: Apple is the odd one out here then, acting more like a mature consumer electronics firm than a pure play AI infrastructure builder. Lucas: They are trying to be both, but their balance sheet allows them to wait. They can buy smaller AI startups instead of building massive data centers from scratch. It is a different risk profile entirely. Luna: So diversification within your tech portfolio becomes critical because you are mixing high-risk builders with lower-risk acquirers. Lucas: Absolutely. And remember, this is September twenty twenty six. The market has absorbed a lot of this news, but the physical constraints of energy and supply chains are starting to bite. Luna: You mentioned energy earlier. Is that the other side of the one trillion dollar coin? Lucas: It is. Data centers are hungry for electricity. If utility costs rise because everyone is trying to build the same one trillion dollars worth of infrastructure, it eats into the bottom line for every single tech stock. Luna: So even if the AI software works, the power bills might kill the profitability. Lucas: Exactly. Which is why I think investors should also look at the utility and energy sector plays that supply these giants. It is a second-order effect that most people miss. Luna: That is a smart hedge. Instead of just betting on the winner of the AI race, bet on the person selling the fuel. Lucas: Right. And speaking of bets, if you find our breakdown of these complex macro forces useful, the way we keep these episodes ad-free relies on listener support. Luna: Yeah, if this conversation helped you clarify your own investment thesis, consider supporting the show. Lucas: You can grab a coffee for us at buy me a coffee dot com slash fexingo. Luna: It really helps us keep digging into the details that matter for your portfolio. Lucas: Back to the numbers then. Let's look at the broader market reaction. The S&P five hundred is up two point eight percent over the week, driven largely by tech. Luna: But the Dow is only up zero point seven percent. That divergence tells you something about where the real momentum is hiding. Lucas: It tells you that the industrial and financial sectors are lagging behind the tech giants. JPMorgan and Berkshire Hathaway are both down this week, which suggests traditional value investors are staying on the sidelines. Luna: Maybe they are waiting to see if that one trillion dollar spending plan actually materializes without causing a recession. Lucas: Or maybe they see the risks we talked about and prefer to sit in cash until the dust settles. It is a valid strategy to watch from the outside. Luna: Patience is definitely a form of alpha in this kind of volatile environment. Lucas: It is. And finally, consider this: when you pick your stocks, are you picking based on the headline number or the underlying unit economics? Luna: Because the headline says one trillion dollars, but the unit economics might say we are overpaying for every single chatbot feature. Lucas: Spot on. Focus on the unit economics. Look at the cost per transaction, the retention rates, and the actual energy consumption per user. Luna: Those are the metrics that survive the hype cycle. Lucas: Exactly. And that is probably the most important takeaway for anyone building a concentrated portfolio right now. Luna: Will the winners be the ones who built the biggest infrastructure, or the ones who used it most efficiently? Lucas: That is the question we will be watching closely in the coming quarters.