Latest / The Bear Market Podcast with Fexingo: Surviving Downturns, Buying the Dip, and Long-Term Resilience / The AI Capex Trap and the Valuation Gap
Transcript
- Lucas: We are looking at an S&P 500 that has climbed to seven thousand seven hundred twenty-one points, but if you look closely at the breadth, it feels less like a broad-based recovery and more like a very narrow bet on infrastructure. Luna: Right, and with the ten-year Treasury yield hovering around five point zero nine percent, the cost of money isn't cheap enough to justify the sheer scale of spending we're seeing from the big tech firms. Lucas: Exactly. The headline number for the index is fine, but the real story here is the capital expenditure cycle. We have companies like Microsoft, Google, Amazon, Meta, and NVIDIA collectively spending hundreds of billions on data centers and AI chips. The question isn't whether they can build it, but whether the revenue actually catches up before their margins get crushed by depreciation and energy costs. Luna: That is the classic infrastructure paradox, isn't it? You spend billions to capture the future, but if the future arrives slower than your depreciation schedule, you are just burning cash for prestige. Lucas: Precisely. And this week, the HSBC CEO came out and explicitly warned that not every data center investment will be a good bet. He noted that while the demand for compute is real, the path to profitability is cluttered with inefficiencies. It is a stark reminder that just because a technology is transformative doesn't mean every player in the supply chain will see a return on invested capital. Luna: So we are essentially watching a gold rush where everyone is buying picks and shovels, but only one or two people are actually finding the gold. Lucas: That is a fair analogy, though I would argue it is even messier. In a gold rush, the price of gold is clear. Here, we do not yet know what the 'unit economics' of an AI inference call will look like in three years. Will it be pennies? Dollars? If it stays high, the model breaks. If it drops too fast, the current spending looks excessive. Luna: And that uncertainty is exactly why the VIX spiked five percent recently. Investors are nervous about the quality of this rally because they cannot price in the downside risk of these massive balance sheets. Lucas: Let us look at the numbers for a second. Real GDP growth is slowing to two point two percent annualized, yet these tech giants are growing their capex at double-digit rates. That disconnect creates a specific type of risk called the 'valuation gap.' The stock prices assume perpetual hyper-growth, but the macro economy is stabilizing, not accelerating. Luna: It feels like we are pricing in a utopia that the current interest rate environment might not support. Five percent borrowing costs really punish projects that take five years to turn a profit. Lucas: Which brings us to the interesting divergence we are seeing right now. While the mega-cap tech stocks are under pressure due to these capex fears, small caps in the Russell 2000 are actually catching up. The Russell is up one point five percent over the last five days, while the Nasdaq is down slightly. Luna: That is counterintuitive. Usually, when rates stay high, small businesses suffer the most. Why are they outperforming right now? Lucas: Because many of them are unencumbered by the AI spending race. They are not trying to build data centers. They are focused on organic efficiency and niche markets. When investors get scared about the margin compression in big tech, they rotate into these quieter, profitable mid-sized companies that don't have a hundred billion dollar burn rate. Luna: So the market is essentially saying, 'I love AI, but I don't want to pay the toll for it.' It is a vote of no confidence in the current business model of the biggest tech firms. Lucas: I think it is more nuanced than that. It is a vote of confidence in alternative uses of capital. If I am a CFO of a manufacturing company or a healthcare provider, I am not going to spend five billion dollars on a GPU cluster. I am going to buy back my own stock or acquire a smaller competitor. The capital is flowing away from the speculative frontier and toward established cash flows. Luna: That makes sense. But does that mean the AI trade is dead? Or is it just entering a consolidation phase where only the winners survive? Lucas: The trade is not dead, but the easy money has been made. We are moving from the 'build everything' phase to the 'monetize something' phase. And that transition is painful for stocks that are valued on potential rather than profit. Luna: Speaking of monetization, I saw headlines today about Anthropic potentially being the last big AI IPO hope for this year. Does that change anything for the existing players? Lucas: It highlights the fragmentation of the sector. We have OpenAI, Anthropic, Google, Meta, Microsoft, Amazon... there are so many models competing that none of them have a clear monopoly yet. An IPO from Anthropic would bring fresh capital into the space, but it would also increase the competitive pressure on the incumbents who are already spending heavily. Luna: So more competition means lower prices for consumers, which is good for us, but bad for the stock prices of the companies building the infrastructure. Lucas: Exactly. Deflationary technology is great for society, but it can be terrible for equity valuations in the short term. If AI makes software cheaper, then the price to earnings ratios of software companies have to adjust downward unless their volume grows exponentially. Luna: It is a fascinating dynamic. We are witnessing the industrialization of intelligence, and industrialization always comes with margin compression before it stabilizes. Lucas: Before we go deeper into how individual investors should position themselves for this volatility, I want to take a brief moment to talk about how this show operates. If our conversations about market dynamics and investment strategy have helped you think about your financial decisions differently, I want to mention that we deliberately do not run ads on these episodes. We believe that keeping the content clean and focused on education is worth protecting. If you want to support that choice directly, the link is buy me a coffee dot com slash fexingo — that's F. E. X. I. N. G. O. It helps cover the production costs and keeps the signal clear. Luna: Yeah, and having that direct line of support means we don't have to chase clicks or sensational headlines to keep the lights on. We can just stick to the data. Lucas: Right. So, getting back to the portfolio level. How do you handle this 'capex trap'? Do you stay fully exposed to the big tech names and wait for the revenue to catch up? Luna: I think staying fully exposed is risky. The margin of safety is shrinking. I would lean towards diversifying into companies that benefit from AI but don't bear the massive capital burden. Lucas: That is a smart distinction. Think about the software companies that integrate AI into their products. They are selling licenses, not building the chips. Their margins are likely to expand as they leverage the underlying infrastructure built by the hardware giants. Luna: Or perhaps looking at the utility side. The companies providing the power and water for these data centers. They have regulated returns and essential assets. They are playing a different game than the tech speculators. Lucas: Utilities are an interesting hedge. They are sensitive to interest rates, obviously, since they carry debt, but they are also beneficiaries of the physical demand for electricity. It is a tangible asset play in a digital world. Luna: But you still have to watch those Treasury yields. If the ten-year yield spikes back up past five point five percent, utilities become less attractive compared to bonds. Lucas: True. So the key is balancing the growth potential of the AI infrastructure with the stability of the entities that support it. It is not about picking one side or the other, but understanding where the value is hiding in the ecosystem. Luna: It really comes down to patience. Waiting for the revenue curves to cross the expense curves. Until then, volatility is going to be the norm. Lucas: Absolutely. And for the listener driving home tonight, the takeaway is simple: look at the balance sheet, not just the press release. Ask yourself who is paying for the party, and who is actually drinking the champagne. Luna: If the bill is coming due faster than the checks are clearing, it is time to step back and reassess your exposure. Lucas: We will see where the market settles as we move through the rest of Q four. For now, keep your powder dry and your margins wide. Luna: Thanks for listening, and let us know which part of the AI infrastructure you think is undervalued right now.