Latest / The Bear Market Podcast with Fexingo: Surviving Downturns, Buying the Dip, and Long-Term Resilience / How Utility Stocks Became the Stealth AI Trade in 2026
Transcript
- Lucas: S&P 500 sits at 7,537 as of this morning. Up one point three percent over the last five days. VIX at 15.57 — low, but not the kind of low that had us worried a few weeks ago. The market looks calm. But underneath that surface, there's a rotation happening that I think a lot of retail investors are missing. Luna: What kind of rotation? We've talked about small caps beating tech, but that's not really the headline anymore. Lucas: Right, that trade has cooled off. The Russell 2000 is basically flat over the last five days — three thousand, barely moving. What I'm looking at instead is utilities. Specifically, regulated electric utilities. And the reason is AI. Luna: AI as in data center power demand? Because I've seen some estimates that are pretty wild. Lucas: Exactly. Goldman Sachs put out a note in May projecting that data center power demand will grow at a compound annual growth rate of 15 percent through 2030. That's roughly triple the overall electricity demand growth rate in the US. And who benefits first? Not the hyperscalers — they're fighting over chip supply. The immediate beneficiary is the utility that can actually deliver the electrons. Luna: So the sleepy regulated monopoly play becomes a growth stock. But aren't utilities super sensitive to rising rates? The ten-year is at 4.49 percent, that's up from 4.48 last week — still high. Lucas: That is the tension. In a normal cycle, a ten-year above 4 percent crushes utilities because they carry a lot of debt on the balance sheet. But this time, the revenue story is so strong that it's offsetting the rate concern — at least for the companies that have a credible ai related buildout plan. Luna: So which names are you watching specifically? Give me one that's playing offense and one that's playing defense. Lucas: Okay. On offense: NextEra Energy. They already have the largest portfolio of wind and solar in North America, but they're also building new transmission lines specifically to interconnect data centers. They announced a joint venture with a large tech company — unnamed — to develop a gigawatt-scale campus in the Southeast. That's one billion watts of capacity dedicated to one facility. Luna: A gigawatt campus. That's roughly the output of a nuclear reactor. And who's the defensive play? Lucas: Dominion Energy. They have a different strategy — they're selling land near their existing nuclear plants to data center developers. Dominion's North Anna nuclear station in Virginia already has a data center park being built on site. They're not spending the capital on transmission; they're monetizing the real estate and the power purchase agreements. Lower risk, lower reward, but the earnings per share visibility is actually better. Luna: So the thesis is: AI isn't just a semiconductor story anymore. It's a grid story. But does the yield curve steepening we've seen — the ten-year minus two-year spread is now 35 basis points — does that change the math for utility debt? Lucas: It does, but only at the margin. The steepening we've seen is mostly because the short end is coming down — the two-year dropped to 4.14 percent, down from 4.17. The ten-year is basically flat. So the cost of new debt for utilities, which is tied more to the long end, hasn't really moved. The bigger risk is if the thirty-year yield keeps climbing — it's at 4.98 now — because that's the rate utilities use for their long-term infrastructure financing. Luna: Right, so if the thirty-year pushes above 5 percent, the interest expense eats into the ai driven revenue growth. How much headroom do these companies have? Lucas: NextEra's management indicated on their last earnings call that they can absorb a fifty-basis-point increase in long-term rates before the net present value of their new projects turns negative. That's not a huge buffer. Dominion has even less room because their regulated structure means they can't pass through all the financing costs immediately. Luna: So it's not a blanket buy on the utility sector. It's a stock-picker's game. And the VIX at 15.57 suggests the market isn't pricing in much rate volatility — but the VVIX, which is the VIX of the VIX, is at 87, still elevated from its long-term average. That implies options traders are hedging for a volatility spike. Lucas: That's a really good point. The VVIX at 87 says option sellers are demanding a premium for tail risk. If that tail risk materializes — say, a surprise CPI print that pushes long rates higher — then utility stocks with high debt loads could get hit disproportionately. Because they've already repriced to reflect the AI growth story. They're not cheap anymore. Luna: And by 'not cheap' you mean the P/E for the utility sector is now around 22 times forward earnings? That's above the ten-year average of 18. Lucas: Exactly. So you're paying a premium for the AI growth, and that premium is sensitive to interest rates. That doesn't make it a bad trade, but it does mean you need to size the position carefully. If the ten-year goes to 5 percent, the sector could correct 10 to 15 percent even if the underlying power demand story remains intact. Luna: That's a risk you can manage with duration hedging, maybe using Treasury futures or even options on the TLT. But for most listeners, the simpler move is to pick one utility with a clear AI catalyst and a manageable debt profile, rather than buying a utility ETF. Lucas: I think that's exactly right. And you know, when we talk about these kinds of allocation decisions — whether it's utilities or something else — a lot of the value of this show is just having a framework to think about the trade-offs. If our money conversations have helped you think about a decision differently, that's really the goal. And we're able to keep doing this ad-free because listeners who find it useful sometimes support the show at buy me a coffee dot com slash fexingo. Luna: Yeah, it makes a difference. It's a simple way to keep the conversation independent. Lucas: Alright — back to the market. So if we look at the VIX term structure, the front month is at 15.57, but the six-month future is around 19. That's a contango of about three and a half points. Normally that's a signal that spot volatility is expected to rise. But the interesting thing is that the contango is narrower than it was a month ago, which tells me the market is starting to price in a higher baseline for volatility, not just a one-off spike. Luna: So the VIX itself is low, but the market is telling you that low volatility isn't going to last. That's consistent with a rotation trade where you want to own assets that have a fundamental catalyst independent of the macro — like utility stocks with AI demand. Lucas: Exactly. And that brings me to another data point from the live feed: nominal GDP is at thirty-one point eight trillion, up from thirty-one point four trillion earlier this year. That's a four hundred billion dollar increase in nominal output. A decent chunk of that is inflation, but the real GDP growth annualized is 2.1 percent, up from 0.5 percent. That's a meaningful acceleration. Strong economic growth, but not so hot that the Fed needs to hike again. Luna: Goldilocks, essentially. And in a Goldilocks environment, utilities with growth stories can re-rate. But if we get a hard landing — which the yield curve steepening sometimes foreshadows — then the debt load becomes a problem. Lucas: Right. So the bottom line for this episode is: AI power demand is real, but it's not a sector-wide rising tide. It's a stock-specific story. NextEra on offense, Dominion on defense. Both have exposure to the same mega-trend, but different risk profiles. And the macro environment — low VIX, steepening curve, moderate GDP growth — is supportive for now, but fragile. Luna: One concrete thing listeners can do this week: look up the debt to capital ratio for any utility they own. If it's above 60 percent, and the company hasn't announced a data center related deal, the stock might be priced for a growth story that isn't there. Lucas: That's a great actionable takeaway. And we should note — this isn't a recommendation to buy or sell anything. It's a framework. The regulated utility business model hasn't changed. But the demand environment has, and the market is still figuring out how to price that. Luna: What's your sense of how long this AI infrastructure buildout can support utility valuations? Three years? Five? Lucas: The capital expenditure cycle for data centers is typically a five- to seven-year build, then a thirty-year operating life. So the heavy spending should continue through at least 2030. But the stock market will price that in long before the revenue shows up. We're probably in year two of a five-year re-rating. So there's still room, but the easy money has been made. Luna: I'll keep an eye on that thirty-year yield. If it breaches 5 percent, I'd take a second look at the thesis. Lucas: Same. That's the line in the sand. Thanks for listening, everyone. We'll be back next week with another angle.