Listen to the audio version of this article (generated by AI).
Democrats sweep Washington… voters turn on data centers… why the pipeline trade may get stronger… is this 1998 or 2000?… the levels Luke Lango is watching
As I write on Wednesday morning, the results are in, and they weren’t close.
Democrats dominated yesterday, winning control of both the House and the Senate. They also flipped governorships in several Republican-held states, including Ohio and Texas, where Gina Hinojosa unseated three-term Gov. Greg Abbott.
In Monday’s Digest, we said Tuesday would be a referendum of sorts on the AI buildout. Well, voters delivered a clear verdict…
Let’s start with Ohio, which is home to 21 of the 41 local data-center measures on ballots nationwide. According to Fortune, at least 14 of the 18 Ohio communities voting on the issue passed bans or other restrictions. In Trenton, a ban aimed at keeping out an 18-building Amazon Web Services campus passed 80% to 20%. And in Wilmington, about 80% of voters rejected rezoning for a data-center project across four separate referendums.
Meanwhile, in Ohio’s Senate race, Sherrod Brown ousted Sen. Jon Husted, after running ads calling Husted the “face of data centers.”
Wall Street’s reaction this morning has two sides. As I write, all three major indexes – the Nasdaq included – are up more than 1%. Markets generally like a divided Washington, since it lowers the odds of sweeping policy changes.
But if your portfolio is loaded with AI data center stocks, you’re having a different experience this morning as the stocks most tied to the data-center buildout are getting hit hard.
For example, Constellation Energy Corp. (CEG) is off more than 9% as I write. And Vistra Corp. (VST) is 15% lower. And Comfort Systems USA, Inc. (FIX) is down more than 17%.
So, is the AI data-center trade dead?
No, but it’s not the same trade that it was yesterday.
A Democratic Congress can hold hearings and stall the permitting reforms the industry wants, but it can’t pass new restrictions over a presidential veto, which President Donald Trump would be almost certain to use.
The more powerful brakes are at the local and state levels, which is exactly where voters acted. In both Texas and Ohio, the governor appoints the utility regulators who decide what data centers pay for power. And Texas is one of the largest data-center markets in the country.
So, the likely outcome isn’t a halt, it’s a buildout that moves slower, costs more, and increasingly must pay its own way – including supplying more of its own power instead of leaning on the public grid.
Before the election, the global equity research shop Bernstein estimated that 55% of the data-center pipeline faced political risk. But the firm also argued that “economic incentives may ultimately outweigh political opposition.”
We’ll see…
Voters may have just made the pipeline trade stronger
If voters and regulators are going to make data centers pay their own way, more of them will need to bring their own power. And much of that power will come from natural gas.
Solar is today’s headline winner, and renewables will get their share, but gas-fired power is the fastest way to deliver electricity around the clock at the scale AI needs, and the companies that move that gas are already lining up the business.
Yesterday, Energy Transfer LP (ET) committed to building a $1.6 billion natural gas pipeline that will run 215 miles from eastern Louisiana across Mississippi, with room to carry up to 1.2 billion cubic feet of gas per day.
Bloomberg put the news in context:
Energy Transfer and other pipeline operators are rushing to put gas conduits into service as consumption of the fuel soars, driven by liquefied natural gas export terminals along the Gulf Coast and new power plants serving AI data centers.
Energy Transfer’s own management was even more direct on yesterday’s earnings call. CEO Tom Long said the company has now signed a 10-year deal to supply “an A-rated hyperscaler,” one of the Big Tech giants building AI infrastructure.
In all, he said, Energy Transfer has signed more than 9.5 million MMBtu (millions of British thermal units) per day of agreements with customers that include “hyperscalers, data centers, and power plants.” That’s roughly eight times the capacity of the pipeline it just approved.
Co-CEO Mackie McCrea put it bluntly:
The demand for our products and what we do, especially natural gas, is going to continue to grow exponentially.
Brian Hunt saw this coming
If you’ve been reading our senior analyst Brian Hunt, editor of Money & Megatrends, none of this is news. Brian has been highlighting this theme repeatedly for nearly two years. Here he is from last Friday’s issue:
In early 2024, I began to see the oil and gas pipeline industry as the best way to generate substantial passive income from the AI boom.
The logic is simple. AI data centers need enormous amounts of electricity, and natural gas is the preferred fuel for the power plants that supply it. But as Brian points out, that only matters if the gas can get where it’s going:
…all the natural gas in the world isn’t worth much if you can’t transport it to customers.
This is where America’s vast natural gas transportation, processing, and storage industry comes in… If we get an AI-driven boom in natural gas consumption, we get a boom in natural gas transportation by default.
So far, the market agrees. Blue chip pipeline operator Enterprise Products (EPD) has returned 14% this year. Fellow blue-chip operator Energy Transfer has climbed 25%.
These individual stock performances have driven the Alerian MLP ETF (AMLP) to a 16% year-to-date gain. Even with AMLP’s big run and shares being near all-time highs, it still yields around 7.8%.
I’ll note that these pipelines offer a way for investors nervous about an AI crash to still participate in AI. Back to Brian on this note:
The typical pipeline operator is not your conventional “high-risk, high-reward” AI play. Instead, it’s a boring, predictable business that generates steady cash flows and shareholder distributions.
But the AI megatrend is giving natural gas a boost that will last for years. Plus, the Iran War and its constriction of Middle Eastern energy flows have made U.S. natural gas exports increasingly more valuable to customers in Europe and Asia.
For more of Brian’s free analysis, he writes Money & Megatrends every day the market is open, highlighting these kinds of opportunities before they become front-page news. His issues are loaded with actionable advice and specific tickers.
Is this 1998 or 2000 for the AI trade?
Now, let’s shift from pipelines back to AI more broadly.
Monday brought another round of AI selling, and by the close, the Nasdaq-100 was in correction territory, more than 10% below its October 6 record high. Yesterday, the script flipped. Chip stocks led a rebound while the Dow struggled. The Nasdaq-100 added 1% as it tried to claw its way out of that correction.
But today, even as the broader market rallies on the election results, the semiconductor index is sliding hard.
So, which is it? Is the AI trade breaking down, or getting ready to bounce?
Here’s our technology expert Luke Lango from Monday’s Innovation Investor Daily Notes with the significance of where the AI trade sits today:
…the whole AI trade is at a major make-or-break point. Nearly every major AI stock, index, and benchmark has fallen to major technical support.
Holding and bouncing here would mean this is late 1998 all over again, right after tech stocks fell 20%-plus and before they soared nearly 300% over the next 18 months.
Breaking down would mean this is potentially late 2000 all over again, right after tech stocks fell 20%-plus and before they crashed another 70%-plus over the next two years.
Luke’s Monday take was that the weight of the fundamental and technical evidence says that we’ve been experiencing a growth scare in a still-alive AI bull market, very similar to late 1998. So, he sees the AI trade roaring back over the next 12 to 24 months. But he doesn’t dismiss the risk that he’s wrong, and that AI has a far bigger slide in front of it.
The levels Luke is watching
Being bullish isn’t the same as being blind, so Luke has mapped out exactly what he’ll do as conditions change.
His focus is the Philadelphia Semiconductor Index (SOX) and its 200-day moving average. That’s the index’s average closing price over roughly the past 10 months, and traders treat it as a dividing line between a healthy uptrend and a broken one.
Luke’s team studied all 26 times since 1995 that the SOX ran more than 10% above that line and then pulled back to it.
Here’s Luke with the results, and what it means for positioning:
Our historical study suggests a modest break below the SOX’s 200-day average often reverses.
We’re holding near 10,800, would begin trimming weaker names below 10,447, and turn more defensive around 10,300.
A close above 11,684 would confirm the rebound and support more aggressive buying.
The key defensive number is roughly 10,300, about 5% below the 200-day. In Luke’s study, the SOX broke below its 200-day moving average but held above it in seven episodes. It was higher 12 months later in six of them, with an average gain of 20%, the best outcome in the study.
But breaks below the 5% line were a coin flip. Seven bounced, averaging 14%. The other six became bear markets, averaging losses of 28%.
He also flags a warning worth respecting:
…the SOX ran 76% above its 200-day this year, second only to 2000, and every past rally that stretched 40%-plus above it and then broke down ended in a crash.
As I write here on Wednesday, the SOX has fallen to 10,203. That’s about 5.5% below its 200-day average, and below the 10,300 line Luke identified as the one that matters most. In his words, that’s the level “where a shakeout turns into a breakdown and we finish tightening.”
To be clear, one day doesn’t confirm a breakdown. Luke’s plan is built around where the SOX settles, not where it trades at a given moment, and his study shows that even breaks below this line bounced about half the time. But this is the scenario where Luke has said he’ll turn defensive, so if the SOX can’t reclaim 10,300 quickly, expect him to be tightening up.
Here’s the full roadmap. Above 11,684, Luke turns more aggressive. Near 10,800, he’s holding. Below 10,447, he starts trimming weaker names. And at 10,300, he turns defensive.
Coming full circle, what’s the takeaway from yesterday’s elections?
The AI buildout isn’t over, but it just got slower, pricier, and more political. That’s bad news for some of yesterday’s AI winners, potentially good news for the pipelines fueling AI, and it puts Luke’s 10,300 line for SOX front and center in the days ahead.
Have a good evening,
Jeff Remsburg
