Listen to the audio version of this article (generated by AI).
AI’s meltdown week… Luke Lango’s take on the selloff… data centers are on tomorrow’s ballot… so is the drone budget… Luke’s Pennsylvania Plan
Friday put an exclamation point on a brutal month for the AI trade.
By the closing bell, the Nasdaq had fallen 2.9% on the day while the S&P 500 had lost 1.8% and the Dow had dropped 1%. Chip stocks took the brunt of it, with the Philadelphia Semiconductor Index (SOX) sinking more than 7%.
For October, the Nasdaq fell roughly 5.2% – its worst month since March 2025 – while the SOX did far worse, dropping more than 15%.
Two earnings reports did most of Friday’s damage…
Intel Corp. (INTC) was the biggest casualty, plunging nearly 29% – its steepest one-day drop on record – after forecasting fourth-quarter revenue of $15 billion to $16 billion, well short of the roughly $17 billion Wall Street expected. CEO Lip-Bu Tan said customers were “signaling a PC slowdown” tied to memory prices and availability.
Amazon.com Inc. (AMZN) fell too, though its story is more nuanced. Its cloud business, AWS, grew 39% to $46 billion, which is hardly a sign of weak AI demand. But Amazon spent $77.3 billion on capital expenditures in the quarter, more than double the $35 billion it spent a year earlier. Over the past 12 months, that spending has pushed Amazon’s free cash flow more than $42 billion into the red – a record.
That number gets at what’s really spooking investors…
Friday’s losses piled onto an already brutal stretch. Alphabet Inc. (GOOGL) shed nearly $500 billion in market value after its earnings report earlier in the week, one of the largest single-day wipeouts in stock market history.
Alphabet and Meta Platforms Inc. (META) now sit more than 20% below their highs, which puts both in bear-market territory.
The fear isn’t that AI is failing. It’s that the bill is coming due. JPMorgan estimates that the tech giants will spend roughly $900 billion on AI this year, and investors are increasingly asking when – or whether – that spending will produce a return worth the price.
So, how should AI investors read this?
Is this the beginning of the end for the AI trade, or a gift for patient investors?
Our technology expert Luke Lango, editor of Early Stage Investor, spent the past few days digging through the hyperscalers’ earnings line by line – and while he doesn’t dismiss the bear case, that’s not where he ends.
Here he is from Friday’s Daily Notes:
We get why that story is spooking people. Parts of it are even true…
[But] the AI trade isn’t broken. It’s between paychecks. The spending is front-loaded, the payoff is back-loaded, and the gap between the two is about to start closing, sooner than most people think.
The key question is whether all that spending earns a return. Luke ran the rough math on each new dollar of AI spending, measured against a cost of capital of roughly 9% to 10%. Microsoft Corp. (MSFT) is earning about 19%, Alphabet about 12%, Amazon about 9% to 10%, and Meta about 8%.
In other words, it’s an uneven story. Back to Luke:
Companies that sell compute are expanding margins, while companies that use it internally are lagging for now.
And the companies selling compute have customers lined up. Microsoft, Amazon, and Google are sitting on roughly $1.8 trillion of contracted cloud backlog between them. AWS growth accelerated for a sixth straight quarter, to its fastest pace in about five years. That isn’t the hyperscalers spending money. It’s customers signing up to pay them.
Luke is also upfront about where the real risk lies:
The bigger uncertainty is 2028, when slower hyperscaler spending growth leaves suppliers more dependent on buyers exposed to tighter financing.
As for Amazon, Luke thinks Wall Street is reading too much into one quarter. Part of the weakness was simple timing.
For example, moving Prime Day into the second quarter knocked about 850 basis points off third-quarter ad growth, and management expects ads to bounce back to high-teens growth as soon as this quarter. And for all the worry about Amazon’s borrowing, it covers its interest costs about 16 times over.
To be clear, Luke isn’t calling a bottom, but his big-picture takeaway is clear:
The market is grading a decade-long build-out on one noisy quarter at peak spending, and that’s short-sighted. Our game plan: for long-term investors, this is a dip to buy with confidence.
The headlines say the AI boom is running out of money. The numbers say it’s about to start making a lot more of it.
Interestingly, the very first item on Luke’s catalyst calendar for the next three months is tomorrow’s midterm elections. That’s where the AI trade faces a different kind of test – one that won’t be decided on Wall Street.
Data centers are on Tuesday’s ballot
That’s not a figure of speech.
According to Ballotpedia, voters in 10 states will weigh in on at least 41 local data-center measures tomorrow. Ohio alone accounts for more than half of them – 21 measures across 18 communities – with proposals ranging from outright bans to caps on how much power a facility can draw.
We’ve already gotten a preview of how these votes can go. Earlier this year, about 88% of voters in Monterey Park, California, backed a policy prohibiting data centers. In Augusta Township, Michigan, roughly 90% voted against a rezoning that would have made room for one.
The issue has also worked its way into some of the biggest races on the map…
In Ohio’s U.S. Senate race, Democrat Sherrod Brown has run ads calling Republican Sen. Jon Husted the “face of data centers,” a reference to tax breaks Husted backed as lieutenant governor. Data centers have also become a talking point in the Texas and Pennsylvania governors’ races. In Pennsylvania, Gov. Josh Shapiro has already tightened the rules for approving new projects.
And this isn’t a one-party story. A Washington Post analysis of candidates’ posts found that politicians in both parties have turned significantly more negative on data centers since 2024 – with Republicans souring even more than Democrats.
The driver is the electric bill. A 2026 Reuters/Ipsos survey, cited by the Brookings Institution, found that 77% of Americans worry data centers will raise their electricity rates. Nearly two-thirds said building data centers at a rapid pace isn’t a good thing.
To be clear, a zoning vote in a small Ohio town won’t decide whether Amazon builds its next campus. But taken together, tomorrow’s results will give us the first nationwide read on how quickly public sentiment is turning against the AI buildout.
That matters more today than it would have a few months ago. As we just saw with Amazon, Big Tech is already spending far more than its operations bring in to build this infrastructure. If local resistance means slower permits, pricier power, and fewer available sites, the payback period on that spending stretches even further. And that’s exactly the question Wall Street is wrestling with right now.
The other spending boom on tomorrow’s ballot
AI isn’t the only corner of the market with a lot riding on tomorrow.
The election will also decide which party controls Congress come January. And that has major implications for a spending package that hasn’t gotten much attention from investors.
Earlier this year, the Pentagon sent Congress a $1.5 trillion budget request for fiscal 2027. Most of it would go through the normal appropriations process. But $350 billion – about 23% of the total – was earmarked for a separate budget reconciliation package.
Reconciliation lets the majority party pass spending with a simple majority, sidestepping the Senate filibuster. The catch is that it depends on party-line votes. According to Air & Space Forces Magazine, if Democrats win control of either the House or the Senate, that path effectively closes.
So, what’s in the package?
Two items stand out…
The first is the Golden Dome, the administration’s proposed missile-defense shield. The second is $53.6 billion for an office most Americans have never heard of: the Defense Autonomous Warfare Group, or DAWG – the Pentagon’s new drone-warfare unit.
That drone figure is staggering. DAWG received about $226 million in fiscal 2026. Including a small slice of base-budget funding, its fiscal 2027 request totals $54.6 billion, roughly 240 times larger.
Why is the Pentagon going all-in on drones?
I wrote about this last spring in our Investing Insider newsletter, and one comparison has stuck with me ever since.
Iran mass-produces a drone called the Shahed, which it fires in swarms. Each one costs roughly $20,000 to build.
The Patriot interceptor the U.S. fires to shoot one Shahed down costs approximately $3 million to $4 million.
The math here just isn’t sustainable. It’s like using a Ferrari to destroy a used Honda Civic – except the Civics keep coming, a thousand at a time.
The economics of warfare have flipped. Cheap, autonomous, mass-produced systems are overwhelming expensive ones, and the Pentagon is racing to catch up.
It’s also racing to restock…
Since July, the Pentagon has signed multiyear contracts worth more than $100 billion combined at their full values to replace missile stockpiles drained by conflict in the Middle East. But industry executives say Congress still hasn’t appropriated the money, which limits how fast contractors can invest in new capacity.
What happens if the package stalls?
It wouldn’t reverse the shift. But it could change the speed – and who ends up paying for it.
Defense analysts have long viewed the post-election “lame duck” session as the likeliest window to pass the package. Tomorrow’s results will tell us whether that window is a starting point or a last chance. And even if reconciliation fails, Pentagon leadership has signaled it may cut expensive legacy programs to fund low-cost autonomous weapons anyway.
And this brings us back to Luke…
He believes America is entering its biggest military buildup since World War II. And he argues that government spending is only part of the story.
He points to a recent gathering organized by Sen. Dave McCormick of Pennsylvania that brought together Wall Street heavyweights, defense leaders, government officials, and emerging technology companies. Luke calls the public-private effort coming out of it the “Pennsylvania Plan” – and he believes it could accelerate the rebuilding of America’s arsenal regardless of how any single budget fight plays out.
Luke has just wrapped up a research project on this rearmament boom – including the five companies he believes are best positioned to profit across its three phases. He’s put everything together in a new video presentation, which you can watch right here, free of charge.
Wrapping up
Tomorrow isn’t just a political event – it’s a market event.
For AI, it’s a referendum on how much patience voters have for the buildout. For defense, it could decide how quickly hundreds of billions of dollars start flowing – and Luke’s new presentation explains where he believes that money is headed.
We’ll report back.
Have a good evening,
Jeff Remsburg
(Disclosure: I own GOOGL, AMZN, MSFT)
