Still Comparing This To The Dot-Com Bubble?
How's that working out for you?
Being Clever
In Fight Club (1999), the narrator—Jack in the screenplay—has a clever name for the fellow frequent flyers he meets, “single-serving friends”:
JACK (V.O.) Everywhere I travel -- tiny life. Single-serving sugar, single-serving cream, single pat of butter.
JACK (V.O.) The people I meet on each flight -- they’re single-serving friends. Between take-off and landing, we have our time together, but that’s all we get.
Then he meets Tyler Durden.
JACK Tyler, you are by far the most interesting “single-serving” friend I’ve ever met. Tyler stares back. Jack, enjoying his own chance to be witty, leans closer to Tyler. JACK You see, when you travel, everything is small, self-contained--
And Tyler cuts straight through it:
TYLER The spork. I get it. You’re very clever. JACK Thank you. TYLER How’s that working out for you? JACK What? TYLER Being clever.
Here’s that scene:
I thought of that Thursday night watching Dell (DELL 0.00%↑) rip higher after earnings.
For months, the clever take has been that AI is 1999 all over again: spectacular technology, ugly economics, reckless capex, and hyperscalers spending huge sums on a future that may not arrive.
Fine.
How’s that working out for you?
Dell Just Answered
Dell reported a blowout quarter Thursday. Revenue rose 88% year-over-year to $43.84 billion, adjusted EPS came in at $4.86, and the company raised its full-year revenue outlook to $165 billion–$169 billion from $138 billion–$142 billion. More importantly for the broader AI debate, Dell lifted its fiscal 2027 AI-server revenue forecast to $60 billion from $50 billion.
That’s the part of the AI story the dot-com-bubble crowd keeps missing.
The revenue is real. The earnings are real. The orders are real. The customers are still building.
Dell’s infrastructure-solutions business revenue jumped 181%, and AI-optimized server demand kept accelerating. One report put AI-optimized server orders at $16.1 billion, up 757% year-over-year.
That is not a sock puppet with a Super Bowl ad.
That is hardware being ordered, shipped, installed, and paid for.
The scoreboard keeps moving in one direction: the physical AI buildout is showing up in the numbers.
The Clever Bear Case
A post making the rounds on X this week argued that the “AI numbers are starting to look very ugly,” citing the FT’s negative estimated AI ROI for Microsoft, Google, Meta, and Oracle, with only Amazon barely positive.
The conclusion was familiar: the technology may be real, but the economics are suspect; the internet survived, but most internet companies did not; hyperscalers are spending massively today on the hope that future demand eventually justifies current capex.
That framework has two big problems.
It treats AI capex as if the only valid scoreboard is a near-term ROI calculation inside the hyperscalers themselves.
There’s another scoreboard already lighting up: the suppliers.
Servers. Memory. Networking. Cooling. Power. Data-center infrastructure. Grid equipment. Photonics. Semiconductors. Defense. Automation.
The physical AI buildout isn’t waiting for a 2030 spreadsheet to become real. It is already showing up in the companies selling the picks, shovels, racks, chips, power systems, and infrastructure.
That was the point of our previous post, “Partying Like It’s Not 1999.”
In 2000, the internet was real, but many internet stocks were priced as if the future had already arrived. Today, the AI buildout is real, and many companies enabling it are still being valued like cyclicals.
That’s a very different setup.
The Hyperscalers Don’t Look Like 1999 Either
The other big problem with the dot-com bubble framework is the hyperscalers don’t look expensive either on a PEG basis. Using a simple PEG-like measure—current P/E divided by latest reported diluted EPS growth—most of them look a lot cheaper than the 1999 analogy would suggest.
Even if you want to parrot the FT’s ROI math, the valuation side of the 1999 comparison is a lot shakier than the clever take makes it sound.
The Trump Effect
There’s another piece of this tape that the top-callers keep underestimating: policy.
Dell had just landed a five-year, $9.7 billion Pentagon contract before it reported earnings. That wasn’t the whole story—the earnings were the main event—but it mattered.
In this market, companies sitting at the intersection of AI infrastructure, defense demand, and government spending can rerate quickly when investors see both revenue growth and Washington’s checkbook pointing in the same direction.
A number of accounts on our Market Watchers X list pointed out the Trump connection too: earlier this month, President Trump publicly praised Dell and told Americans to “go out and buy a Dell.”
Nineteen days later, Dell had the Pentagon contract. Then came Thursday’s earnings.
We weren’t in Dell, but we got a taste of the same phenomenon Thursday with Aevex Aerospace (AVEX), a drone name we placed a trade on Thursday morning after reports that the Trump administration was considering funding selected U.S. drone companies.
AVEX closed at $40.48, up 31.39% on the day.
That’s how this market has been working.
Policy support matters. AI infrastructure matters. Defense and autonomy matter. When those themes line up with contracts, revenue growth, or capital flows, stocks can move fast.
Back To Work
Plenty of smart people have spent the last few months predicting crashes, fading the AI trade, and waiting for a better entry point.
Meanwhile, the market keeps rewarding those in the right themes, from AI, to drones, to space. We’ve been in all of them.
While others stay on the sidelines polishing their 1999 analogies, we’ll keep doing what has worked: scanning for names where the story, the screens, and the setup line up.
If you want a heads up when we place our next trade, check your inbox later.
If you’re still expecting the market to crash, we’ve got you covered there too
Perhaps your system needs a systematic political-policy signal, too. Government involvement was always a red flag to me in the past, but we seem to have entered a new era in that regard—certainly, being clever by eschewing governmental influence hasn't been helpful to me over the last few years; if you can't *avoid* 'em, join 'em.