The AI Collapse of 2026: Chaikin’s Warning
“The AI collapse of 2026” is the fear-side phrase in Marc Chaikin’s current “The End of Elon?” presentation, and it is worth reading carefully because the warning is more precise than the headline. Chaikin is not predicting the end of AI, and he says so explicitly: this is not “some kind of anti-AI trade… like the HALO trade so many people are talking about.” It is a rotation thesis. The claim is that a new class of precision hardware will shift where AI value is created, and that some of today’s most-loved AI stocks are on the wrong side of the shift.
What the warning actually says
The host, Molly Hendrickson, delivers the sharpest version: “Today’s AI darlings are, in some cases, about to get clobbered… Along with anyone holding shares.” And the personalized frame: “if you have a dime invested in Tesla… Or Nvidia… Or Microsoft… or Alphabet… or Meta… You need to understand what’s coming next.” Chaikin crystallizes it as “a lot of AI darling stocks are actually doomed right now,” and the companion report in the offer is titled “The Worst AI ‘Landmine Stocks’ to Avoid Right Now.”
The mechanism behind the warning is the presentation’s genuinely substantive part: the floating point problem. Today’s generative AI runs on low-precision FP16 hardware, which is fast but rounds numbers. Chaikin’s on-camera demonstration, asking ChatGPT to count the C’s in “broccoli,” shows the limitation, and his explanation is technically sound: “That lack of precision can’t be solved with better programming. You can’t solve it by feeding more data into the models. Because the floating point problem isn’t a software problem. It’s a hardware problem.” Scientific work, fusion simulation, drug discovery, materials science, needs FP64 double precision, and the DOE is building FP64 AI supercomputers, starting with Lux at Oak Ridge, expected to fire up “before the end of this year.”
What is real and what is framing
The honest accounting requires separating three layers. The technology is real: FP16 versus FP64 is textbook computer science, the DOE machines are real, and the shift of scientific computing onto AI-era hardware is a genuine trend. The rotation logic is defensible: new infrastructure classes historically created new winner lists, and Chaikin’s own track record on calling the big turns, the COVID crash in January 2020, the bottom “just 24 hours before the greatest market recovery in American history,” the 2022 bear, the 2023 rebound, Liberation Day, gives the call standing.
The framing is where skepticism is fair. “Doomed” is stronger than the argument supports, because FP16 generative AI and FP64 scientific computing solve different problems, and the presentation itself qualifies its comparisons with “regarding major breakthroughs” every time. The urgency device, “Do not buy or sell shares of a single AI stock until you’ve heard what he has to say,” is manufactured pressure no honest analyst would literally apply to an entire portfolio. And “tear the AI industry down to the studs” is rhetoric, not analysis. Chaikin’s own gentler formulation is the accurate one: “if you’re already positioned in today’s top AI plays, that’s a good thing. What I’m showing you today is a different layer of the same opportunity. I want you to have both.”
The constructive side
The collapse warning is the door; the American Atlas thesis is the room. The free pick is AMD, “the first company on my American Atlas ‘buy’ list,” whose Instinct chips already power Frontier and El Capitan. The paid American Atlas Portfolio covers “17 companies that have signed deals… They’re all working directly on building American Atlas,” and the Landmine Stocks report covers the avoid side. The offer is Power Gauge Report, normally $499/year, discounted for the promotion, with a 30-day full-refund guarantee, “That puts all the risk on Marc’s side of the table.”
For readers who hold AI stocks, the useful takeaway is not paralysis but discrimination: the thesis argues that precision hardware, networking, and the DOE buildout are the next layer, and that within AI, positioning matters more than the sector banner.
Where to learn more
Start with our End of Elon? American Atlas review, then the FP64 explainer and the Oak Ridge and Frontier detail. The full presentation is at the Chaikin Analytics offer page.
NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.