Coffeezilla

The Rabbit r1 was a $199 orange gadget that promised to run your life through something called a “Large Action Model” — an AI that would, its makers said, learn to operate any app the way you do and then do it for you. It launched to a wave of tech-press wonder. Then Stephen Findeisen sat down with one, and in “$30,000,000 AI Is Hiding a Scam” (21 May 2024) showed that the Large Action Model was, functionally, scripted automation wearing the costume of intelligence — and that the company behind it had a concealed prior life as an NFT project. The video did the one thing the entire AI news cycle had temporarily forgotten how to do. It checked whether the machine was real.

That is Coffeezilla’s whole method, and it is why he belongs in an issue about the machines. He is not an AI channel. He is a fraud channel that keeps finding AI at the scene, because the current confidence bubble is exactly the kind of environment his instincts were built for: an atmosphere of enormous claims, technical opacity, and money moving faster than anyone can audit it.

What It Does

Findeisen is an ex-chemical-engineer out of Texas A&M who spent 2017–2020 grinding as “Coffee Break” before the Coffeezilla project found its beat. The beat is financial fraud, told as documentary. The format is instantly legible: green screen, Blender-built sets, money-flow graphics that trace where the funds actually went, and an animated robot bartender named Maxwell for comic relief in what are otherwise fairly grim stories. The signature is the receipt. Findeisen does not describe a scam in the abstract; he shows you the wallet, the timestamp, the on-chain transfer, the deleted tweet.

The landmark remains “I Accidentally Got SBF To Admit to Fraud” (December 2022), in which he pressed Sam Bankman-Fried into a self-incriminating answer on a Twitter Spaces call. The CryptoZoo series — his multi-part investigation into Logan Paul’s collapsed NFT game — is the fullest expression of the model: patient, sourced, and willing to follow a story across months. And the $HAWK / “Hawk Tuah” investigation (5 December 2024, 6.4M views) caught a meme-coin launch in something close to real time, the token cratering from a roughly $500M valuation to around $60M inside twenty minutes while snipers reportedly extracted $1.3M-plus.

What It Does Extraordinarily Well

The craft is the argument. A Coffeezilla video is built like a legal filing that happens to be entertaining — thesis, evidence, evidence, evidence, conclusion — and the entertainment never softens the rigour. This matters more than it sounds, because the genre’s gravity is always toward the cheap version: outrage without proof. Findeisen resists it. When he cannot prove a thing, he says so. That discipline is why his accusations land, and why they survive contact with lawyers.

Which brings us to the X-factor, and the reason the score is where it is. Coffeezilla does journalism that carries genuine consequence — real enough that Logan Paul filed a defamation suit against him in June 2024 (heading, as of mid-2026, toward federal court in San Antonio; the allegations remain contested and the matter is ongoing), and real enough that Andrew Tate doxxed him in October 2024. You do not get sued and doxxed for making reaction content. You get sued and doxxed for finding something. In the AI moment specifically, he functions as an immune response the platform grew on its own — the mechanism that shows up when the hype is thickest and asks the unfashionable question of whether the product does what the demo claimed.

Coffeezilla is the closest thing YouTube has to an immune system — and like an immune system, he mostly arrives after the infection.

Where It Falls Short

Two things, and they are related. The first is cadence. Coffeezilla published only eight videos in all of 2025 — a real figure, and a real problem for a channel whose Consistency we score at 55. Investigation is slow, we accept that; you cannot rush a case to a weekly schedule without becoming the thing you investigate. But eight videos a year is an event calendar, not a diet, and the score reflects it honestly rather than waving it away as artisanal scarcity.

The second is the immune-system caveat in the pull quote, which is also the channel’s most-cited internal critique: he is reactive, not proactive. The great Coffeezilla videos are mostly autopsies. He is exceptional at documenting a collapse and rarer at calling one before it happens — the Rabbit and $HAWK pieces are the exceptions that prove how hard the live version is. In a machine cycle that mints new claims weekly, the lag between the scam and the exposé is the space where the damage gets done.

The verdict. None of that dislodges the case. What the high Content Quality and X-Factor say together, against that low Consistency, is a channel whose every output matters more than most channels’ entire year — an EXCELLENT built on rarity and consequence rather than volume. In an issue about what happens when machines make the claims, Coffeezilla is the one asking for the receipts. EXCELLENT, 84, entering the Top 50 at the threshold.