Paradox has built something genuinely rare: 1.28 million subscribers, 117.5 million views, fourteen years of catalogue. The hard part is done. The part that isn't is what happens after someone watches.
There is a version of this piece that opens with the subscriber count and stops there. It would be true and it would be useless. So here is what those numbers actually consist of, and what I could and couldn't verify about them.
Paradox runs four YouTube channels carrying 1,281,000 subscribers and 117.5 million views between them, 92.8 million of those on the founder's own channel, which has been publishing since March 2012. Around 1,700 videos, including 197 episodes of one interview format and 103 of another, with guests most French media businesses could not book. There is a product behind it that already ships. On the evidence available from outside, the expensive, slow, unrepeatable part of building a media business is finished.
The company also states 10,000 students, a 4.9 out of 5 rating and €50M in revenue generated. I have no way to check any of those from outside, so I am reporting them as claims rather than treating them as findings.
What the public record shows is a company operating as two separate machines that don't touch.
One is paid, recent and well made. As of 26 August there were 30 active ads in France across three advertiser identities, the most recent launched on 30 July. They point at a 23-question public diagnostic and a workshop scheduled for 3 September, with UTM tags named consistently all the way through the redirects. Someone competent is running it.
The other is organic, and it stopped. Across all four channels, the last thing published went out on 13 May.
| Channel | Subscribers | Last published |
|---|---|---|
| Founder channel (FR) | 914,000 | 16 March 2026 |
| Business | 138,000 | 25 March 2026 |
| Unfiltered (FR) | 121,000 | 13 May 2026 |
| Unfiltered (EN) | 108,000 | June 2024 |
The catalogue keeps being watched daily regardless. That is the useful part: this is a stock that goes on working with nobody at the controls, which makes it the cheapest thing in the business to restart.
Of the last 45 videos, 8 carry a link to anything Paradox owns. Of the 14 links I found across them, 8 point to a partner investment app rather than to Paradox itself. None point to the flagship product, none to a programme in the catalogue, none to the qualification quiz.
Then the homepage. It carries 101 links. Not one of them reaches the diagnostic, the workshop, or either of the landing domains the advertising is actively paying to fill. Two reach the app. So the paid engine spends money sending people somewhere the owned audience is never told exists, and the owned audience is handed to a partner.
An earlier draft of this analysis said that 35 of 36 recent videos carried a Paradox link, which would have made the routing problem disappear entirely. That count had swept up the channel signature displayed automatically beneath every video. The real figure, checked a second time by a different method, is 8 of 45 — and it reverses the conclusion.
I mention it because it is the whole reason the rest of this is worth reading. A number that flatters the story is the one to re-measure, not the one to lead with. Every figure above was taken at the source, dated, and where it carries a conclusion, taken twice by two different methods.
Sources. Subscriber and view counts from the About panel of the four channels, 26 August 2026. Publication dates from channel RSS feeds and cross-checked against the rendered /videos pages. Video descriptions read in the feed, then six of them reopened in a browser with the description expanded. Redirect destinations resolved individually. Active advertising from the Meta ad library, France, status "active". Homepage link counts from the rendered DOM. All landing pages reopened in a logged-out context.
Out of reach from outside: analytics, CRM, ad spend and cost per lead, retention cohorts, anything behind a paid account, product roadmap.