THESIS: Every operator in a restricted vertical fights for the same blocked channels — and loses the same way. Stake never entered that fight. Locked out of Google, Meta, and most regulated markets, they did something structurally different: instead of renting attention from ad platforms, they bought creators — and when the platform hosting those creators banned them, they bought the platform layer itself. Revenue went from $100M to $2B+ in two years on the creator play, and $4.7B annually with ~$80B in yearly transactions on the full machine. The lesson isn't "sponsor streamers." It's the ladder: rent → buy → own — and knowing which rung you're on before someone kicks it out from under you.
CONTEXT: Stake is arguably the most-banned major brand on the internet — blocked in the US and most of Europe, UK licence revoked by the Gambling Commission (Dec 2024). Paid social and search were never an option. What they built instead: sponsored gambling streams on Twitch (millions/year to streamers gambling live on camera) drove the $100M→$2B run. Then, in late 2022, Twitch banned unlicensed gambling streams — the exact platform-risk event every creator strategy fears. Two months later, Stake's founders launched Kick: their own streaming platform, favorable to the content Twitch exiled, with a 95/5 sub split that vacuumed in creators. By 2026: 100M+ users, 4.5B watch-hours in 2025 (+131% YoY), the #4 most-watched live platform on earth behind YouTube, TikTok, and Twitch — funded by ~$1B of the founders' own money. No platform can ban them tomorrow, because they are the platform.
Four mechanisms — each mapped to what an operator in iGaming, crypto, dating, or sweepstakes can actually run.
═══
What it is: Stake didn't sponsor celebrities or buy pre-roll. They picked creators whose content format was the product demo: streamers gambling live, on camera, with real balances. The viewer wasn't watching an ad for a casino — they were watching the casino being used, for hours, by someone they parasocially trusted.
Evidence: The Twitch-era deals (Trainwreck, xQc and the slots meta) made gambling streams one of the platform's biggest categories before the ban. The mechanism: watch-time in restricted niches converts because the session is the funnel — a 3-hour stream is 3 hours of product exposure, social proof, and codes on screen, at a depth no 30-second ad can reach.
Application: In a restricted vertical, select creators by three filters, in order: (1) format-fit — their native content is your product in use (slots streamers for casino, trading streamers for crypto, unboxing/lifestyle for sweeps), not adjacent lifestyle with your logo bolted on; (2) audience qualification — 18+/21+ skew, geo-match to your licensed markets (a creator whose audience can't legally convert is a liability, not reach); (3) conversion evidence over follower count — rank on attributable signups/deposits per stream, not viewers. The nano/mid tier usually beats marquee names on cost-per-deposit — the marquee names are for a different job (Mechanism 3).
═══
What it is: Stake's deals were never "post about us" — they were long-term, exclusive, deeply-integrated partnerships where the creator's income depended on the relationship, not the platform. When Twitch banned the category, the creators didn't drop Stake — they followed the money to Kick. The deal structure made the relationship the durable asset, and the platform a replaceable surface.
Evidence: The gambling-stream exodus after the Twitch ban didn't scatter — it moved as a bloc, because the contracts and the economics (reported multi-million monthly deals for top streamers) lived outside Twitch. Content distribution collapsed for a category; Stake's creator roster stayed intact.
Application: Build ban-survivability into the contract:
═══