Faceless YouTube Channels: How to Start One (Real Numbers)

My faceless YouTube channel has 7 subscribers.

I'm telling you that in the first sentence because almost everything else you'll read about faceless YouTube channels is trying to sell you something, a course, an "autopilot" video tool, a dream. I checked what currently ranks for this search while writing this. Two of the top ten results are literally landing pages for faceless-video generators, and the realest number in any of the articles was someone mentioning they'd made 9 sales of something. So here's the version with actual numbers in it.

Quick answer if you're skimming. Yes, faceless channels can make money, a small minority make very good money, and the bar to reach monetization is much higher than anyone selling you the dream admits. YouTube wants 1,000 subscribers plus 4,000 public watch hours in a year, or 1,000 subscribers plus 10 million Shorts views inside 90 days, before you earn your first cent of ad revenue. My channel is 993 subscribers short. I keep it running anyway, and later in this post I'll explain why that's actually rational.

What a Faceless Channel Actually Is

Any channel where the creator never shows their face. That's the whole definition, and it covers way more than the AI slop people associate with the term. Kurzgesagt is a faceless channel with north of 20 million subscribers, it's animated science explainers made by a whole studio. Ambient music channels, history documentaries, gaming compilations, ASMR, phone-screen tutorials, all faceless.

The reason the term exploded recently is that AI collapsed the production cost. Script, voiceover, visuals, music, all of it can now be generated, which pulled thousands of people in with the promise of content that makes itself.

The promise is half true. That's the interesting part, and the trap.

Do Faceless Channels Really Make Money

Some do, genuinely. But let me do the math the thumbnails skip, because once you see it you can't unsee it.

To earn anything from ads you need to get into the YouTube Partner Program first, and for Shorts that means 10 million valid public views in 90 days plus a thousand subscribers. Ten million. For long-form it's 4,000 watch hours in a year, which sounds gentler until you realize that's 240,000 minutes of humans actually watching your AI content.

And after you're in? Ad revenue depends on your RPM, the amount you earn per thousand views, and creators publish RPMs anywhere from around a dollar to over $20 depending on niche. Here's a back-of-envelope you can redo yourself. At a $5 RPM, the often-quoted goal of $10k a month means about 2 million monthly views. Every month. At a $2 RPM, it's 5 million. This is why the answer to "how do I make ten grand a month on YouTube without making videos" is honestly that you mostly don't, and the people who do have either fat-RPM niches (finance, business) or numbers so large they're effectively running media companies.

One more myth while I'm here. "Without making videos" isn't a real category. Automated channels still require somebody to pick topics, run the pipeline, review output, and fix the constant weird failures. I know because I do it. You're not removing the work, you're changing what the work is.

What Running Mine Actually Looks Like

I built my channel as an experiment in full automation, and I mean full. The music is generated by an open-source model running on my own computer. The visuals too, my machine takes something like fifty seconds per image. Assembly into a finished video is scripted, uploads are scheduled, and my hands-on time per video is a few minutes.

Building that pipeline took weeks. I want to be honest about the trade, because "15 minutes per video" claims float around this niche constantly and they skip the part where the system behind those 15 minutes took real engineering time to exist.

Current results, months in. 13 videos live, 7 subscribers, revenue zero (see the threshold math above, I'm nowhere near it). My marginal cost per video is close to zero because everything runs locally instead of on paid AI services, so the experiment costs me basically electricity.

And since nobody else writes this part, here's what "automated" actually breaks like. Renders fail silently and you find out the video is 40 minutes of black frames. The music model occasionally produces something unusable and you regenerate three times. A thumbnail comes out with mangled text and you don't notice until it's live. None of these are disasters, but each one is a human catching a machine, which is the recurring theme of every automation story I have. The pipeline does the producing. I do the noticing.

So why keep going? Because at zero marginal cost, a channel is a lottery ticket I got for free, and the pipeline skills transfer directly into work that actually pays now, client videos, product promos, my own projects. I wrote more about that framing in the main guide on how to make money with AI, where faceless YouTube deliberately ranks last out of nine methods.

The Stack, Free and Local vs Paid and Cloud

Two honest ways to build the pipeline. Mine is the left column.

The paid route ships your first video this weekend, the local route takes weeks to wire but then produces at no cost forever. If you're testing whether you even like this, honestly, take the paid route with the cheapest tiers and decide later. The tool subscriptions are where most beginners bleed money though, so cancel fast if you're not uploading.

How to Start One This Month

The steps that matter, in the order I'd redo them today.

First, pick the niche by demand, not vibes. I keep a whole post on youtube automation niches, but the short version is you want topics people search and rewatch (ambient music, explainers, top-10s in evergreen subjects) rather than whatever's fun to generate.

Second, commit to one format you can repeat fifty times. One. The channels that grow are recognizable within half a second of the thumbnail. My 13 videos are all the same format on purpose, even at 7 subscribers I'm testing a repeatable thing, not throwing spaghetti.

Third, disclose the AI. YouTube requires disclosure when content is realistic and AI-generated or meaningfully altered, and repeated non-disclosure can get you removed from the Partner Program entirely. Non-photorealistic stuff like animated explainers gets a lighter label in the description. Either way, just tick the box, the penalty asymmetry makes hiding it stupid.

Fourth, thumbnails and titles are the actual product. Nobody sees your video quality before they click. I generate thumbnail candidates in batches and pick, which costs me nothing locally, and it's still the highest-leverage manual step in the whole pipeline.

Fifth, upload on a schedule and judge nothing for 90 days. YouTube's recommendation system needs data, and channels routinely sit at nothing for months before any movement. Mine still sits at nothing, I'm aware, the difference is mine costs zero to continue.

Sixth, when you do start judging, only two numbers matter early. Click-through rate tells you whether the thumbnail-title combo earns the click, and average view duration tells you whether the video deserves it. Views themselves are a lagging output of those two, so when a video flops I change the packaging or the pacing, not my mood. I check mine about once a week, takes five minutes, and honestly with 7 subscribers the weekly check is mostly a discipline ritual. But it's the same ritual that will matter if this ever moves, so it runs.

Should You Actually Do This

Depends what you're buying with the effort, and I'd genuinely think about it in those terms.

As a primary income plan, no. The threshold math above is brutal, the competition ballooned the moment AI dropped production costs, and the income ranges you see quoted in ranking articles ($12K to $120K a year appears in one, with no receipts attached) describe survivors, not averages.

As a cheap experiment that teaches you AI video production, scripting, and packaging, while holding a free lottery ticket on the side? That's the version I run, and I'd defend it. The skills already pay me through other work even though the channel pays nothing, which is a strange sentence, but it's the true one.

If money this quarter is the goal, the faster routes are freelancing and digital products, and I laid out all nine options with the same honesty in the full guide. And whichever way you go, my monthly numbers, including this channel's slow crawl past 7 subscribers, are in the income ledger. Zeros included. Especially the zeros.