Adavia Davis understands that better than perhaps anyone else. Since dropping out of Mississippi State University in 2020, the 22-year-old has built a thriving content-creation business out of what has come to be called “slop”— that high-volume, AI-generated background noise that thrives in the gaps of our focus. Davis’ most successful videos aren’t meant to be watched, shared, or even remembered. Often, Davis told Fortune, his viewers are asleep.

Davis has assembled a sprawling network of YouTube channels that operates as a near-autonomous revenue engine, requiring only about two hours of his oversight a day. He currently runs five active channels, but his broader portfolio includes multiple Minecraft channels aimed at children as well as channels devoted to funny-animal compilations, prank videos, anime edits, Bollywood clips, and celebrity gossip. Most lucrative is a “Boring History” channel built around six-hour “history to sleep to” documentaries, narrated by what sounds like a languid David Attenborough.

The channels belong to a genre that has come to dominate YouTube, known as “faceless” content–-videos designed to be scalable, easily replicated. Nearly all of Davis’ videos are generated with artificial intelligence, anchored by TubeGen, a proprietary software pipeline built by his partner, fellow 22-year-old Eddie Eizner, that automates nearly every step of production. Scripts and visuals are generated with Claude, the silky British narration from ElevenLabs, then assembled into long-form videos. The results can run as long as six hours, costing as little as $60 to produce from start to finish. 

Davis told Fortune that his network of videos generates roughly $40,000 to $60,000 a month in revenue. His operating costs—primarily small salaried teams overseeing the different niches—run at about $6,500 per month, he added. The margins are 85%-89%, extraordinary by tech standards. 

Fortune reviewed screenshots from Davis’ social media analytics dashboards, as well as recent AdSense payout records, which show tens to hundreds of thousands of dollars in monthly earnings from individual channels, equating to annual gross revenue of roughly $700,000. He talked to Fortune more about what is turning into his career, how it got started, and why college wasn’t part of the equation for him.

How Davis hacks the attention economy

Growing up on YouTube, Davis was a product of the platform’s golden era. When he was 10 years old in 2014, he said, he would spend six hours a day scripting and editing Minecraft and Fortnite playthroughs. He said he mourns the passing of this era, a time when creators were driven by “a love of the game, not necessarily to sell something.” 

But by 2022, the launch of ChatGPT shifted the internet’s market logic. Davis said he saw the writing on the wall early: the era of the personal brand was being eclipsed by the large-scale-content farm. But he was also, frankly, surprised by what turned from a hobby to a side hustle to something resembling a business. “I didn’t start [making content on] YouTube to make AI videos,” he said, adding that it was just for fun at first, but money started coming in from his various channels. “Then, if all my competitors are uploading more than me, and I’m waiting on my scriptwriter to get done, then I’m just falling behind.”

Davis was a 19-year-old college student when he felt the internet world shifting under his feet. He sold his first YouTube channel to a brand, which converted the account into a marketing feed for its product (Davis said he routinely accepts this kind of deal, even if it rarely pays off for the buyer: “they don’t know what they’re doing”). To celebrate, he spent what he describes as the last of his savings on a Tesla Model 3, at the time retailing at $55,000, not leaving any funds for tuition. Davis had enrolled in school largely for the experience, he said, but quickly realized he couldn’t juggle classes and content creation without killing both. “If I stayed in school, I was going to be broke and distracted,” he said. “That was just a setback for no reason.”

Davis turned fully to making YouTube channels with the new AI tools at his disposal, with the internet that he grew up with now gone forever, in his opinion. “The ethics have gotten really, really bad from these higher-up companies that have their number one goal as attention,” Davis said. “Because attention is the number one currency. Whoever has the most influence controls the most.” He described the system that he’s monetized as very “psychological,” even destructive—“trying to destroy minds to make them easier to sell to.”

Davis explained his understanding of the business model as YouTube needing to cater to advertisers, “the puppet masters” of the platform, in order to stay alive. The only way to survive in this system, he argued, is to understand it, or even teach it. (In fact, Davis said that he offers an online course for people looking to supplement their income, including his belief that “social media is a social science.”)

Recent data suggests that so-called “AI slop” has rapidly expanded across YouTube. Researchers at the video-editing company Kapwing found that more than 20% of the videos shown to new users fall into that category. The study further found that channels posting nothing but that AI low-quality content have collectively amassed over 63 billion views, 221 million subscribers, and an estimated $117 million a year in advertising revenue. YouTube, meanwhile, has emerged as a major player in both TV and streaming, with the 2020s marking a turning point in the popularity of podcasts with video, and YouTube’s more traditional TV offerings such as NFL (or, next year, the Oscars) combining with its dominance in user-generated content (UGC) to make it an engagement giant. Melissa Otto, head of research at S&P Global Visible Alpha, previously told Fortune that YouTube’s dominance in UGC is the real reason Netflix is spending so heavily to try to acquire Warner Bros. Disney’s subsequent $1 billion licensing deal with OpenAI fits into a similar category, per Nicholas Grous, director of research for consumer internet and fintech at Ark Invest.

Against this backdrop, Davis remains a comparatively small fish: he has built and sold faceless AI-driven channels ranging from roughly 400,000 subscribers to just over one million. Yet, he said his network of videos now averages about two million views per day. “When you understand psychology, everything else just falls into place,” he said.

Over the past several years running channels on YouTube as well as shows on TikTok, Instagram, and Snapchat, Davis said that he’s learned to optimize for social media’s most unforgiving metric: watch time. Some tactics are straightforward. Davis obsessively engineers the opening seconds, or the “hook,” of a video—the bright contrast of colors on screen, the first facial expression or vocal inflection you hear—because that initial moment determines whether a viewer stays or clicks away.

Others are more mischievous. In compilation videos, Davis sometimes turns to shock tactics such as a sudden flash of a spiders on screen for a split second at the beginning, just long enough to make viewers rewind and check whether they actually saw what they think they saw. In short-form clips, he has intentionally misspelled words on screen to bait viewers to pause, comment and correct him, stretching watch time in the process.

“I do everything in my power to trick watch time,” he said. “Because that’s the metric that’s going to pay you at the end of the day.”

The 2027 deadline

So far, Davis has had something of a first-mover advantage, given how early he was to spot the arbitrage opportunity and also his long-developed intuition for the sort of video that performs well.

But now, with AI advancing beyond scripts into video production and further collapsing barriers to entry, competition has grown fiercer. He said the biggest career mistake he ever made was posting a promotional video for TubeGen showing how he made his long-form Boring History sleep videos using AI. Within days, Davis said that he saw scores of copycats posting similar videos, crowding out the niche that he had built and monopolized, until then.

But more threatening than the individual imitators, he said, are the companies with capital. Davis describes himself as “kind of a doomer” about the future of the space, estimating that individual creators have until around 2027 to meaningfully profit from AI-generated long-form YouTube content.

After that, he predicted the “sharks” will arrive: large media companies with the capital to industrialize any format the moment it proves lucrative. “At that point,” he said, “you’re just competing against the big fish.”

​​Davis pointed to a World War II history channel that he admired, full of thoughtfully produced videos that seemed to come from a student, posting every other day. Once an unnamed media company noticed the niche, it began uploading three times a day. Those sorts of videos cost roughly $110 to produce, he estimated, whereas posting at the media company’s speed would cost over $300. “You can’t compete unless you have the budget,” he said. 

Still, he said he was optimistic that he’ll find a way to “seep through the cracks,” as he has for three years now. Rather than inventing new genres, Davis said he looks for small edges inside formats that already work. Most recently, he has been experimenting with a twist on a familiar setup: pairing narrated Reddit posts with looping Minecraft footage—but instead of a classic Reddit story, swapping in narrated horror stories for the “psychopaths,” as he put it, who like to fall asleep to them.

“The proof of concept is there,” Davis said.

But Davis hopes that one day, soon, none of his content will be much in demand at all. As AI content floods the internet and trust erodes, he believes authenticity itself will become scarce,and therefore valuable. He already sees a growing audience for creators who reject heavy editing and algorithmic tricks.

“It’ll get worse before it gets better,” he said, but eventually, “True longevity,” he said, “is going to come within brands and real influencers with real faces.”